Sie sind auf Seite 1von 426

Microstrip Patch Antennas:

A Designer's Guide
Microstrip Patch Antennas:

A Designer's Guide

Dr R. B. Waterhouse

RMIT University

Springer Science+Business Media, LLC


.t Electronic Services < http://www.wkap.nl >

Library of Congress Cataloging-in-Publication

Waterhouse, R.B. (Rodney B.)


Microstrip patch antennas: a designer's guide / R.B. Waterhouse.
p.cm.
Inludes bibliographical references.

1. Microstrip antennas. 1. Title

TK7871.67.M5 W38 2003


621.382'4--dc21
2002041376

ISBN 978-1-4419-5338-4 ISBN 978-1-4757-3791-2 (eBook)


DOI 10.1007/978-1-4757-3791-2

Copyright © 2003 by Springer Science+Business Media New York


Originally published by Kluwer Academic Publishers in 2003.
Softcover reprint ofthe hardcover Ist edition 2003

AlI rights reserved. No part of this publication may be reproduced, stored in a retrieval
system or transmitted in any form or by any means, electronic, mechanical, photo-copying,
microfilming, recording, or otherwise, without the prior written permission of the
publisher, with the exception of any material supplied specifically for the purpose of being
entered and executed on a computer system, for exclusive use by the purchaser of the
work.
Permissions for books published in the USA: perrn; ss; ODs@wkap corn
Permissions for books published in Europe: permissions@wkap.nl
Printed on acid-free paper.
Dedication

This book is dedicated to my wife Dalma; without her love,


inspiration and support this work would not have been possible.
Contents

Chapter 1 Introduction 1

1.1 History 1
1.2 Advantages and Issues 7
1.3 Applications 10
1.4 Summary of Book 16
1.5 Bibliography 19

Chapter 2 Fundamental Properties of Single


Layer Microstrip Patch Antennas 21
2.1 Introduction 21
2.2 General Theory of Operation and Design Tools
22
2.3 The Effect of Conductor Shape 26
2.4 Impedance and Radiation Performance of Single
Layer Patches 31
2.5 Excitation Methods of Microstrip Patches 35
2.6 Circular Polarization Generation 60
2.7 Summary 64
2.8 Bibliography 65

Chapter 3 Enhancing the Bandwidth of


Microstrip Patch Antennas 69

3.1 Introduction 69
3.2 Intuitive Procedures 70
3.3 Horizontally Coupled Parasitic Patches 75
3.4 Stacked Patches 83
3.5 Large Slot Excited Patches 111
3.6 Aperture Stacked Patches 123
3.7 Ultra-wideband ASPs 150
3.8 Summary 162
3.9 Bibliography 164

Chapter 4 Improving the Efficiency of Microstrip


Patch Antennas 167

4.1 Introduction 167


4.2 Surface Waves 168
4.3 Patches that do not Excite TMo Surface Waves
169
4.4 Hi-Io Stacked Patches 178
4.5 Photonic Band-gap Structures 182
4.6 Summary 193
4.7 Bibliography 194

Chapter 5 Small Microstrip Patch Antennas 197


5.1 Introduction 197
5.2 Shorted Microstrip Patches 202
5.3 Further Size Reduction Techniques for Shorted
Patches 216
5.4 Winged Shorted Patch 219
5.5 Shorted Spiral Patches 226
5.6 Improving the Performances of Shorted
Microstrip Patches 231
5.7 Performance of Shorted Microstrip Patch
Antennas for Mobile Communications Handsets
at 1800 MHz 262
5.8 Summary 272
5.9 Bibliography 274

Chapter 6 Direct Integration of Microstrip


Antennas 277

6.1 Overview for Requirements for Integration 277


6.2 Slot Coupled Procedures and Solutions 283
6.3 Direct Contact Procedures and Solutions 309
6.4 Summary 323
6.5 Bibliography 324

Chapter 7 Microstrip Patch Arrays 327

7.1 Introduction 327


7.2 Series Fed Arrays 328
7.3 ParaUel Fed Arrays 330
7.4 Combination Fed Arrays 361
7.5 Large Scanned Arrays of Microstrip Patches 364
7.6 Alternatives to Large Arrays of Microstrip
Patches 382
7.7 Wraparound Patch Antenna Arrays 394
7.8 Summary 407
7.9 Bibliography 409

Chapter 8 Summary 413

8.1 Overview 413


8.2 Future Directions of Microstrip Patch Technology
413
8.3 Bibliography 414
List of Contributors
When presenting a large volume of work, there are always going to be many
contributors, other than the main author. Perhaps the easiest means of
acknowledging all their efforts is by listing the people on a chapter-by-chapter
basis:

Chapter 1: Dr R. Waterhouse

Chapter 2: Dr R. Waterhouse, Dr D. Novak, Dr D.-K. Park, Dr Y.Qian


and Prof. T. Itoh

Chapter 3: Dr R. Waterhouse, Dr. J. T. Aberle, Dr D. M. Kokotoff, Dr


A. Mitchell, Dr M. Lech, Dr S. D. Targonski, Mr M. Lye, Dr
F. Zavosh, Dr K. Ghorbani, Dr D. Novak, Dr A. Nirmalathas
and Dr C. Lim

Chapter 4: Dr R. Waterhouse, Mr D. Pavlickovski, Dr D. M. Kokotoff


and Dr J. T. Aberle

Chapter 5: Dr R. Waterhouse, Dr H. K. Kan, Dr D. M. Kokotoff, Dr S.


D. Targonski, Dr J. T. Rowley and Mr D. Pavlickovski

Chapter 6: Dr R. Waterhouse, Dr W. S. T. Rowe, Dr D. Novak, Dr A.


Nirmalathas and Dr C. Lim

Chapter 7: Dr R. Waterhouse, Dr K. Ghorbani, Dr W. S. T. Rowe, Dr S.


D. Targonski, Mr L. Mall, Dr H. K. Kan, Dr D. Novak, Dr A.
Nirmalathas and Dr C. Lim

Chapter 8: Dr R. Waterhouse
Acknowledgements
Firstly, I would like to thank all my colleagues and former students who
directly helped contribute to the material presented in this book. These people
are listed on page x and I very much enjoyed all the interactions we had over
the years. I would also like to thank all my technical colleagues who
contributed indirectly to this book by providing feedback on work presented
at conferences or reviewed in journals, or just discussed at technical meetings.
I would also like to thank members of the antenna community whose work
over the years inspired me to undertake some of the work presented herein. A
special thank you goes to the professorial staff from the University of
Massachusetts and their former graduate students for their kindness
throughout the years.

I would like to thank RMIT University for supporting the activities that led to
the body of work presented within this book. I would also like to thank the
institutions that helped fund much of this work, including the Australian
Research Council, CSIRO Australian National Telescope Facility, British
Aerospace and the Australian Army Agency.

I would also like to thank my family and friends whose support over the years
is much appreciated. Special mention must be made to two wonderful
woman: Ilona and Margaret for all their love and devotion to Dalma and
myself over the years. And finally a special thank-you goes to Renee for
suggestions on the title of the book - maybe next time.
Chapter 1 Introduction

1.1 History

Often described as one of the most exciting developments in antenna


and electromagnetic history, the microstrip patch antenna has matured into
probably the most versatile solutions to many systems requiring a radiating
element. Microstrip patch antennas fall into the category of printed antennas:
radiating elements that utilize printed circuit manufacturing processes to
develop the feed and radiating structure. Of all the printed antennas,
including dipoles, slots and tapered slots, microstrip patches are by far the
most popular and adaptable. This is because of all their salient features:
including ease of integration, good radiation control and low cost of
production.

The expression 'patch' is derived from the shape of the printed


conductor of the antenna: traditionally rectangular or circular. Figure 1.1.1
shows a photograph of a rectangular patch antenna (the patch conductor is
under the gray square. Microstrip patch antennas are typically resonant in
style as opposed to traveling wave and therefore are characterized by being
quite efficient over a relatively narrow operation bandwidth. Having said this,
over many years of research and development, some of the more advanced
forms of microstrip patches have responses that are more aligned to
characteristics of traveling wave antennas: of course at the expense of the
simplicity of the antenna.

Figure 1.1.1 Photograph of a Microstrip Patch Antenna (with coverlayer)

R. B. Waterhouse, Microstrip Patch Antennas: A Designer’s Guide


© Springer Science+Business Media New York 2003
Research and development on the topic of microstrip patch antennas
has been undertaken for the past 30 or so years with contributions from many
companies, government organizations and universities throughout the world.
It is the author's opinion that because of such a diverse group of institutions
undertaking to solve some of the inherent issues related to microstrip and
printed antennas in general, that such advancements in this technology could
be made. One underlining key as to why so many organizations could
contribute to this comes down to the associated costs with developing this
form of antenna, which are extremely low. Thus unlike the areas of
developing MEMs and photonic devices, almost any professor and their
students can afford to manufacture these antennas. Honorable mention should
be made of the following research and development groups throughout the
world who have made major contributions to the area of microstrip patch
technology: The University of Massachusetts; Polytechnic of Switzerland;
Hanscom Air Force Base; Ball Aerospace; Japan; and the University of
Birmingham. Of course there have been many other institutions that have
made significant contributions to this field and the author acknowledges that
he has probably offended them by not giving the credit that is due: I humbly
apologize. Below is a summary of probably the more pertinent developments
in the history of microstrip patch technology:

1970s

In the mid-1970s probably saw the first real contributions to the area
of microstrip patch technology. Although there were some patents on this
subject filed many years beforehand, it was really in this timeframe when the
first significant advancements were made. In particular, two styles of
exciting/feeding a microstrip patch antenna were developed: the edge-fed
patch and the probe fed patch [1]. These two types of excitation methods are
really the 'god fathers' of all patches, highlighting many of the inherent
advantages of microstrip patch antennas and unfortunately some of their
disadvantages too. Despite the many cynical researcher and developer (there
are still some people today who do not see the all the potential glory of the
microstrip patch!), the early pioneers continued in their endeavors developing
patches of all shapes, linear and planar arrays of these elements [2], both large
and small and even microstrip patches that could be shaped to reside on the
curved surface of a missile: a 'wrap-around' antenna. Albeit, all these
developments involved printed antennas of extremely narrow operation
bandwidth, typically 1 - 2%, not at all suitable for most communication
systems and hence the early skepticism. It is interesting to note that one of
the first attempts to enhance the impedance bandwidth, stacking patches was
proposed in the late 1970s [3], however, the method was unsuccessful until
2
the nineties, when it was fmally mastered. One key disadvantage of the
microstrip patch antenna was the relative difficulty in accurately predicting its
input impedance response. This was because of its resonant nature and the
inhomogeneous medium in which the patch resided. Despite these difficulties
two modeling techniques were introduced with respect to the microstrip patch
in the 1970s: the transmission line model and the cavity model. These
analytical tools, although limited in what they can 'accurately' model,
provided a means of understanding the behavior of the microstrip patch and
how its performance can be improved upon, to a limited extent. To this day,
no other software tools can provide the same intuition to the designer as the
transmission line and cavity models, although some of the more advanced
software packages, with their impressive visualization interfaces can provide
many clues to still unsolved issues related to the microstrip patch.

1980s

Research and development into microstrip patch technology


continued into the 1980s, with major contributions once again coming from
the defense industries, whether in the form of direct R&D or in grants.
Probably the more significant advancements came as a result of trying to
improve on the inherent issues of the microstrip patch: the most well
documented being the narrow impedance bandwidth of the patch antenna.
Both edge and probe fed style patches had impedance bandwidths less a few
percent and this value really depended on the material used as the grounded
substrate. Contrary to this the radiation bandwidth (where the gain drops by 3
dB) of these microstrip patches was quite broad in access of 30 %. This is a
fundamental reason as to why the majority of research on the microstrip patch
antenna has been conducted on ways of improving the impedance nature of
the antenna, rather than its radiation limits. One simplistic view as to why the
impedance bandwidth of edge and probe fed patches was so limited can be
explained by circuit theory. As these excitation methods are inductive in
nature, this inductance dominated the frequency response of the antenna at
low frequencies, below resonance. Now because the microstrip patch is a
resonant style antenna, at its most efficient operation, the reactance must be
close to zero and to do so the capacitive and inductive components must
effectively cancel. The increased inductive element reduced the allowable
thickness of the material, which we will see in later chapters has a direct
consequence on the obtainable bandwidth. So two important findings were
based upon finding new means to feed the antenna: the aperture-coupled patch
[4] and the proximity coupled patch [5]. The first is easy to put into context
with respect to the previous discussion: it is capacitive in nature (we will
discuss this more in chapter 2) and bandwidth enhancements were achievable.
Aperture-coupling to a patch is the magnetic equivalent of the edge- or probe
3
feeding procedures (which are often referred to as direct contact feeding
techniques). So an aperture coupled patches original bandwidth performance
was actually quite similar to a direct contact fed patch. However, the lack of a
current discontinuity in this excitation procedure later proved to very
important, especially in achieving good impedance and radiation
performance.

As a direct result in improved computer capabilities, in terms of speed


and processing ability, more sophisticated software tools were able to be
developed. Integral Equation techniques were introduced in the 1980s [6-7],
with respect to microstrip patches and these methods allowed more accurate
predictions of the antenna, in terms of its impedance response. Computational
speeds were not fast enough to allow for using these new numerical methods
to be used for design algorithms and also the tools became cumbersome when
non-conical shapes were considered. Infinite array codes were also developed
to predict the performance of large scanned arrays of microstrip patches and
scan blindness phenomenon (where power is trapped within the array and not
radiated) were studied [8-9].

Using patches in arrays were further investigated in the 1980s and


phased arrays were developed and analyzed [10]. The effects of the feed
network on large arrays and the eventual limitations on the overall size of a
patch array were explored [11]. Investigations into the radiation properties of
the microstrip patch were undertaken including means of efficiently
producing circular polarization and other types of polarization [12, 13].
Taking advantage of the ease in which microstrip patch antennas can be
integrated with microwave components was considered in the 1980s, and
several preliminary studies were undertaken. Issues were raised with respect
to the surface wave efficiency of a microstrip patch antenna that was to be
directly developed on material with high dielectric constants. Also bandwidth
enhancement procedures were further investigated and matured during the
1980s, with stacked patches: coupling a second patch to the driven element
[14], horizontal coupling patches together [15] and using a large slot to excite
an aperture-eoupled patch successfully explored [16]. By the end of the
1980s a lot of the major concerns with microstrip patch technology were well
established and the R&D committee working in this area had a good
understanding of how some of these issues could be resolved. One thing for
sure was that to get around some of the inherent problems of the microstrip
patch a more complicated structure would be needed.

4
1990s

The 1990s saw the development of microstrip patch technology into


commercial systems. One primary application was for mobile communication
base station terminals. Several versions of the microstrip patch were
developed depending on the company, all with different capabilities and
performance specifications. Most were developed using sheet metal
manufacturing techniques, so the purest may argue that these antennas were
not true printed antennas. However, despite this, it had become apparent that
the commercial applications were now truly driving the design. In some
respects the years of hype that had surrounded the microstrip patch were now
forcing patch antenna designers to 'deliver the goods'.

Some excellent development really dominated the microstrip patch


landscape in the 1990s and associated with this, which is typically the case,
some very good research resulted. Sophisticated analytical tools were now
available as software packages and there was completion between software
houses. The integral equation technique based codes were fast enough that
designers could use them to yield minimal design iterations. Also Finite
Difference Time Domain (FDTD) (for example [17]) and Finite Element
Method (FEM) [18] were introduced to analyzing microstrip patch problems
with the added feature of being able to analyze the surrounding environment.
At this stage though these tools were too slow to use as a true designer
software package.

Bandwidth enhancement techniques were further investigated in the


1990s and designs were created that started to push the radiation
limitslbandwidth of the patch antenna. Designs with impedance bandwidths
approaching an octave (67 %) were designed and produced [19]. One cannot
highlight the significance of fast, accurate computational tools in achieving
these results. Had the structures changed so much from the printed antennas
proposed in the 1970s and 1980s? Not really: the fundamental structures
were very similar and it really comes down to something pretty elementary.
Although transmission line and cavity models can give useful, simple trend
information for a microstrip patch, something more sophisticated is required
to get a true understanding of these rather complicated radiators. And by this
stage stacked elements excited with large slots were being incorporated: too
complicated for a simple analytical tool.

Other advances in the 1990s included the initial investigation of


highly efficient printed antennas that had reduced surface wave effects [20];
further exploring methods to integrated microstrip patches with MMIC
technology and also examples of this [21]; microstrip patches that could
5
operate over multiple bandwidths (very useful for providing a solution that
can be used for different communication systems without redesigning the
entire antenna) [22]; and very high gain printed antennas based on patch
antenna technology, referred to as reflectarrays [23]. These antennas combine
the gain advantages of using a reflector configuration with the manufacturing
ease of printed antenna technology to provide the required phase distribution.

Another area in microstrip patch technology that received quite a bit


of attention during the 1990s was the concept of size reduction of the patch
conductor (for examples, [24, 25]. The explosion of the wireless industry
really pushed this R&D. Companies and clients were wanting small,
unobtrusive antennas in their communication terminals and once again the
touted advantages of microstrip patches were ringing loudly in everyone's
ears. Of course a conventional microstrip patch antennas is not suited to such
an application at low microwave frequencies, simply due to its resonant
nature and the overall size required to be resonant at frequencies less than 2
GHz. And so techniques were undertaken to reduce the size of the patch
antenna: albeit at the price of performance.

2000s

By the tum of the century microstrip patches were pretty well being
used in most 'free-space ' communication systems or at least one of the first
considered for them! (Probably with the exemption of multi-octave radar and
earth stations) Further enhancements in bandwidth have been achieved (some
using variations of the excitation method). More and more examples of
integration are surfacing, including for spatial power combiner applications
[26], as well as other high gain solutions that had minimal feed network loss.
Utilizing Photonic Bandgap (PBG) structures to enhance the efficiency and
improve the overall radiation performance of the patch antenna have been
explored [27]. Attempts have been made to use optimization procedures (for
example, genetic algorithms) to ensure the 'optimum' performance could be
achieved [28], [29].

One very important advancement has occurred now which is really a


consequence of the R&D that has been undertaken over the past 20 or so
years: a good understanding of the performance and limitations of the
microstrip patch antenna has been developed. The latest exotic designs could
not have been undertaken without such an understanding. It is interesting to
see that as the limitations are pushed for these radiators, they start to share
characteristics of other antennas: the solutions almost become hybrid
antennas. For example, a shorted patch antenna (using shorting pins to reduce
the size of the patch conductor [24]) in its limit is similar to an inverted F
6
antenna [30]. An Aperture Stacked Patch (used to obtain impedance
bandwidths in excess of an octave) with a reflector patch (used to reduce
backward directed radiation) [31]: is it a printed, vertical version of a Yagi-
Uda log periodic antenna? Thus it is apparent that the microstrip patch
designer can learn from more classical styled antennas and may be these other
antenna designers can learn new things from the patch developers.

1.2 Advantages and Issues

Now that a critique of the R&D that has impact the area of microstrip
patches has been given, its appropriate to give the advantages and demerits of
this form of radiator in some detail. Of course this list is dynamic and so
some issues have resolved, at the expense of some other feature. Having said
this, these disadvantages will still be listed here, with their possible
resolutions.

Advantages

Size and Profile: In its conventional form, the volume of microstrip


patch antenna is relatively small compared to other radiators. Figure 1.1.2
shows a schematic diagram of a rectangular microstrip patch (on the left) and
a circular patch. Here the resonant rectangular patch of length, L and width,
W is etched on a grounded substrate, with a dielectric constant, e, and height,
d. The single layer solution use substrates less than 0.05 Ao (where Ao is the
free space wavelength) and multi-layered broadband solutions (stacked
patches) have a combined layer thickness of no more than 0.1 Ao. So a
microstrip patch or a variant of it can be mounted close to a ground-plane and
perform its functions (unlike a wire dipole). Thus the volume occupied by a
microstrip patch antenna (to perform satisfactory operation) is smaller than
most wire and electrical aperture (horn) antennas. The minimal thickness of
the material, or profile allow the microstrip patch to be easily integrated into
the skins of various objects, for example the fuselage of an airplane or missile,
or even a computer.
Po<'"

""' ...... "robo foot


101'- llooloan<

--
Stba.!,a:.

rr ~
_v..
e,
GI ~P tarlO

Figure 1.2.1 Conventional microstrip patch antenna: a) square edge-fed, b) circular


probe-fed
7
Ease of Manufacturing, Integration and Low Cost: As mentioned previously,
the microstrip patch can readily be developed using standard printed circuit
etching techniques, which should equate to low manufacturing costs. It is
interesting to note that these costs were considered relatively high for
traditional antenna companies trying to make the conversion to printed
antennas. Also if air-filled patches (no dielectric substrate) are being
considered, standard sheet metal fabrication procedures can be incorporated.
For an edge-fed solution, it is possible to develop the whole antenna,
including the feed network in one process. Microstrip patches are easily
integrated with their feed networks, unlike wire antennas or waveguide based
antennas. Thus promoting the ease in which these antennas can be developed
with the active devices of the terminal creating a possible one-board solution.
This, however, would be extremely difficult to achieve, or at least would give
a non-optimal performing terminal. It has in the past been deduced that
because the manufacturing processes were greatly simplified compared to
other antennas and that the material used were just sheets of dielectric
laminates, the over cost of developing a microstrip patch is low. This is
indeed the case, however in its early history the material costs of the patch
antenna were deemed expensive. In fact even in the mid-1990s it was still
considered expensive, only because most companies were trying to maximize
returns. Inexpensive materials such as standard printed circuit board,
commonly referred to as FR4, can be used if the operation frequency is less
than 1 GHz. Much beyond this the dielectric losses can become high
compromising the overall performance of the antenna.

Ease offorming arrays: Microstrip patch antennas are considered as medium


gain radiators (typically less than 8 dBi) and so for applications where more
gain is required, arraying techniques must be applied. Fortuitously, it is a
relatively straightforward procedure to form an array of microstrip patches.
This is fundamentally due to the feed network for a single microstrip element
being developed using printed circuit technology and so incorporating more
feed lines for more radiating elements is a ,simple extension. As microstrip
patches are basically 2-dimensional antennas (ignoring the height of the
materials used), it is possible to fabricate all the array feed network on one
layer and the radiating elements either on that layer too or on a layer in close
proximity to it. The final coupling from the feed to the radiating elements can
be done in a variety of ways: via probes, slots or just electromagnetic
coupling. So unlike arrays of hom antennas or even wire antennas where in
both cases the feed layout can be extremely complicated (effectively 3-
dimensonal), the cost of developing an array of microstrip patch antennas
does not increase substantially from that of a single element.

8
Efficient: Microstrip patch antennas are basically efficient radiators.
This can be attributed to one simple fact: they are resonant-style radiators.
Resonant antennas are more efficient than their broadband traveling-wave
style counterparts. Compared to other resonant antennas microstrip patches
are generally reasonably efficient, although this will depend on whether good
microstrip procedures are held to in the design procedure. There are three
loss mechanisms that need to be addressed in printed antenna designs:
conductor loss; dielectric loss and surface wave loss. All are material
dependent and so careful selection of the material can lead to good overall
efficiency. The efficiency of the microstrip patch is dependent also upon
what feeding procedure is adopted: direct contact techniques are more
efficient than the non-contact methods.

Disadvantages

Impedance Bandwidth: In their conventional form, the impedance bandwidth


of a microstrip patch antenna is narrow, limited to values below a couple
percent of the operating frequency. This can be attributed to two factors: it's
a resonant style antenna (and so the condition of efficient resonance is only
satisfied over a narrow band of frequencies) and it has a thin thickness,
typically less than 0.05 ~ (where ~ is the free-space wavelength of the
operating frequency). Such a narrow bandwidth would prohibit the use of
microstrip patch antenna technology is a variety of applications. However, as
was mentioned previously there have been several procedures developed over
the past few decades to overcome this limitation and these will be discussed in
a subsequent chapter.

Excitation ofSurface Waves: Due to the presence of the dielectric


substrate, conventional microstrip patch antennas will always excite a TMo
surface wave (unless the material used is air). This excitation can lead to
decreased efficiency (the amount depends on the material used) and also
coupling of the launched surface wave to other aspects of the antenna and
feed network. The surface waves can diffract off the finite sized ground-plane
of the antenna causing increased cross-polarization levels and even 'scalping'
of the co-polarized radiation pattern. On the feed network aspect and if a
dual-polarized printed antenna is considered, the surface wave can reduce the
isolation between the different polarization feed networks compromising the
overall performance of the antenna. Once again, several procedures have
been developed that mitigate this inherent problem and these will be
addressed later.

9
Size (again!): The size of a microstrip patch antenna appears in
both its advantages and disadvantages, simply because there are applications
where even this form of printed antenna is too large. These applications are
associated with the wireless industry, in particular for the development of
hand-held communication terminals operating at frequencies below 2 GHz.
As a microstrip patch is a resonant antenna, it must have a length comparable
to half a guided wavelength at the relevant operation frequency. Of course at
low microwave frequencies and for small volumes, as required for handset
terminals, there is not a simple solution to this problem using microstrip patch
technology. Over the years several techniques have been developed and are
currently in use for commercial mobile communication handsets and these
will be presented in a later chapter.

Radiation Performance:Although the radiation performance of the microstrip


patch is reasonably well behaved and to achieve dual polarization is relatively
straightforward compared to other antennas, there are cases where microstrip
technology cannot perform adequately to meet the requirements of the system.
For example, if 50 dB side lobe levels are required for an array of microstrip
patches, it is very difficult to meet this, simply due to the nature of the
antenna. Being a resonant antenna as well as exciting surface waves leads to
difficulties in fine resolution control of the radiation patterns. Having said
this, such high performance requirements typically only apply to a selected
few military systems and so this form of printed antenna can still find many
an application.

1.3 Applications

As mentioned before microstrip patch antennas are being used in a


large variety of applications, however this was not always the case. Because
of the inherent limitations of this printed antenna, in its early days of
development it was only considered for a hand-full of systems. It wasn't until
the mid-1990s that the microstrip patch (a somewhat mutant version of the
original form) began to be universally accepted as serious competitor to over
antenna technologies.

Radar (Fixed frequencyllow bandwidth Radar and low cost detection


systems): Most early applications of the microstrip patch antenna took
advantage of its salient features, in particular the small volume of the antenna
and the low manufacturing cost. Of course the applications were typically not
mainstream (communication systems) as the inherent disadvantages (the most
compelling was the very limited bandwidth) would not permit them to be used

10
in more prevalent systems. Fixed frequency/low bandwidth Radar
applications were such examples where microstrip patch technology could
readily be incorporated. Several spacebome systems were launched using
microstrip patches as the fundamental radiator to monitor environmental
issues on Earth [32]. The light-weight feature of printed antenna technology
was very advantageous reducing the payload of the launched satellites.
Figure 1.3.1 shows an example of a microwave patch array used for this
application.

More recently low cost motion detection systems and tagging systems
have been developed using microstrip patch arrays. These systems do not
require large bandwidths and probably the most important issue is to keep the
cost of development low. As microstrip technology allows for the antenna to
be easily integrated with the feed network and any active devices (surface
mount oscillators, mixers and detectors), patch antennas are an obvious
choice. Also, these systems tend to be in the moderate microwave frequency
bands (5 GHz - 10 GHz) and so the overall sizes of these antennas are quite
small, although there has also been many millimeter-wave printed antennas
developed for collision avoidance systems. Figure 1.3.2 shows an example of
a microstrip patch array developed for this application.

Figure 1.3.1 Exampleof MicrostripPatch Arrayused for SAR Applications

11
Figure 1.3.2 MicrostripPatch Array for Collision Avoidance System

Missiles: One of the first applications of the microstrip patch was for
telemetry systems communicating with missiles or other forms of projectiles.
The requirements for such antennas on the missiles were/are: low proflle and
conformal nature (they could not impact the aerodynamics of the projectile)
and a broad radiation pattern (to maximize communication spatial coverage).
Thus from the previously given characteristics of the microstrip patch it is
apparent that this antenna can easily conform to these specifications.
Microwave dielectric layers used in the construction of microstrip patches are
usually soft laminates and can easily be molded to comply the skin of a
missile. An added feature of the microstrip patch specifically applicable to
military systems is that it has a low impedance bandwidth, or large
electromagnetic signature. This means that it only operates effectively at a
very narrow range of frequencies and therefore is less susceptible to possible
jamming frequencies or interference. The initial telemetry applications were
in the mid to late 1970s and variations of the 'wrap-around' patch are still
being used today. A photograph of a wraparound microstrip patch antenna
array is shown in Figure 1.3.3.

12
Figure 1.3.3 TelemetryMicrostripPatch Antenna

Aircraft: Microstrip patch antennas are now being used on commercial


aircraft for providing passengers with communication services via satellite
systems. This has only come to light recently due to the higher frequencies
(greater than 5 GHz) being assigned for such services. The conformal nature
of the patch antenna (see the previous subsection) makes an array of patch
elements an obvious choice. It should be noted that most antennas on aircraft
for pilot communication and tracking purposes do not use printed antennas
simply because the frequencies of operation are too low for these services.
Antennas that incorporate the fuselage of the aircraft as part of the radiating
element are used to increase the overall size of the antenna and so allow the
antenna to operate at low operating frequencies. A photograph of a rnicrostrip
patch array and its appropriate phasing circuitry used for an aircraft
communication service is ziven in Figure 1.3.4.

Figure 1.3.4 AirborneMicrostripPatch Phased Array


13
Satellite Communications: Although microstrip patches found use in
satellite systems early on in its history, it wasn't until the early 1990s that this
form of printed antenna was to be incorporated into satellite communication
services as the 'main mission' antenna. A key example of this was the
Iridium communication system [33]. Although this system is now defunct, it
facilitated many important advancements in communication technologies: one
being able to develop a network of microstrip patch antennas that could
satisfy the bandwidth and radiation performance necessary to make the
technology viable. Most satellite communication antennas now use (or
propose to use) microstrip patch elements. A schematic diagram of the main
mission antenna for one of the Iridium satellites is given in Figure 1.3.5.

Figure 1.3.5 Iridium Main Mission Printed Antenna

Mobile Communication Base Stations: Probably one of the most common


applications today for the microstrip patch is for mobile communication base
stations. Like for satellite communication systems, here it is only recently,
since the mid to late 1990s that a variant of the microstrip patch antenna could
satisfy the stringent requirements of a base station terminal. Now there are a
variety of microstrip patches used for this application designed and developed
by different companies. Each approach has its relative features, although
overall they generally provide the same performance. For mobile systems at
near 900 MHz there are probably as many wire antenna solutions as patch
antenna versions. However, for higher frequency systems, such as 30,
microstrip patch base station terminals are much more prevalent. A
photograph of an array of microstrip patches used as a mobile
communications base station terminal is provided in Figure 1.3.6.

14
Figure 1.3.6 Mobile Communications Base Station Antenna

Mobile Communication Handsets (and applications of limited real-estate):


One of the most recently considered applications for printed antennas and in
particular microstrip patches are for communication systems handset
terminals . This is to facilitate the requirement of customers for an
aesthetically pleasing handset terminal with no visible antenna. As microstrip
antennas can easily be integrated with active and passive components and are
also conformal, it was deemed that microstrip patches could be used for this
application. The main issue, however, is the size of the antenna. For
operation in the low microwave frequency spectrum, a conventional patch
antenna is just too large to be accommodated for on a handset terminal. So
since the mid - 1990s much attention has focused on developing patches are
small in size. Several mobile communication handset terminals today use
variants of the microstrip patch antenna as their radiator. Of course these
antennas have minimal resemblance to the original microstrip patch antenna
of the early 1970s, however, their roots can be traced back to these antennas.
As the operation frequency of mobile/wireless services increases, more
conventional-styled patches could be used for the handset terminals. A very
important issue here is the compromise in performance that has to be
accommodated for because of the major size reductions required. A typical
handset with internal antenna is provided in Figure 1.3.7.

15
Figure 1.3.7 Internal Antenna for Cellular Systems

Another application receiving attention is the utilization of patch


antennas for biomedical telemetry services. Once again because of the small
size and the ease of integration with microwave components, the microstrip
patch is an obvious choice. Here because often the communicating sensor
may be located within the skin of the user, the size of the antenna must be
minimized. Fortuitously, such applications require relatively narrow
bandwidths, compared to mobile communications systems.

1.4 Summary of Book

The objectives of this book are threefold: to give the reader


techniques that have been incorporated over the years to solve some of the
inherent problems of microstrip patch antennas; to provide the reader with
parameter studies and performance trends of these variations of the microstrip
patch antenna which will hopefully give the reader insight into how these
antennas operate; and to give the reader some examples of microstrip patch
antenna designs that have been utilized in systems and how these designs
were derived. It should be noted that this book tries to provide a designer's
prospective of microstrip patch antennas and so it is not a typical book on this
subject. Firstly the reader will notice there are not many equations throughout
this book related to the design of microstrip patch antennas, nor is there a
chapter dedicated to the analysis of microstrip patches. The reason for this is
quite simple: apart from very fundamental microstrip patch antennas (patches
mounted on single layer, thin substrate material), simple design equations do
not hold for an accurate representation of the performance of the antenna.
Unfortunately. to achieve performance goals typically required in a system
requiring a radiating element, the microstrip patch needs to be modified from
its inherently simple form of the early 1970s, into something more
complicated. Thus a full-wave analysis tool is required to model this antenna.
So the key to good microstrip patch antenna engineering is a fast, versatile
and accurate analysis tool. As was mentioned earlier, there are many software

16
packages available today that accurately analyze these printed antennas and a
microstrip patch antenna designer will require one of these to successfully
design their antenna . Having said this, the author encourages the reader to
develop their own analysis tool based on these well-established techniques to
get a better appreciation of what is involved in developing these somewhat
complicated software tools. Also fundamental theories, such as the cavity
model and the transmission line model should be consulted but will not be
presented here in their entirety. As stated before, one of the objectives here is
to give a designer's prospective of microstrip patch antennas.

Chapter 2 lays the groundwork for the microstrip patch' antenna


designer. Here the single layer patch is descriptively analyzed in terms of
how the various parameters of the antenna affect its overall performance.
Patch conductor shape is examined and the effects of the dielectric properties
of the substrate are investigated. A very brief critique is given on numerical
analyses tools that can be used to analyze microstrip patch antennas and the
pros and cons of each style of tool are given. In this chapter the different
excitation methods for the microstrip patch antenna are examined and advice
is given as to when each technique should be used based upon the advantages
and disadvantages of each procedure.

The remaining chapters are dedicated to methods used to make this


form of printed antenna compliant with typical specifications required in
communication systems of today. For each method performance trends are
given to hopefully give the reader a feel as to what parameters are important
for the various designs. Of course solving some of the inherent problems of
microstrip patches tends to open new 'cans of worms' and these new
difficulties the designer may encounter are addressed for all the designs
presented . Chapter 2 also looks into procedures to generate circular
polarization (CP) from a microstrip patch antenna. Here the advantages and
issues with the three common methods , single feed, dual feed and
synchronous subarrays are investigated and design examples of each
procedure are presented.

Chapter 3 introduces the techniques used to enhance the impedance


bandwidth of the microstrip patch antenna. The somewhat obvious, intuitive
procedures of a single layer design are touched upon and then methods to
truly enhance the bandwidth performance of the antenna are presented. These
include using horizontally coupled parasitic elements, stacked patches, large
slot coupled patches and finally aperture stacked patches. This final version
of the humble patch has been shown to give impedance bandwidths in excess
of an octave! A procedure will be shown when this bandwidth can even be
further advanced. Of course, as mentioned before, 'what you gain in the
17
straights, you lose in the roundabouts' and so issues related to these
bandwidth enhancement techniques are discussed and some possible solutions
to these problems are suggested and investigated. Chapter 3 is probably the
most applicable chapter to current day designs of the single radiating element,
for say base station antennas, satellite communication antennas and radar
systems. Design trends to give insight into the performance of each modified
patch antenna are provided and for the case of the stacked patch an
investigation into what appears to be the optimum dielectric constant material
selection are provided.

Chapter 4 is dedicated to methods used to improve the efficiency of


the microstrip patch antenna. Techniques that enlarge the size of the patch
conductor to ensure the fundamental surface wave mode is weakly excited are
presented. Also a simple technique that gives other unexpected advantages is
summarized and the details on how to design this antenna (and the previous
cases) are also provided. The concept of using photonic bandgap structures to
improve the efficiency of the antenna is briefly investigated.

In chapter 5 small microstrip patches are investigated. The objective


of this chapter is to provide the reader with techniques used to reduce the size
of the patch conductor to make it more compliant with applications requiring
minimal antenna real estate. In particular we focus on the shorted patch
structure and several of its variants . Of course there are several issues with
using the shorting post procedure and these are discussed in this chapter.
Methods are then introduced to overcome some of the inherent problems of
small microstrip patch antennas, including increasing the impedance
bandwidth, reducing the cross-polarized fields and increasing the radiation
efficiency (or gain). Finally techniques for developing circularly polarized
versions of small patch antennas are examined. .

Chapter 6 investigates and presents material on how microstrip patch


antennas can readily be integrated with microwave active circuitry and
photonic devices. This chapter importantly summarizes procedures that are
and will be used to provide compact mobile communication base station
modules of the future . In particular the focus of this chapter is to provide
efficient, broadband antenna solutions. Two procedures are given both with
their relative merits and issues. Procedures to generate circularly polarized
and dual linearly polarized integrated antennas are also presented.

Chapter 7 presents the variety of methods of developing arrays of


microstrip patches and design examples of each procedure. Details of arrays
designed and developed in the low microwave frequency spectrum as well as
ones designed for use at millimeter-wave frequencies .are given . A procedure
18
is provided for designing wraparound arrays and also an investigation of a
linear phased array is given. A brief examination on what design parameters
that need to be varied from when designing a broadband single stacked patch
to a large array of stacked patches is presented, as well as for the case when
aperture stacked patches are used. Finally printed alternatives to large arrays
of microstrip patches are briefly examined.

In chapter 8 an overview of this book is given and a summary of


possible future directions into the research and development of microstrip
patch antennas is given. Microstrip patches are finally becoming one of the
main antenna alternatives or most free-space systems and with the
development of more sophisticated fabrication procedures and elaborate
applications it appears that these radiators will be around for a long time.

1.5 Bibliography
[1] R. E. Munson, "Conformal Microstrip Antennas and Microstrip Phased Arrays,"
IEEE Trans. Antennas Propagat., vol. AP-22, pp. 74 - 78, January 1974.
[2] R.1. Mailloux, 1. F. Mcllvenna and N. P. Kemweis, "Microstrip array technology,"
IEEE Trans. Antennas & Propagation , vol. 29, pp. 25 - 37, Jan. 1981.
[3] S. A. Long and D. M. Walton, "A dual-frequency stacked circular-disc antenna,"
IEEE Trans. Antennas Propagat., Vol. AP-27, pp. 270-273, March 1979.
[4] D. M. Pozar, "A Microstrip Antenna Aperture Coupled to a Microstrip Line,"
Electronics Letters, Vol. 21, pp. 49 - 50, January 1985.
[5] D. M. Pozar and B. Kaufman, "Increas ing the Bandwidth of a Microstrip Antenna by
Proximity Coupling," Electronics Letters, Vol. 23, pp. 368 - 369, April 1987.
[6] 1. R. Mosig and F. E. Gardiol, "General Integral Equation Formulation for Microstrip
Antennas and Scatterers,' Proc. lnst. Elect. Eng., pt. H, Vol. 132, pp. 424 - 432,
1985.
[7] D. M. Pozar, "Input Impedance and Mutual Coupling of Rectangular Microstrip
Antennas," IEEE Trans. Antennas Propagat., Vol. AP-30, pp. 1191 - 1196,
November 1982.
[8] D. M. Pozar and D. H. Schaubert, "Scan blindness in infinite arrays of printed
dipoles ," IEEE Trans. Antennas & Propagation, vol. 32, pp. 602 - 610, June 1984.
[9] J. T. Aberle and D. M. Pozar, "Analysis of infinite arrays of probe fed rectangular
microstrip patches using a rigorous feed model", Proc. Inst. Elec. Eng., Pt. H, vol.
136, pp. 110-119, April 1989.
[10] D. L. Rascoe , et al, "Ka-band MMIC beam-steered transmitter array", IEEE Trans.
Microwave Theory & Techniques, vol. 37, pp. 2165 - 2168, Dec. 1989.
[11] E. Levine , G. Malamud, S. Shtrikman and D. Treves, "A study of microstrip array
antennas with the feed network," IEEE Trans. Antennas & Propagation, vol. 37, pp.
426-434, April 1989.
[12] 1. Huang, "A Technique for an Array to Generate Circular Polarization with Linearly
Polarized Elements", IEEE Trans. Antennas Propagat., AP-34, pp. 1113 - 1124,
September 1986.
[13] P. S. Hall, 1. S. Dahele and 1. R. James, "Design Principles of Sequentially Fed Wide
Bandwidth, Circularly Polarized Microstrip Antennas," IEE Proc. H, vol. 136, pp.
381- 389, October 1989.

19
[14] S. Sabban, "A New Broadband Stacked Two-Layer Microstrip Antenna," 1983 IEEE
Antennas Propagat. Symposium, pp. 63 - 66, June 1983.
[15] Kumar and K. C. Gupta, "Non radiating edges and four edges gap-coupled multiple
resonator broad-band microstrip antennas", IEEE Trans. Antennas Propagat., Vol.
AP-33, pp. 173-178, February 1985.
[16] J.F. Zurcher, 'The SSFIP: A global concept for high performance broadband
planar antennas", Electronics Letters, vol. 24, pp. 1433-1435, Nov. 1988.
[17] XFDTD, Remcom: www.fdtd.com
[18] J.-M. Jin and J. L. Volakis, "A Hybrid Finite Element Method for Scattering and
Radiation by Microstrip Patch Antennas and Arrays Residing in a Cavity," IEEE
Trans. Antennas Propagat., vol. AP-39, pp, 1598 -1604, November 1991.
[19] S. D. Targonski, R. B. Waterhouse and D. M. Pozar, "Design of wideband aperture-
stacked patch microstrip antennas", IEEE Transactions Antennas & Propagation,
vol. 46, pp. 1246 - 1251, Sept. 1998.
[20] D. R. Jackson, J. T. Williams, A. K. Bhattacharyya, R. L. Smith, S. J. Buchheitt and
S. A. Long, "Microstrip Patch Designs That Do Not Excite Surface Waves," IEEE
Trans. Antennas Propagat., vol. AP-41, pp. 1026 -1037, August 1993.
[21] Y. Qian and T. Itoh, "Progress in Active Integrated Antennas and Their
Applications", IEEE Trans. Microwave Theory and Techniques, 1998, vol. 46, no.
11, pp. 1891-1900.
[22] E. Lee, P. S. Hall, P. Gardner and D. Kitchener, "Multi-band Antennas," 1999 IEEE
Antennas Propagat. Symposium, Orlando Florida, July 1999.
[23J D. M. Pozar, S. D. Targonski and H. D. Syrigos, "Design of millimeter-wave
microstrip reflectarrays," IEEE Trans. Antennas & Propagation, vol. 45, pp. 287 -
296, Feb. 1997.
[24J R. B. Waterhouse, "Small microstrip patch antenna," Electronics Letters, vol. 31, pp.
604-605, Apr. 1995.
[25] I. Park and R. Mittra, "Aperture-coupled small microstrip antenna", Electron. Lett.,
Vol. 32, pp. 1741-1742, Sept. 1996.
[26] M. P. DeLisio and R. A. York, "Quasi-optical and Spatial Power Combining," IEEE
Trans. Microwave Theory Techn., Vol. MIT-50, pp. 929 - 936, March 2002.
[27] R. F. Jimenez Broas, D. F. Sievenpiper, E. Yablonovitch, "A High-Impedance
Ground Plane Applied to a Cellphone Handset Geometry," IEEE Trans. Microwave
Theory and Tech., vol. 49, pp, 1262-1265, Jul. 2001.
[28] H. Choo and H. Ling, "Design of Multiband Microstrip Antennas Using a Genetic
Algorithm," IEEE Microwave and Wireless Components Letters, Vol. 12, pp, 345 -
347, September 2002.
[29] A. Mitchell, M. Lech and R. B. Waterhouse, "Optimization of broadband microstrip
patch antennas," 2000 Asia Pacific Microwave Conference, APMC'oo, Sydney
Australia, pp, 711 - 714, Dec. 2000.
[30] C. R. Rowell and R. D. Murch, "A capacitively Loaded PIFA for Compact Mobile
Telephone Handset," IEEE Trans. Antennas Propagat., vol. AP-45, pp, 837 - 843,
August 1997.
[31] R. B. Waterhouse, D. Novak, A. Nirmalathas and C. Lim, "Broadband printed
sectoral coverage antennas for millimetre-wave wireless applications," IEEE
Transactions on Antennas & Propagation, vol. 50, pp. 12 -16, Jan. 2002.
[32] K. R. Carver, "Antenna Technology Requirements for Next-Generation Spacebome
SAR Systems," 1983 IEEE Antennas Propagat. Symposium, pp. 365 - 368, July
1983.
[33] J. J. Schuss, J. Upton, B. Myers, T. Sikina, A. Rohwer, P. Makridakas, R. Francois,
L. Wardle and R. Smith, 'The IRIDIUM main mission antenna concept", IEEE
Trans. Antennas & Propagation, vol. AP-47, pp. 416 -425, Mar. 1999.

20
Chapter 2 Fundamental Properties of Single
Layer Microstrip Patch Antennas

2.1 Introduction

As mentioned in the previous chapter, microstrip patch antennas, in a


variety of forms, are being used in numerous wireless communication
applications. The highlights of this radiator summarized in Section 1.2 are the
reasons as to why the microstrip patch antenna has in some cases become a
key feature of these systems, especially in the age of unobtrusive structural
requirements and minimal hardware real-estate.

Despite the hype associated with these printed antennas, microstrip


patches are relatively complicated radiating structures, in particular when
trying to enhance their generic performance to achieve the electrical and
mechanical requirements for a radiator in a commercial system. Even in its
simplest form, a single layer geometry, to accurate predict the performance of
the patch antenna is not trivial. Having said this, approximate performance
trends have been established using simplest numerical techniques and these
have given useful insight into how the basic form of the microstrip patch
operates. Of course, as the form of the patch antenna becomes more exotic,
these trends have somewhat limited use.

In this chapter we lay the foundation required for understanding the


microstrip patch antenna and some of its basic characteristics. In Section 2.2
we look at the general theory of operation of the patch antenna and also
briefly examine some methods to analyze these radiators. In Section 2.3, we
look at how the shape of the microstrip patch conductor affects the
performance of the antenna and whether there are advantages or
disadvantages and associated with the chosen conductor shape. Section 2.4
presents some fundamental trends associated with the microstrip patch
antenna, namely how its impedance and radiation characteristics vary as a
function of the material used as a substrate. These trends are extremely
important, especially in helping resolve some of the issues associated with
microstrip patch antennas. In Section 2.5 the different feeding/excitation
procedures for coupling power to and from the patch antenna are introduced
and discussed. The relative features and cons associated with each method of
excitation are given. Also design examples and procedures are provided to
help understand these forms of microstrip patch antennas. Finally in Section
2.6, methods to generate circular polarization (CP) from a microstrip patch

R. B. Waterhouse, Microstrip Patch Antennas: A Designer’s Guide


© Springer Science+Business Media New York 2003
antenna are presented . As with the feeding techniques, each CP generation
method has its relative advantages and disadvantages and these will be
discussed.

From this chapter it is hoped that the reader understands the general
properties of the microstrip patch and has some insight into the features and
issues associated with changing the original structure, whether it be changing
the feeding mechanism or the conductor shape or the material used. An
understanding of the material presented here is imperative to the techniques
discussed in subsequent chapters on how to enhance the performance of the
patch antenna . For example, there will be circularly polarized patch arrays
presented in Chapter 6 using the techniques presented here.

2.2 GeneralTheory of Operation and Design Tools

2.2-1 Introduction

A general schematic diagram of a microstrip patch antenna is shown


in Figure 2.2.1. Here an arbitrarily shaped metallic conductor is etched on a
grounded dielectric laminate . A microstrip patch antenna is a resonant-style
radiator so one of its dimensions must be approximately A/2, where Ag is a
guided wavelength taking into consideration the surrounding environment of
the printed antenna. The resonant dimension depends on the shape of the
patch conductor. Having said this it will be become apparent that the
properties of the substrate, namely its dielectric constant, e, and its height will
playa fundamental role in the performance of the printed antenna .

Patch conductor
Edge feeding

Grounded substrate

Figure 2.2.1 Microstrip Patch Antenna Concept

22
Simplistically when a voltage is applied to the feeding point the
radiator, a current (or mode) will be excited on the patch and vertical electric
fields will be generated between the patch and the ground plane. Because the
now formed slots (between the edges of the patch and the ground-plane) are
A/2 apart, the radiated fields will add constructively creating an efficient,
resonant radiator. The radiation efficiency will depend upon the material
used, as the confinement of the fields will be determined by the dielectric
constant of the substrate and its height. The efficiency will be discussed
further in Section 2.4.

From an equivalent circuit point of view, when the feeding point is


close to the open circuit edge of the patch a large voltage and a minimal
current will be seen by the feed point, resulting in a high input impedance.
When the feeding point is at the center of the patch, the input impedance will
be close to zero as the voltage is a minimum (being a Ag /4 from the open
circuit) and so the current will be a maximum, presenting a low impedance at
the input port. Thus the location of the feed can be used to control the input
impedance of the antenna.

2.2-2 Methods of Analysis

As mentioned before, the microstrip patch is a relatively complicated


radiator. Essentially it is a resonant antenna located in a non-homogeneous
environment. Over the years there have been several analytical/numerical
methods proposed and used to analyze this radiator. These methods can be
grouped as either approximate techniques or full-wave analyses. The
approximate techniques, the transmission line model and the cavity model use
simplifying assumptions to reduce the complexity of the analysis of the
antenna. These design tools are useful in giving general performance trends
of the patch antenna and under certain conditions, these techniques can be
relatively accurate. One such condition is that the substrate is relatively thin
(less than 0.02 1.0). Another condition is that the dielectric constant be
relatively low. A distinct advantage of the approximate techniques is the
speed in determining the predicted performance and hence the use of these
methods for obtaining overall trends. However, if minimal iterations of a
design are required and the performance goals are difficult for a conventional
microstrip patch to meet, then the approximate techniques should not be relied
upon .

Full-wave analyses obviously give the most accurate results as they


apply Maxwell's Equations to the problem at hand and then ensure the
boundary conditions associated with the structure are satisfied. Full-wave

23
analysis techniques can be sub-divided into two camps, depending on how
Maxwell's equations are solved: differentially or by integrating. Of the full-
wave analysis procedures, the Integral Equation approach is the most mature
with the Spectral Domain formulation and the Spatial Domain formulation
both being applied to predict the performance of a microstrip patch antenna
since the early 1980s. Over the years very good agreement between theory
and experiment has been achieved using both procedures and the degree of
accuracy really depends on the degree in which the programmer wants to
represent the structure to be analyzed. The initial problem of using the
Integral Equation procedures was the large computation time required to solve
the antenna under question. However, the speed of present day computers has
resolved this issue. As mentioned in Chapter 1 there are several commercially
available software packages based on these procedures.

Applying differential full-wave analyses to microstrip patch antennas


has really been practiced since the late 1980s. The Finite Difference Time
Domain method was the first and later the Finite Element Method. Initially
both suffered from excessively large computational times, however, these
techniques allowed a more realistic representation of the surrounding
environment of the patch antenna. For example, was the patch residing within
a cavity, or on a small handset? If so these procedures could solve such a
problem, unlike the original Integral Equation techniques that typically
required the approximation that the ground-plane and substrate extended
infinitely in horizontal plane.

Designing Microstrip Patches

From a personal perspective I have used all six analytical techniques


to design microstrip patches and here's a summary of my opinion on each.

Transmission Line Model! 1, 2]

To design a microstrip patch based on this approach is not


recommended. If by chance you are lucky enough to be given a problem
where the accuracy of this technique is reasonable (within 10% for the
resonant frequency), expect still to undergo several design iterations. For a
microstrip patch antenna designer this means using either conductive tape to
increase the size of the patch, or a scalpel, to scratch away some of the
conductor. Once this has been done, then the input impedance will need to be
adjusted using the above-mentioned techniques as the transmission line model
is not accurate here either. Realistically, the transmission line model is only
useful to give you a 'ball-park' figure on what material the designer should be
using.
24
Cavity Model [3, 4J

The cavity model is more dependable than the transmission line


model, although again expect to some "patch modifiers" (see previous
comments) to help get the necessary response. For substrate thicknesses
greater than 0.03 Ao it is strongly advised not to use either approximate
method as erroneous predictions are common. Once again the cavity model is
useful for obtaining 'ball-park' figures on what material should be used and
approximate dimensions of the patches. Typically I use either one of these
approximate techniques (preferably the cavity model) as a first design
iteration before using a more rigorous analysis tool. The cavity model does
handle non-contact feed patch solutions (such as an aperture coupled patch)
more accurately than it does for a direct contact (for example probe fed) fed
patch. This can be attributed to lack of a current discontinuity for non-contact
fed patch antennas, which the cavity model has difficulty is predicting the
associated modes.

Spectral Domain Integral Equation Method [5, 6J

Of all the analytical tools applied to the microstrip patch antenna and
its variants, the Spectral Domain Integral Equation technique is the one I am
most familiar with having formulated and developed several versions of it.
Thus my opinion may be somewhat biased, however I have found it to be
extremely useful in the design of patch antennas and accurate, when the
surrounding environmental conditions are not too adverse (such as when the
substrate needs to be truncated directly below the patch radiator). For such
conditions, "patch modifiers" will need to be applied. Close to half of the
designs presented in this book were based upon using the spectral domain
technique. One distinct advantage it has over other full-wave analysis
methods is its speed when conically shaped patches are considered. For such
cases entire domain basis functions can be used significantly reducing the
computational time. A fast analysis tool is imperative when attempting a
design, especially when it is quite complicated.

Spatial Domain Integral Equation Method [7J

The Spatial Domain formulation of the Integral Equation is an


extremely popular full-wave analysis, especially the Mixed Potential version
[7]. Software tools such as Ensemble and IE3D are based on this. It is
probably a little bit more versatile than the Spectral Domain technique and
more recently I have used these tools on numerous occasions to formulate
designs. Both software tools I have found to be extremely useful in designing
25
patch antennas, although I probably have a preference to Ensemble simply
because I've used it more often and the manuals I find easier to follow.

Finite Difference TimeDomain(FDTD)Method [8]

The Finite Difference Time Domain method can be a very accurate


tool if the time is taken to formulate the problem at hand accurately. Once
again the advantage of using this procedure is its ability to analyze usual
surroundings (such as the user of a mobile handset). For such conditions it is
not advised to use FDTD as a design tool, simply because the computation
time would be very long. Several years ago I used this technique to attempt to
design a small printed antenna and found it fairly cumbersome. However,
more recently I have not had these issues.

Finite ElementMethod (FEM)[9]

FEM had similar traits as to FDTD when it came to modeling


microstrip patch antennas in its early days of formulation. However, more
recently I have used it (in the form of HFSS [10]) to design several
complicated patch structures and have had little difficulty with it. The
differential techniques are inherently slower than the Integral Equation
procedures, however, the gained flexibility is extremely advantageous,
especially when considering integrating microstrip patches with active
devices and their housing structures.

2.3 The Effect of Conductor Shape

2.3-1 Introduction

Over the years there have been many conductor shapes proposed and
investigated for a microstrip patch antenna. Schematic diagrams of these are
shown in Figure 2.3.1 and a brief summary of the advantages and
disadvantages are given below. Here we are assuming operation only in the
lowest order mode, a condition a patch is normally operated under simply
because its size is smallest.

2.3-2 Rectangular and Square Patches

The first and probably the most utilized patch conductor geometry
were the rectangular and square shapes. Figure 2.3.1a and b show these
geometries. For the rectangular patch the antenna is excited at some point

26
along the resonant dimension, L, to generate a mode in this direction. Figure
2.3.2 shows the currents excited on a rectangular patch, in particular the
direction they are excited. In general the length of the patch controls the
resonant frequency and the width of the patch affects the impedance level at
resonance as well as the bandwidth (a second order effect): the larger width of
the patch, the smaller the input impedance of the antenna. Keep in mind this
statement is only valid under certain conditions (relatively thin substrate
material). As the thickness of the material increases to greater than 0.03 An,
these relationships are not mutually exclusive and the feeding procedure and
location can dramatically change all the measures of performance. The case
shown in Figure 2.3.2 is for a linearly polarized radiator, so the currents
should predominantly be in one direction. By moving the feed location off
the symmetry point with respect to the W dimension, currents can also be
generated in this direction. Once again, under conservative conditions, if a
square patch is considered and the feed is located along the diagonal plane,
dual polarized radiation at the frequency of operation can be generated.

(a) (b) (c) (d)

(e) (f) (g)

Figure 2.3.1 Examples of ConductorShapes for MicrostripPatches

In general, of all the conductor shapes rectangular patches tend to


have the largest impedance bandwidth, simply because they are larger than
the other shapes. Square patches can be used to generate dual or circular
polarization.

2.3-3 Circular and Elliptical Patches

Figure 2.3.1c and d show schematic diagrams of circular and elliptical


patch geometries. These shapes are probably the second most common
geometry. Circular and elliptical patches are slightly smaller than their
27
rectangular counterpart and as a result have slightly lower gain and
bandwidth. The dominant modal distribution on a circular patch is different
to that for a rectangular/square patch conductor. Figure 2.3.3 shows a typical
current distribution. Even though this current variation is somewhat different
to that presented in Figure 2.3.2, the overall effect is similar: the current
magnitude (given by the size of the arrows) is largest in one direction, the
same as for the rectangular patch case.

Figure 2.3.2 Excited Fields on a Rectangular Patch Antenna

A circular patch, like a square patch only has one degree of freedom
in its conductor shape and that's its radius. Thus changing the radius will
control the resonant frequency of the circular patch. Once again under
conservative conditions, the feed position will control the input impedance of
the antenna at the chosen resonant frequency.

One of the primary reasons the circular geometry was quite


expensively investigated in the past was due to its inherent symmetry. This
allowed full-wave analysis tools utilizing a spectral domain technique to be
written that were computationally more efficient than their rectangular
counterpart. This was important in the early stages of patch design and
development for it allowed performance trends of more complicated
structures (such as stacked patches) to be explored and optimized efficiently.
Importantly these trends could then be relayed back to other geometries,
simply because most of the differences in performances of different conductor
shaped patches are minimal. With the advent of several rigorous,
28
computationally fast full-wave design tools , such as Ensemble and IE3D,
systems incorporating circular patch antennas are becoming increasingly rare .

Figure 2.3.3 Typical Fields Excited on a Circular Patch Antenna

2.3-4 Triangular and Disc Sector Patches

Triangular and disc sector patch geometries are smaller than their
rectangular and circular counterparts, although at the expense of further
reduction in bandwidth and gain. Figure 2.3.le and f show schematic
diagrams of these conductor shapes. Triangular patches also tend to generate
higher cross-polarization levels, due to their lack of symmetry in the
configuration. Figure 2.3.4 shows the current distribution on a triangular
patch conductor. Contributions to cross-polarization fields come from
currents directed in the orthogonal direction to the main polarization. The
triangular and disc sector patches have a similar number of design freedoms
as the rectangular patch. For a disc sector antenna, if the TMot mode is
excited, then the radius of the disc controls the resonant frequency and the
sector angle controls the bandwidth and impedance. Depending on the aspect
of the triangle and the disc sector, dual frequency and dual polarized patches
can be developed using either of these conductor shapes, however the
bandwidth is typically very narrow. Also for the dual frequency patch, the
polarizations for each frequency band are orthogonal.

29
Figure 2.3.4 Typical Fields Excited on a Triangular PatchAntenna

2.3-5 Annular Rings

Annular ring geometries are the smallest conductor shape, once again
at the expense of bandwidth and gain. Figure 2.3.1g shows this conductor
shape. One problem associated with an annular ring is that it is not simple
process to excite the lowest order mode and obtain an input impedance close
to 50 Q at resonance. In fact, impedance values ranges from 150 - 250 Q are
quite normal. Non-contact forms of excitation (see Section 2.5) are typically
required to feed this element at the expense of antenna efficiency. The
symmetry issues mentioned for the circular patch cases also apply here. The
current distribution of a probe fed annular ring is shown in Figure 2.3.5.

The annular ring has one more design variable than the circular patch
and therefore its response should be easier to control. Both the inner and
outer ring dimensions can be used to control the resonant frequency of the
printed antenna, which is very advantageous. However, as the inner radius
approaches the outer radius dimension, the impedance bandwidth becomes
narrower.

30
Figure 2.3.5 Typical CurrentslFields Excited on an Annular RingPatch

2.4 Impedance and Radiation Performance of Single Layer


Patches
Figure 2.4.1 shows some fundamental, very important performance
trends of a single layer microstrip patch antenna performance as a function of
the laminate properties used to fabricate the substrate. These performance
trends represent the properties of microstrip patch antennas in their pure form
with a simple, ideal excitation method. Although the trends are for
rectangular patches, other conductor shapes have similar responses. The
impedance bandwidth (defined here as a 10 dB return loss bandwidth) for
various dielectric constant values as a function of the electrical thickness of
the laminate is given in Figure 2.4.1a. As can be seen from this plot, the
thicker the substrate material, the greater the bandwidth of the microstrip
patch antenna. An important observation from Figure 2.4.1a is that the lower
the dielectric constant, the greater the bandwidth that can be achieved from
the antenna. Please note that the trends are not continued for very thick
material due issues making microstrip patches radiate efficiently at these
thicknesses. A discussion on this is provided in Chapter 3.

The directivity of a patch antenna once again for different dielectric


constants as a function of the electrical thickness is shown in Figure 2.4.1b.
Simplistically what is observant here is that because the microstrip patch
antenna mounted on the low dielectric constant material is physically bigger
than the antenna on the high dielectric constant laminate (recall from Section
2.2 that the resonant dimension is ')../2), it has a larger collecting area and
therefore greater directivity. The directivity slightly increases as the thickness
31
increases due to the volume of the antenna becoming larger. Please note that
efficiency is not included in this plot.

10D .....---------------"""'T"-----.

~. 2.2 •••••
. .....
...... "
"......

.>
. .....
.
. ......
.......
...... '
..- ~=
-"
102_-

0.0- . : " - - - - - - - - - - 1 - - - - - - - - - - - : : - : '


0.025 0.050
s.6~ thido_, dlle

(a)

10.0 r - - - - - - - - - - - - - - - - - - - - ,

tr-ID _ t -22
-
: ! .

~ -- -- ---------
;: :; fo--------- --- -
t.-
~

t 5D f- 102
l!!

0.050

(b)

32
ID """""..-=------
~ ~ ~
------------,
"" " .
" , • • • •••• •••• +

t.=22
~ .
-,

"" -,
"" ... ...
,
M'
... ...
•~ 0.5 I- ....
~
"-
... ...
" ... E.= 10.2

"

I
0.05 0.10
sm.irate tJddmess, dI10

Figure 2.4.1 Performance trends of single layered patch antennas: (a) impedance
bandwidth; (b) directivity; (c) surface wave efficiency

There are three forms of loss associated with a microstrip patch


antenna: conductor loss, dielectric loss and surface wave loss. The first two
of these loss mechanisms depend on the quality of the material used as the
substrate. The latter is due to the characteristics of the material, namely the
dielectric constant and the thickness. Surface waves are modes of
propagation supported by the grounded substrate used to form the patch
antenna. Figure 2.4.lc shows the surface wave efficiency of a microstrip
patch antenna for several dielectric constants as a function of thickness of the
substrate. As can be seen from this figure, the higher the dielectric constant,
the more power lost to the surface wave and therefore the less the efficiency
of the antenna. Please note there are no surface waves excited for the case
when e, = 1.0. Figure 2.4.2 shows the conditions for when each surface mode
is launched for a typical microstrip patch substrate (e, = 2.55). As can be seen
for most practical cases, the only surface wave mode that needs to be
considered is the TMol mode, which is always present (unless €r = 1.0).

From Figure 2.4.1 there appears to be a fundamental problem


concerning the integration of microstrip patch antennas with MMIC
(Microwave Monolithic Integrated Circuits) and OEIC (Opto-Electronic
Integrated Circuits) technology, one of the oft-touted advantages of microstrip
patch technology. MMICs and OEICs are typically developed on high
dielectric constant, thin material (note: the dielectric constant for GaAs and
33
AIGasInP, common materials for MMICs and aBICs is approximately 13).
Trying to develop a microstrip patch antenna in this environment would result
in an antenna with poor bandwidth and radiation performance. Even trying to
directly integrate microstrip patch technology with passive microwave circuits
such as filters and couplers presents a problem, as to make these circuits
compact, once again high dielectric constant and thin laminates are typically
utilized, such as Alumina materials (e, = 10.2). Fortunately there are ways to
overcome this problem and they will be addressed later in Chapters 4 and 6.

A typical E- and H-plane co-polar radiation pattern is presented in


Figure 2.4.3. As can be seen from this plot the radiation pattern is relatively
broad in both orthogonal directions. E-plane patterns are generally broader
than H-plane due to the presence of the surface wave mode and other higher
order modes being launched towards endfrre (8 = 90°).

180

Figure 2.4.3 RadiationPatternof a MicrostripPatch Antenna

From Figure 2.4.1 the initial design procedure of any microstrip patch
solution can be established. If only a low impedance bandwidth solution is
required, say less than 5 % of the operating frequency, then the material to be
used for the patch can be selected by looking at Figure 2A.la. Once this has
been determined the next phase: the conductor shape can be selected. This
34
will be based upon the space available for the antenna and the polarization
required. Typically the gain specified for the antenna is considered much
later in the process as the gain a single microstrip patch element is relatively
constant and can be easily increased by using arraying techniques. Once the
above has been chosen, the next stage is to look into the feeding technique, or
how to couple power to and from the patch antenna.

2.5 Excitation Methods of Microstrip Patches

2.5-1 Introduction

Now that the general responses of a simple microstrip patch antenna


have been discussed, it is now timely to look into methods of launching power
to and form this printed radiator. The way a microstrip patch antenna is
excited will determine the achievable impedance bandwidth (indirectly), the
purity and direction of the radiated fields, the efficiency of the overall
antenna, the ease of manufacturing of the antenna and its robustness. There
are four fundamental techniques to feed or excite a microstrip patch antenna,
namely edge-fed, probe-fed, aperture-coupled and proximity-coupled. These
can be further simplified into direct (edge and probe) and non-contact
(aperture and proximity-coupled) methods. Recently, there has been mention
of some new excitation techniques, such as the L-shape probe; however, this
is really a hybrid representation of the probe and proximity-coupled versions.
The properties of each feeding method are summarized below.

2.5-2 Edge fed patches

2.5-2.1 Configuration and Characteristics

One of the original excitation methods for a microstrip patch antenna


is the edge-fed, or microstrip-line fed technique [11]. A schematic diagram
representing this method is shown in Figure 2.5.1. Here a microstrip feed-line
of width Wf is in direct contact with a rectangular patch conductor of length L
and width W. The patch resides on a grounded substrate of thickness, d and
dielectric constant, e, Typically the microstrip feed-line comes in contact
with one of the radiating edges of the patch, as shown in Figure 2.5.1,
although cases where the contact is located along the width of the patch have
also been examined. As the excitation source is in direct electrical contact
with the patch radiator, edge (or sometimes known as microstrip-line) feed
falls into the category of a direct contact excitation means.

35
Microstrip feedline, Wf

Grounded substrate

Figure 2.5.1 Edge-fed Microslrip Patch Antenna

There are several advantages of edge-feeding a microstrip patch over


other feeding techniques. One of the key features of this technology is its
ease of fabrication, as the feed network and radiating patches can be etched on
the one board. It is for this reason that many large planar arrays have been
developed using edge-fed patches. It is also very easy to control the level of
the input impedance of an edge-fed patch. Simply by inserting the feed into
the patch conductor the impedance at resonance can be adjusted from very
high 150 - 250 ,Q when the contact point of the feed-line and the patch at the
radiating edge of the patch, down to a couple of ohms if the contact point is
near the center of the patch.

The equivalent circuit for an edge-fed microstrip patch is shown in


Figure 2.5.2 [12]. Here the traditional RLC tank circuit represents the patch
antenna and the stand-alone inductor represents the feed. Far below
resonance the feed inductor dominates the impedance response of the antenna.
This is why when plotted on a Smith chart, at very low frequencies (below
resonance) the impedance is not zero, but a finite inductance. As the
thickness of the substrate increases, the feed inductance becomes larger and
so the offset from the short circuit point is bigger.

Edge-fed patches in their simplest form are relatively easy to model,


if electrically thin material is used (d < 0.015 Ao). Simple transmission line
models can be utilized to give estimations of the input impedance
performance of the antenna. However, as mentioned before, for cases when
thicker materials are used, the modeling of the performance is not too
straightforward. This is because of the current distribution of the

36
discontinuities associated with the contact point between the microstrip line
and the patch antenna.

Figure 2.5.2 Equivalent Circuitof an EdgeFed Patch

Edge-fed microstrip patches have bandwidth and gain characteristics


consistent with Figure 2.4.1, that is, these patches are relatively narrow
bandwidth antennas. As is evident from Figure 2.4.1, if thick, high dielectric
constant material is used as the substrate, the surface wave efficiency is poor.
Also this form of feeding technique can suffer from relatively high spurious
feed radiation. This is simply because the feed-network is not separated from
the antenna and so material suitable for efficient radiation for the antenna will
also cause the feed network to radiate too.

As the feeding mechanism described here is in direct contact with the


patch antenna, the overall efficiency of this antenna is relatively high,
approximately 90 % if the surface wave losses are low. Conductor losses are
important to minimize for this style of microstrip patch.

2.5-2.2 Design Procedure

For edge-fed patches mounted on relatively thin substrates, the design


procedure for these antennas is relatively simple. Using the selected material
that will give the appropriate bandwidth (according to the relationship
summarized in Figure 2.4.1) the width of a 50 Q feed line is calculated and
the resonant length of the patch is calculated. A good initial estimate of this
length is:

L=Ae!2 (2.1)
where Ae is the effective wavelength given by:

37
(2.2)

where Ee = (e, + 1)/2

There are probably better approximations for Ee , however, as this is only a


starting point to an initial estimation, the extra effort to obtain a more accurate
e, is not worth it.

Once the approximate length has been calculated, the transmission


line model or the cavity model can be used to give an estimation of the
performance of the antenna. The length is typically adjusted to ensure the
correct frequency of operation is obtained. To correctly match the input
impedance of the antenna, the designer has two choices: to modify the width
of the patch or change the insert feed position. As mentioned before, as the
width of the patch is increased, the input impedance decreases. The input
impedance of the edge-fed patch can also be decreased by extending the feed
insert further into the patch conductor. With respect to the feed insert, a
common rule of thumb is to make the gaps between the feed-line track and the
inner edge of the patch at least the same width as the track itself. Whether the
width is increased or the insert position is moved further into the patch
conductor really depends on other requirements for the antenna. For example,
is dual polarization a requirement and therefore a square patch will be
needed? Or is an array of patches going to be developed and so the width
must not be too large to ensure grating lobes occur in the radiation patterns.

After the initial simulations runs have satisfactorily been completed,


it is recommended a full-wave simulator be used to then give a design ready
for developing. If not then the designer should be prepared for "patch
modifiers" as mentioned before.

2.5-2.3 Design examples

Introduction

Two design cases are presented here for single layered edge fed
patches: an edge fed rectangular patch etched on a high dielectric constant,
electrically thick material [13]; and a dual frequency sectored disc patch for a
self-diplexing transceiver [14]. The example is a case of essentially what not
to do, as it highlights some of the shortcomings of this feeding procedure.
The second design gives an example of where edge feeding is very useful and
easy to do. More fundamental designs of edge fed patches can be readily
found in the literature and so they will not be repeated here.

38
Millimeter-wave Edge Fed Rectangular Patch Antenna

An edge-fed rectangular microstrip patch is considered for rnillimeter-


wave (mm-wave) communication applications where the antenna needs to be
directly integrated with active components at the transceiver. The radiating
element is connected to the mm-wave and photonic circuitry via a simple
microstrip line. This type of microstrip patch radiator will suffer from the
compromises mentioned previously as the material used to create the substrate
for the patch has a high dielectric constant, approximately 13.

To accurately predict the performance of the edge-fed rnicrostrip


patch configuration a rigorous analysis is required. The analysis used here is
based on the Spectral Domain Integral Equation technique [15] and
importantly includes all surface wave effects. The antenna surface wave
efficiency, can be written as:

11 = 1 - P sw I Plot (2.3)

where Plot is the total power delivered by the source to the antenna
and Psw is the power delivered by the source to the surface waves.

Figure 2.5.3 InputImpedance Response of mm-wave Edge-fed Patch

The required operation frequency for the patch antenna is 37 GHz.


The substrate, where the photonic and/or RF circuitry is located, was assumed
to be 0.254 mm thick and have a dielectric constant of 12.5 (consistent with
the relative dielectric constant of GaAs and InP). The design procedure
39
outlined earlier was applied to this material. Figure 2.5.3 shows the
impedance behaviour of an edge fed patch designed for 50 ~ resonance at the
desired frequency. The 10 dB return loss bandwidth of this configuration is
200 MHz (approximately 1%). The surface wave efficiency, of the edge-fed
patch is 76 %. As previously mentioned, to increase the bandwidth of the
edge fed patch, thicker substrate can be used. By doubling the thickness of
the substrate, the bandwidth can be increased to 370 MHz, however, this is
achieved at the price significantly reducing the efficiency, to less than 45%.
The gain of this edge-fed patch was 4 dBi.

The example here shows some of the definite shortcomings of using


an edge fed microstrip patch in an inappropriate application. Design cases
will be presented in Chapter 6 that will overcome the issues encountered here.

Self-Diplexed Integrated Transceiver for Wireless Applications

Introduction

Due to the advent of unlicensed wireless systems, such as Bluetooth


and WLAN, there has been much attention drawn to the development of low
cost wireless integrated antenna transceivers. Microstrip patch technology
would seem to be a good choice of antennas for these systems due to their
already mentioned ease in fabrication as well as the ease in which they can be
integrated with active devices. In fact the inherent large electromagnetic
signature of a microstrip patch antenna can be taken advantage of here helping
to provide good isolation between the up- and down-link channels of the
transceiver, if a dual frequency antenna can be developed with good isolation
between the two frequency bands. To do this requires a printed antenna that
excites two independent modes. Such an antenna not only removes the
necessity of high order filters/diplexers but also reduces the overall
component count of the transceiver, thereby improving its efficiency and
minimizing its cost. Importantly for wireless systems such as WLAN, the base
station antennas typically use some form of polarization diversity to counter
multipath effects and so the polarization of the terminal antenna is not critical.

In this sub-section we present a compact dual-frequency printed


antenna with orthogonal polarized modes and high isolation between the two
feeding ports. Here the patch conductor shape is a circular sector disc and
high dielectric constant material is used to provide good isolation between the
up and down-link bands. This good isolation provided by the patch antenna is
used as the basis for the transmit-receive filtering.

40
Antenna Design

Figure 2.5.4 shows a photograph of the circular sector microstrip


antenna for dual-frequency operation. The circular sector patch of radius
a = 14.7 mm and sector angle ¢o =70° is printed on a substrate (RT/DuToid
6010) of dielectric constant cr =10.2 and thickness h =1.96 mm, The electric
field under the patch can be calculated using a cavity model and is given by
[16]:

s, =- [ca LLAnmJ n(kvnmP)cosv n¢


V (2.4)
n=Om=1

where to is the angular frequency, A nm is an arbitrary constant,


J v. is the Bessel function of the first kind, v n =nx]¢o' I
k v•m =x v•m a and
x vnm is a solution of aJ n(x)lax = o.
V The resonant frequencies of the
proposed antenna is given by:
x vnm · c (2.5)
I" _
J vnm -
21la
eff
.JE:
where c is the velocity of light in free space and a eff is the effective
radius of the circular patch due to fringing fields and is given in [17].

Figure 2.5.4 Photograph of CircularSectorPatch Antenna

When the sector angle ¢o is 70°, the first and second modes of the
circular sector patch antenna are TM vII' VI =2.571 and TM01 mode,

41
respectively, and the electric field intensity of the two modes, calculated by
using the FDTD method [18], are shown in Figure 2.5.5. From the electric
field distribution of TM vII and TM 01 mode, we found that the two modes
have orthogonal polarization and broadside radiation patterns. The feeding
position is critical to get good isolation between two modes as can be seen
from field plots.

". ""
..,
D'
o

Figure 2.5.5 Field Intensities of Modeson CircularSector Disk Antenna

In Figure 2.5.4, port 1 excites the TM 01 mode and port 2 excites the
TM vII mode. The measured S parameters of the two-port patch antenna are
shown in Figure 2.5.6. The isolation between the two ports (S21) is greater
than 30dB. The radiation patterns of the proposed antenna at the two
operating frequencies are measured and shown in Figure 2.5.7. It can be seen
that the two operating modes have broadside radiation patterns and orthogonal
planes. The cross-polarization radiation levels are more than 20 dB below the
co-polar radiation levels. The difference in the antenna gain in the broadside
direction at the transmit and receive frequencies is approximately 0.2dB.
o .......\ x~·7 ··
·10
~ V
!
522
-20 L - 5 11

! -30
_.- .- .... ... ......-. .
-40
""
\
_. ........ . -.....

,, 521
I~
\

·50

-60
5 5.2 5.4 5.6 5.8 6
Frequency [GHz]

Figure 2.5.6 Measured S-parameters of CircularSectorDisk Patch Antenna

42
10 10

110 0

270 270
(a) (b)

Figure 2.5.7 Measured Radiation Patterns at: (a) 5.4 GHz; (b) 5.5. GHz

2.5-3 Probe Fed Patches

2.5-3.1 1ntroduction

Another original excitation method proposed and demonstrated in the


mid 1970s is probe feeding [11]. A schematic diagram representing this
configuration is shown in Figure 2.5.8. Here a probe of radius ro extends
through the ground-plane and is connected to the patch conductor, typically
soldered to it. The probe or feeding pin is usually the inner conductor of a
coaxial line; hence probe-feeding is often referred to as a coaxial feed. The
probe position provides the impedance control in a similar manner to inserting
the feed for an edge-fed patch. Because of the direct contact between the feed
transmission line and the patch antenna, probe feeding is referred to as a
direct contact excitation mechanism.

2.5-3.2 Characteristics ofProbe-fed Patches

There are several key advantages of the probe-fed patch. Firstly, the
feed network, where phase shifters and filters may be located, is isolated from
the radiating elements via a ground-plane. This feature allows independent
optimization of each functional layer. Of all the excitation methods, probe
feeding is probably the most efficient as the feed mechanism is in direct
contact with the antenna as well as most of the feed network is isolated from
the patch, minimizing spurious radiation. The high efficiency of this printed
antenna has seen a somewhat renaissance of the probe-fed styled patch,
despite the added complexity of developing a via connection. For a probe-fed
43
patch mounted on the same material considered for the edge-fed case
presented in Section 2.5-2.1 is approximately 96 %.

y'

Figure 2.5.8 Probe-fed Micorstrip Patch Antenna

Probe-fed microstrip patches have similar issues to edge-fed patches,


namely their bandwidth is somewhat small and these printed antennas are
somewhat difficult to accurately analyze. The probe used to couple power to
the patch can generate somewhat high cross-polarized fields if electrically
thick substrates are used. Also because this antenna is no longer a single layer
geometry, due to the location of the feed network it is more complicated to
manufacture.

The equivalent circuit of a probe fed microstrip patch antenna is


shown in Figure 2.5.9. As is the case for the edge-fed patch, here the feed is
modeled by an equivalent inductance. The inductance becomes larger as the
length of the feed pin is increased. It is in fact this feed inductance that limits
how thick the substrate for a direct contact fed patch can be. Beyond about
0.06 Ao, it is difficult to achieve a resonance in the impedance response and
therefore the radiation efficiency is compromised.

2.5-3.3 Design Procedurefor Probe Fed Patches

The design procedure for probe fed patches is similar to that for edge
fed patches and so it will not be repeated here, however some additional
defmitions and comments will be made. Figure 2.5.10 shows the effect on the
input impedance response of varying the position of the probe feed for a
44
microstrip patch antenna. As mentioned before, varying this parameter
(equivalent to varying the position of the insert for an edge fed patch) changes
the input impedance of the antenna from a high value when the probe is
located close to the edge, to a low value when the probe is positioned near the
center. When the input impedance response is will matched at resonance (50
Q) we will define this as a case where the patch is critically coupled. This
means that at this feed location the fields are at the appropriate levels to give
this impedance. For the case where the input impedance is less than 50 Q at
resonance we will define this as being under coupled and for when the
impedance is greater than 50 Q we will say the patch is over-coupled. Figure
2.5.10 displays this representation. These definitions will be used throughout
this book and are very important when it comes to investigating the effects of
parameter variation on the response of the antenna.

Figure 2.5.9 Equivalent Circuit of Probe-fed Patch Antenna

There are several points that should be made here with respect to
designing probe fed patches and these are related to how the probe feed
should be connected to patch antenna. SMA connectors are the usual choice
for this application, except when very high frequencies, into the mm-wave
range are considered. Here K-type connectors are typically used. Firstly, the
feed pin radius should be small to minimize conductor loss associated with
the pin. Also, thinner pins yield low cross-polarization levels. Before
soldering the pin to the patch conductor, ensure it does not extend much
beyond the top of the copper track, relative to the frequency of operation. If
the pin extends beyond the patch conductor, it is likely to contribute to
generating more cross-polarized radiation. Care should be taken when
drilling the hole through the patch conductor to the ground-plane to avoid
potential short-circuits occurring from copper ribbons coming in contact with
the ground-plane. It is also important to make sure the hole in the ground-
45
plane, where the center conductor of the SMA comes through the ground-
plane physically matches the outer radius of the connector.

Under-coupled Crilically-coupled

Over-coupled

Figure 2.5.10 Input Impedance of Probe-fed Patch as a Function of Probe Position

2.5-3.4 Design Example - Low Cost Printed Antenna for Indoor Wireless
Base Stations

Introduction

There are currently two competing strategies for the delivery of


mobile communications to the customer. These technologies incorporate
either telecommunication satellites or terrestrial cellular communication
networks. The latter technology has received considerable attention recently
with the proposed optically distributed microcellular systems which will
satisfy increased user demands as well as reduce the likelihood of detrimental
phenomenon such as outage [19]. However, there is one important
disadvantage of microcellular technology and this is the large number of sites
required, typically several thousand within a central business district.
Although the cost of implementation may be justified in such an environment
where user demand is high, within a suburban area it is not. For this reason
mobile satellite communication systems have received increased popularity in
recent times (examples of proposed systems are given in [20]). There are
however, some environments where the performance of the satellite
communications link will be poor, in particular in large buildings or tunnels.
In these situations some form of interconnect system between the satellite

46
communication topology and a microcellular or indoor wireless system is
required to ensure no loss of continuity for the user.

In this section an antenna module suitable for a satellite/indoor


wireless interconnect is proposed. The configuration is low cost, unobtrusive
and consists of two antennas, one for the transmit band of frequencies and the
other for the receive band. Since the module consists of two separate transmit
and receive antennas, bidirectional transmission over optical fiber can be
easily implemented between the satellite terminal (located at a good reception
point) and the proposed indoor base station site. Among other advantages,
optical fiber has low loss, small size and light weight. This paper describes
the antenna module and presents its characteristics, including input impedance
behavior and far-field radiation patterns. The current prototype is appropriate
for the Australian MobileSat system (transmit frequency band: 1.5450 -
1.5590 GHz and receive frequency band: 1.6465 - 1.6605 GHz).

Configuration and Design

A schematic diagram of the proposed antenna configuration is shown


in Figure 2.5.11. The antenna module consists of two separate circular
microstrip patches, with radii R) and R2, respectively, etched on a common
grounded substrate of height, d and with dielectric constant, e, Each patch is
fed by a probe located at (Xph ypi) from its center and is designed to cover the
appropriate band of frequencies. To minimize the interaction between patches
or mutual coupling, the large electromagnetic signature typical of a microstrip
patch etched on an electrically thin substrate, is utilized. Therefore, whereas
in most applications the inherent narrow bandwidth of a microstrip patch
antenna is a hindrance, here it is exploited. To numerically model the circular
probe-fed microstrip patches in the module, the Spectral Domain Integral
Equation analysis presented in [21] was implemented.

An important factor in any communication system is the cost. To


minimize the cost of the printed antenna module, standard PCB was used as
the substrate. Using this dielectric material and the analysis presented in [21],
the patch dimensions required for the frequency bands are: transmit band: R)
= 27.06 mm, xp) = 9.70 mm; receive band: R2 = 25.65 mm, Xp2 = 10 mm.
Another important issue in mobile communication systems is size of the
module. Therefore a compromise must be made between the overall size of
the module and the mutual coupling between the elements. A distance
between patch centres in the H-plane of 10 em (0.53 Ao) was chosen and the
overall size of the module was 231 ern',

47
x


r,

+ ~
(X", Y,,) (x". Y,, )

(a) (b)
Ftgure 2.5.11 Schematic uf Arueuua Configuration (a) Top View; (b) Side View

Results and Discussion

Figure 2.5.12 shows the predicted return loss of the two circular
patches as a function of frequency. Also shown in this diagram are the
measured ISIII and ISzzI results obtained using a Wiltron 360 B network
analyser connected to the two ports of the module. As the results show, the
amount of coupling between the patches is relatively small, since experiment
and theory are in good agreement. The coupling between the ports was
measured and the maximum was -33 dB at 1.56 GHz. Figure 2.5.12 also
shows that the impedance bandwidth requirement for both transmission and
reception bands has been satisfied using this substrate.

-5

-10

-15
~
·20
~
a.
E -25
a:: · 30
••••• 'Trans. Theory
- - Trans. Exp.
- •• - Reo. Theory
- • - • Rec. Exp.
-35

-40

-45
1.5 1.53 1.56 1.59 1.608 1.638 1.668 1.698

Frequency (Gllz)

Figure. 2.5.12 Predicted and measured returnloss of antennaconfiguration

48
The E- and If-plane far-field radiations patterns of the antenna
module were measured at the center frequencies of both transmission and
reception bands. Figure 2.5.13 presents the measured co-polar E- and H-
plane patterns of the transmit patch at 1.54 GHz (with the receive patch
terminated with a 50 Q load). The E-plane results are typical for a microstrip
patch, whereas the H-plane pattern displays an interesting phenomenon. A
'reflector-like' behavior is experienced due to the presence of the other
element. This is a similar pattern distortion to that experienced in a printed
Yagi antenna [22]. The patterns for the receive patch at 1.632 GHz are very
similar to those presented in Figure 2.5.13, with the 'broadening' of the H-
plane pattern on the other side of the maximum, as expected. The measured
cross-polarization levels were less than -25 dB at all frequencies in both
planes. From the results presented in Figure 2.5.13 it is evident that the usual
requirement for wireless base stations has been satisfied, namely a relatively
broad 3 dB beamwidth, in this case 90 0 in the E-plane and 850 in the H-plane.
o

225

180

Figure 2.5.13 Measured E-plane and Il-plane Patterns at 1.54 GHz

Other printed antenna solutions for indoor wireless systems are


possible. These include using a single patch mounted on a low dielectric
constant, electrically thick substrate (approximately 0.06 Ao for the required
overall bandwidth) or a stacked patch configuration. One obvious
disadvantage of these choices is the need to isolate the transmit and receive

49
carriers from the one port, typically using a diplexer which increases the cost
of the RF circuitry.

2.5-4 Proximity coupled patches

2.5-4.1 Introduction

The second form of non-contact fed patches created to overcome the


shortcomings of the direct contact fed patches is the proximity-coupled patch
[23]. Figure 2.5.14 shows a schematic diagram of this printed antenna. The
microstrip antenna consists of a grounded substrate where a microstrip feed-
line terminated with an open circuit is located. Above this material is another
dielectric laminate with a microstrip patch etched on its top surface. Please
note there is no ground-plane separating the two dielectric layers. The power
from the feed network is coupled to the patch electromagnetically, as opposed
to a direct contact. This is why this form of microstrip patch is sometimes
referred to as an electromagnetically coupled patch antenna.

pal.chCL, W)

microslrip feed line, w,

Figure 2.5.14 Proximity-coupled Microstrip PatchAntenna

2.5-4.2 Characteristics ofProximity Coupled Patches

A key attribute of the proximity-coupled patch is that its coupling


mechanism is capacitive in nature. This is in contrast to the direct contact
methods, which, as mentioned before, are predominantly inductive. The
difference in coupling affects the obtainable impedance bandwidth, as the
inductive coupling of the edge- and probe-fed geometries limits the thickness
of the material useable. Thus bandwidth of a proximity-coupled patch is
50
inherently greater than the direct contact feed patches. The capacitive nature
of this non-contact excitation method to the patch is represented in the
equivalent circuit shown in Figure 2.5.15. One important note is that when
examining the input impedance of this antenna on a Smith chart, because of
the capacitive nature of the feed, the 'origin' of the frequency response will
start near the open circuit point, be capacitive below resonance and then
inductive above resonance, before approaching the open circuit point again.

Figure 2.5.15 Equivalent Circuit of Proximity-Coupled Microstrip Patch


Antenna

Full-wave analyses are not too difficult to develop for the proximity-
coupled patch due to the lack of a current discontinuity between the feed
network and the radiating element [15]. Thus unlike for edge-fed and probe-
fed patches, special attention does not need to be undertaken near the feed
point, and no 'attachment mode' or heavy segmentation of the conductor area
in this vicinity is necessary. It is interesting to note that the formulation of the
Spectral Domain Integral Equation in [15] that can be applied to both edge-
fed and proximity-coupled patches is more accurate for the latter case, simply
because of the lack of a current discontinuity.

There are some shortcomings of the proximity-coupled microstrip


patch. First of all, the feed and antenna layers are not fully independent, as
power must be coupled efficiently to the antenna. Therefore these printed
antennas can have relatively high spurious feed radiation, albeit, not as high
as for an edge-fed case. The antenna is a multilevel structure and so
alignment procedures are important. Small air-gaps between the feed
substrate and the laminate for the antenna can affect the coupling to the patch
and so care must be taken when fabricating these antennas.

The proximity coupled patch is slightly less efficient than the edge
fed patch, given the same material parameters for the design. In the case
51
considered previously for the edge-fed patch, the proximity coupled solution
has an efficiency of approximately 88 %. The efficiency for a proximity
coupled patch is an issue, especially when thicker material is to be used to
enhance the bandwidth (considering the trends presented in Section 2.4). This
will be looked into further in Chapter 3.

2.5-4.3 Design Procedure for Proximity CoupledPatches

One of the salient features of proximity coupled patches are the extra
degrees of freedom available in this form of microstrip patch, compared to the
direct contact feeding procedures. Thus to achieve a well-matched impedance
response, the designer can change several parameters, relatively
independently. Here we will consider 6 parameters and their effects on the
input impedance response.

Antenna substrate

Using a low dielectric constant dielectric layer here can obviously


enhance the achievable bandwidth and surface wave efficiency here. This
will be shown in the design example presented in Section 2.5-4.3. Care needs
to be taken, as using too thick a layer can make it difficult to couple energy to
and from the patch conductor. A general rule of thumb here is if the antenna
layer is equal to or less than double the feed layer, critical coupling can still
be achieved. If it is greater than this, the patch will be under-coupled and it
will be extremely difficult to correct the impedance response, despite the
many degrees of freedom presented here.

Feed substrate

Although this is typically not a design parameter as it is set by the


remaining circuitry, the feed layer does affect the achievable impedance
bandwidth of the antenna. Using high dielectric constant material here,
requires the antenna substrate to be thinner, simply because more of the feed
layer field is confined within the feed layer. Thus the two substrate layers are
not mutually exclusive. Assuming that there is no restriction on the feed
material, making this layer thicker allows for the antenna layer to become
thicker, although this will increase the likelihood of spurious radiation.

Patch Dimensions (Length and Width)

The patch length once again predominantly controls the resonant


frequency of the antenna and so the design formula applied to the edge and
probe-fed patches can be applied here too. A more accurate representation of
52
the effective dielectric constant considering the thicknesses of each layer can
be used, however once again, the extra effort required gives "limited returns"
as this is only a starting point for design process.

If dual polarization is required, then this dimension can obviously not


be varied. However, for linearly polarized cases this parameter can give a
useful extra degree of freedom. The width of the patch can have two
consequences on the overall impedance response. Firstly, as the width is
reduced, the resonant frequency of the antenna is reduced. Also the width of
the patch can be used to help achieve critical coupling of the antenna to the
feed network. Consider a design where the parameters of the proximity
coupled patch have been optimized to give critical coupling. If a thicker
antenna layer is now used to achieve better bandwidth performance, and if the
other parameters are left unchanged, the patch will be under-coupled as less
energy will be fed to the patch. By reducing the width of the patch, critical
coupling can then be achieved again, to a degree. Please note that for a
proximity coupled patch under coupling means a higher input impedance than
50 .0 is achieved at resonance and over coupling means a lower input
impedance (less than 50 .0) at resonance has resulted. Both phenomena can
be simply explained by representing the coupling mechanism by an equivalent
capacitance. As the width of the patch is reduced, the equivalent capacitance
is also decreased thereby creating both of the described effects. Of course, a
potential shorting coming is that the overall area of the antenna is decreased,
and therefore the increase in bandwidth associated with increasing the volume
could be somewhat compromised by this decrease in area.

Feed-linewidth

Obviously this should be made to match the impedance of the


circuitry that's connected to the antenna, however, as is the case for the width
of the patch, the feed-line width can be changed to improve the coupling
between the feed-line and the radiator. Care would need to be taken to
minimize current discontinuities associated with the transition from the
optimized feed-line width associated with the proximity coupled patch and the
50.0 track.

Location of open circuitwith respect to patch

This parameter, along with the thickness of the materials used, is the
most critical in determining whether the patch is critically coupled or not.
Moving the location of the open circuit termination closer to beneath one of
the radiating edges of the patch, increases the coupling between the feed-line

53
and radiator. Simplistically this action can be interpolated as attempting to
overlap the location of the maximum fields from the resonant patch and the
fields associated with the open circuit termination. Once again if a proximity
coupled patch has been designed to be critically coupled and then the location
of the open circuit termination is moved closer to the patch edge, the antenna
will become over-coupled.

2.5-4.4 Design Example - mm-wave Proximity Coupled Patch

Introduction

In this design example, we show a proximity coupled microstrip patch


antenna mounted on the same material considered in the first design case
presented for the edge-fed patch. Doing so will allow us to get a true insight
into when this patch technology is extremely useful.

Configurations and Theory

The proximity coupled microstrip patch configuration considered is


similar in form to that presented in Figure 2.5.14. Here, the patch conductor
is etched on a dielectric layer that is parasitically coupled to the microstrip
feed line. The advantage of this form of printed antenna over the
conventionally feed patch is that the dielectric layer can be optimized for
efficient radiation independently of the material used for the RF circuitry
(with a high dielectric constant).

To accurately predict the performance of the microstrip patch


configuration a rigorous analysis was used. Once again, the analysis used
here is based on the Spectral Domain Integral Equation technique [15] and
importantly includes all surface wave effects. The required operation
frequency was set to 37 GHz. The feed substrate, where the photonic and RF
circuitry is located, was assumed to be 0.254 rom thick and have a dielectric
constant of 12.5 (consistent with the relative dielectric constant of GaAs and
InP).

Results and Discussion

The predicted input impedance behavior of a proximity coupled patch


designed for 50 Q resonance at 37 GHz is presented in Figure 2.5.16 . The
dielectric layer between the feed circuitry and the antenna conductor was
chosen as RT Duriod 5880 (E,- = 2.2, d = 0.254 rom). The 10 dB return loss
bandwidth of this patch is 500 MHz (or 8 %) and the surface wave efficiency

54
was approximately 72 %. Thus comparing this result to that presented in
Section 2.5-2, larger bandwidths can be achieved with this configuration
without serious degradation to the efficiency. The gain of the proximity
coupled patch was 6 dBi. More designs incorporating proximity coupled
patches are presented later in this book.

Figure 2.5.16 InputImpedance of mm-waveProximity-Coupled Patch

2.5-5 Aperture Coupled Patches

2.5-5.1 Introduction

As a result of the shortcomings of the direct contract feeding


techniques, namely the small inherent bandwidth and the detrimental effect of
surface waves, the non-contact excitation mechanisms were introduced.
Aperture coupling was another non-contact feeding mechanism investigated
to overcome these issues [24]. A schematic diagram of this microstrip patch
antenna is shown in Figure 2.5.17. Here separate laminates are used for the
feed network and the patch antenna. The laminates are separated by a ground-
plane and coupling between the feed, in this case a microstrip line, and the
patch antenna is achieved via a small slot in the ground-plane. The feed-line
is terminated either with an open circuit or a short circuit stub (in Figure
2.5.17 an open circuit stub is used). Aperture-coupled microstrip patch
antennas are probably the most utilized microstrip patches in today's global
market.

55
Oroun~plan.

~-- Feed Substrata (tol> tan 6. 4)

Figure 2.5.17 Aperture-Coupled Microstrip Patch Antenna

2.5-5.2 Characteristics ofAperture CoupledPatches

The configuration shown in Figure 2.5.17 has several advantages over


its direct contact and non-contact counterparts. Firstly, unlike the edge-fed
patch antenna and the proximity coupled patch, independent optimization of
the feed and antenna substrates can be achieved here. Of course whenever
such a statement is made it can be readily countered and in fact we will
investigate configurations in Chapter 6 showing that this is not the case.
However, they are independent, at least to a first order approximation. Thus
these microstrip patch antennas are suitable for integration with active
devices. Unlike the probe-fed configuration, no vertical interconnects are
required, simplifying the fabrication processes and also adhering to the
conformal nature of printed circuit technology. Having said that, alignment
issues can be important here and also multi-level fabrication processes are
typically required. Being a multi-layered antenna can create other problems.
The presence of small gaps between the layers of dielectric can significantly
alter the input impedance nature of the antenna, especially at high frequencies,
where these gaps appear larger electrically. Also the material required to
bond the layers could play a significant role in the performance of the
antenna. If the bonding material is lossy and is located near say the slot, the
efficiency of the antenna will be reduced.

The equivalent circuit of an aperture-coupled patch is shown in Figure


2.5.18 [12]. Here the patch effectively is a load for the slot and then this
combination is terminated with the open circuit stub (or capacitance). One
can think of this as a 'magnetic' equivalence of the equivalent circuit for the
probe and edge fed patches, with some subtle differences. Because of this, in
its original form the aperture-coupled patch has similar bandwidth and gain
56
responses as the direct fed patches, however it is very easy to significantly
enhance the impedance bandwidth of this antenna. We will discuss how to
achieve enhanced bandwidth in the Chapter 3. The antenna efficiency of an
aperture coupled patch is slightly lower than the other feeding arrangements
and for the parameters considered before an aperture coupled patch antenna
would have an efficiency of 85 %. One very distinct advantage of an aperture
coupled patch over its microstrip counterparts is its capability of producing
very pure radiation, that is, minimal cross-polarization levels. This is because
a typically designed aperture coupled patch will use a thin slot to excite the
antenna, reducing the likelihood of generating cross-polarized currents. One
consequence of using a slot as an excitation source is the increased likelihood
of diffracted fields interacting with the finite sized ground-plane of the
antenna. This will be addressed in subsequent chapters.

1Il!E:=:: r$:--- -.,r =; ~G.


----- T
Figure 2.5.18 Equivalent Circuit of Aperture Coupled Patch Antenna

The aperture-coupled patch has more design parameters than the


direct contact fed patches and the proximity coupled approach and therefore
has more flexibility or degrees of freedom for the antenna designer. Despite
its somewhat complex appearance, the aperture-coupled microstrip patch
antenna is relatively easy to accurately model, even when using full-wave
analyses. Once again, the reason for this is that unlike for the direct contact
fed patches, there are no abrupt current discontinuities. Thus relatively
simple, accurate, computationally fast full-wave analyses are easy to develop.

2.5-5.3 Design Parameters and Procedure for Aperture CoupledPatches

Of all the different feed configurations, aperture coupled patch have


the most degrees of freedom and therefore for the patch novice, they are
probably the hardest to design. However, once a designer becomes familiar
with them, aperture coupled patches are relatively straight-forward to create.
As there is a slot involved with this patch radiator, another parameter must be
monitored, rather than just the input impedance response and this is the front-
to-back ratio (FBR). FBR is the ratio of far-fields radiated from the antenna
(patch and slot) in the forward direction, that is, beyond the patch and the far-
57
fields radiated from the slot in the backward direction. This parameter is
important in applications where antenna radiation coverage and isolation is
important, such as for sectorized mobile communication systems. For an
antenna designer considering an aperture coupled patch there are five degrees
of freedom typically used to control the input impedance response and their
affects on the input impedance performance of the antenna are discussed
below. It is assumed that the feed substrate cannot be varied.

Antenna substrate

If a patch is critically coupled, increasing the antenna substrate will


cause the patch to become under-coupled. This will manifest itself as a lower
input impedance at resonance. Using thicker antenna substrates allows
greater bandwidth to be achieved, although other parameters have to be varied
to obtain this goal.

Patch dimensions (length and width)

As for all styles of excitation, increasing the length of the patch


reduces the resonant frequency of the aperture coupled patch. In general, a
wider patch increases the bandwidth of this form of printed antenna, once
again as is the case for all microstrip patches. For a critically coupled
aperture coupled patch, reducing the patch width, causes the antenna is be
over-eoupled. A consequence of this is that the front-to-back ratio decreases.
A simple explanation of this effect is that the relative radiation from the slot
becomes larger as the patch size is decreased and therefore since a slot
radiates equally on both sides of a ground-plane, the FBR will be reduced.

Open circuitstub termination

The stub termination allows the designer to readily match the input
impedance response of the aperture coupled patch. The microwave circuitry
analogy to this parameter is the single stub matching procedure. For the patch
configuration, the terminated stub allows the frequency response to be rotated
on a constant conductance circle on the Smith chart.

Slot length

The slot length controls the amount of coupling to the patch radiator.
As it is increased in size, the amount of coupling to the patch increases. The
slot length is directly related to the antenna substrate used: the thicker the
substrate; the longer the slot.

58
The design procedure for the aperture coupled patch, is similar to that
for all microstrip antennas, in that it is conducted by examining the impedance
performance of the antenna on a Smith chart. The starting points mentioned
for the other feeding techniques, namely the patch length and substrate
selection is the same. The key to an aperture coupled design is to get a
starting impedance response that not confined to the edge of the chart. Once
there is a hint of a finite real part to the input impedance, it is relatively
straightforward to design a critically coupled patch using the parameter study
given above. I typically start with a slot length about half of the length of the
patch, although this will depend on the thickness of the material being used.
The start point for the open circuit stub, is usually near directly under the
center of the patch. As can be seen from the above 'parameter dependence of
the aperture coupled patch, different parameters can have countering effects
on the impedance and radiation performance. This is very useful as it gives
the designer the capability to tailor their design for the given application. For
example if minimal backward directed radiation is important, then rather than
increasing the slot length to couple to the patch mounted on a relatively thick
layer, the width of the patch can be decreased, to a degree.

2.5-5.4 Design Example - mm-wave Aperture Coupled Patch

Introduction

In this sub-section we will present an equivalent aperture coupled


patch solution to the mm-wave edge-fed and proximity coupled solutions.
Once again it will be shown that a non-contact feed procedure can easily out-
perform the edge-fed patch technique.

Configurations and Theory

The final configuration investigated for the base station antenna was
the aperture coupled microstrip patch (refer to Figure 2.5.17 for details of the
layout). Here the microstrip line and hence photonic and mm-wave devices is
coupled to the radiating element via a slot in the ground plane. Importantly
once again, the antenna substrate can be optimized for efficient radiation and
bandwidth, relatively independently of the photonic/rf circuitry material.

To accurately predict the performance of the aperture coupled


microstrip patch configurations a rigorous analysis is required. The analysis
used here is based on the Spectral Domain Integral Equation technique and
incorporates reciprocity and importantly includes all surface wave effects.

59
Results and Discussion

Once again, the required operation frequency for the mm-wave


application is 37 GHz. The feed substrate, where the photonic and RF
circuitry is located, was assumed to be 0.254 rom thick and have a dielectric
constant of 12.5 (consistent with the relative dielectric constant of GaAs and
InP). An aperture coupled patch configuration was investigated, and as was
the case for the proximity coupled patch, RT Duroid 5880 as the antenna
substrate. The impedance locus is shown in Figure 2.5.19. The 10 dB return
loss bandwidth is approximately 1 GHz (6 %) and the surface wave efficiency
is 70 %. The gain of the aperture coupled patch was 6.2 dBi and the FBR was
19 dB. From these results, the aperture coupled microstrip patches offer
enhanced performance compared to the edge-fed solution for the application
of rom-wave microcellular systems. Many more aperture coupled microstrip
patch design examples are presented throughout this book, in particular in
Chapters 3, 6 and 7.

Figure 2.5.19 Inputimpedance behaviour of aperture coupledmicrostrip patch

2.6 Circular Polarization Generation

2.6-1 Introduction

One of the advantages of microstrip patch technology is the relative


ease to generate dual and hence circular polarization (CP) with these antennas.
This feature is one reason why microstrip patch antennas have been utilized
for space-borne communication antennas on satellites. There are effectively
three methods for generating CP: using a single feed and perturbation
approach; a dual feed excitation technique; or a synchronous subarray

60
approach. Each method has its merits and disadvantages of which will be
summarized herein.

2.6-2 Single Feed Method

This approach requires no external circuitry for the microstrip patch


antenna to generate CP [25]. The principle relies on the fact that CP is
obtained when two orthogonal but otherwise identical modes are excited with
a 90° phase difference between them. To do so on a single patch with only
one feed is not too difficult to implement. Figure 2.6.1 shows a schematic
diagram of how this can be done. The patch is feed on one of its diagonal
planes therefore exciting two orthogonal modes, one resonant in the x-
direction and the other in the y-direction, satisfying one of the requirements
for CPo To achieve 90° phase shift between the two modes, one mode (or
field) is slightly perturbed with respect to the other. This can be done by
making one resonant dimension slightly longer than the other, truncating a set
of the comers of the patch or even putting a slot within the patch. The
philosophy here is that by slightly adjusting the resonant frequency of one
mode with respect to the other, if properly designed, then the overall phase
difference between the self-impedances of the modes and therefore the fields
will be 90°, satisfying the second requirement for CP generation.

Figure 2.6.1 Singlefeed CP edge fed Microstrip Patch Antenna

Although relatively simple to implement, there is a critical flaw in


this design procedure: the CP bandwidth (axial ratio less than 3 dB) is
extremely narrow, typically a fraction of the impedance (10 dB return loss)
bandwidth. Effectively the problem here is to ensure the phase relationship
(90 ° phase difference) holds across a range of frequencies. This is quite
difficult to do, especially when considering the rapid impedance variation of a
typical microstrip patch antenna near resonance. A case will be presented in
Chapter 6 where a single feed microstrip patch antenna was a CP bandwidth
(and impedance bandwidth) greater than 10 % was achieved. The key to
61
successfully broadening the CP bandwidth of this antenna is reducing the
impedance variation of each mode as a function of frequency.

2.6-3 Dual Feed Method

Probably the simplest and most common means of producing CP on a


microstrip patch antenna is to excite two orthogonal modes using separate
feeds and ensuring there is 90° phase difference between the modes by
incorporating a 90° phase shift within one of the feed lines [26], as shown in
Figure 2.6.2. CP bandwidths comparable to the impedance bandwidth of the
patch antenna are usually observed for this configuration. It can be somewhat
difficult to obtain true CP (0 dB axial ratio) for a direct contact dual feed CP
patch antenna due to the inherent asymmetry of the configuration as well as
the asymmetric cross-polarization levels generated by the patch (note: the H-
plane levels are higher than the E-plane levels off broadside, 9 = 0°). Also
spurious radiation from the feed network of the antenna will degree its axial
ratio performance. For this reason probe fed and aperture coupled solutions
are popular choices. Several examples of dual feed CP patches will be
presented throughout this book, in particular in Chapters 6 and 7.

Figure 2.6.2 Dual feed CP edge-fed Microstrip Patch Antenna

2.6-4 Synchronous Subarrays

By far the best CP bandwidths and lowest axial ratios can be achieved
using a synchronous subarray [27]. A four-element synchronous subarray is
shown in Figure 2.6.3. Here each patch is spatially rotated 90° with respect to
the previous patch. In addition to the spatial rotation, the feeds are also phase
rotated sequentially by 90°. Thus port 2 leads port 1 by 90°, port 3 leads port
1 by 180° and port 4 leads port 1 by 270°. Having such a feed arrangement
cancels the generated cross-polarization fields of the individual patches. It
has been shown in the past that very good axial ratio bandwidths can be
achieved using a synchronous subarray of elements that individually do not
radiate CP [28]. Also the impedance bandwidth of this CP antenna is
62
typically larger than that of the single element due to the feed network.
Synchronous subarrays do not necessarily consist of four elements; cases of 3
(0°, 120° and 240°) and higher order cases have been investigated. As long as
there is a spatial symmetry with respect to the center of the configuration as
well as the appropriate phase rotation, this approach should provide good
axial ratio performance. A typical axial ratio plot of a 4 element synchronous
subarray is shown in Figure 2.6.4. Note the very good axial ratio, in
particular between ± 45°.
Palch4 Patch 3

Patch I
Patch2

Figure 2.6.3 4 ElementSynchronous Subarrayof MicrostripPatches

Figure 2.6.4 Typical Axial Ratio Responseof Synchronous Subarray of Patches

There is one significant disadvantage of the synchronous subarray and


that's its electrical size. Because of this a synchronous subarray can have
severe grating lobe problems when implemented in an array. It is potentially
possible to utilize the synchronous subarray concept on a single patch by
incorporating 4 feeds, sequentially rotated, however the limited patch real
63
estate as well as the affect of the coupling of each feed on the input
impedance of the printed antenna could make this difficult to implement.
Once again several design cases of using synchronous subarrays to generate
CP are presented in this book, in particular in Chapters 5 and 7.

2.6-5 Comparison of Techniques

A comparison of the axial ratio performance of the three procedures is


shown in Figure 2.6.5. Here probe fed versions of the 3 configurations are
examined. A high dielectric constant material (s, = 10.2) was deliberately
chosen to highlight the issues associated with using a single feed arrangement.
Once again the inability of the single feed arrangement to maintain a 90
degrees phase relationship between the orthogonally radiated fields severely
limits the axial ratio bandwidth. As mentioned before the synchronous
subarray significantly out-performs the dual and single feed solutions.
However, in an array environment the beam efficiency is dramatically
reduced due to power radiated from the grating lobe due to the overall size of
the subarray.

10 , I.
,,
,, ,
,:"
--3dBAR
,
,,
, ,:" --SFednf.
,, t ........... OF eonf.
,, I
,
"
J --- SS eonf.

, "
___ f'f

o
'" <, ",'
, '"
"I
0.8 0.9 I.I 1.2
Normalized Frequency

Figure 2.6.5 Comparison of Axial RatioPerformance of 3 CP Techniques

2.7 Summary

In this chapter the fundamental properties and performance


characteristics of single layer microstrip patch antennas were presented.
Effects on the overall performance of the substrate material and the patch
conductor shape were investigated. It has been shown that using different
conductor shapes can give subtle differences in the impedance and radiation
performance of the printed antenna. It was also shown that to enhance the
64
bandwidth and improve the radiation performance (efficiency), thick, low
dielectric constant material should be used. In this chapter techniques to
numerically predict the performance of microstrip patch antennas was also
discussed and it was recommended that a full-wave analysis tool be used to
accurately determine the performance of the microstrip patch design.

The four general classes of excitation methods for a microstrip patch


antenna were summarized, highlighting the advantages and issues of each
type. Edge fed patches provide probably the simplest form of microstrip
patch, probe feeding is the most efficient and the non-contact feed approaches
overcome some of the inherent difficulties associated with the direct contact
methods. Aperture coupled patches provide the most flexibility of all the
excitation procedures and for this reason it is the most commonly used.
Design procedures for each configuration were provided. Design examples
highlighting the merits and issues associated with each procedure were also
given in this chapter.

Finally methods for generating circular polarization using microstrip


patch technology were outlined. The three procedures were discussed and the
advantages and disadvantages of each technique were given.

As mentioned before, this chapter lays the foundation for the rest of
the book. To get maximum benefit from this book, it is necessary to have a
thorough understanding of the material presented here. Many concepts based
on the material and trends of chapter 2 will be assumed in subsequent
chapters.

2.8 Bibliography
[1] K. R. Carver and J. W. Mink, "Microstrip Antenna Technology," IEEE Trans.
AntennasPropagat., Vol. AP-29, pp. 2 - 24, January 1981.
[2] E. H. Van Lil and A. R. Van de Capelle, ''Transmission Line Model for Mutual
Coupling Between Microstrip Antennas," IEEE Trans. AntennasPropagat., Vol. AP-
32, pp. 816 - 821, August 1984.
[3] Y. T. Lo, D. Solomon and W. F. Richards, "Theory and Experiment on Microstrip
Antennas," IEEE Trans. Antennas Propagat. , Vol. AP-27, pp. 137 - 145, March
1979.
[4] W. F. Richards, Y. T. Lo and D. D. Harrison, "An Improved Theory of Microstrip
Antennas with Applications," IEEETrans. AntennasPropagat. , Vol. AP-29, pp. 25 -
27, January 1981.
[5] M. D. Deshpande and M. C. Bailey, "Input Impedance of Microstrip Antennas,"
IEEE Trans. Antennas Propagat., Vol. AP-30, pp. 651- 656, July 1982.

65
[6] D. M. Pozar, "Input Impedance and Mutual Coupling of Rectangular Microstrip
Antennas," IEEE Trans. Antennas Propagat., Vol. AP-30, pp. 1191 - 1196,
November 1982.
[7] J. R. Mosig and F. E. Gardiol, "General Integral Equation Formulation for Microstrip
Antennas and Scatterers," Proc. lnst. Elect. Eng., pt. H, Vol. 132, pp. 424 - 432,
1985.
[8] A. Reineix and B. Jecko, "Analysis of Microstrip Patch Antennas Using Finite
Difference Time Domain Method," IEEETrans. AntennasPropagat., vol. AP-37, pp.
1361 - 1369, November 1989.
[9] J.-M. Jin and J. L. Volakis, "A Hybrid Finite Element Method for Scattering and
Radiation by Microstrip Patch Antennas and Arrays Residing in a Cavity," IEEE
Trans. AntennasPropagat., vol. AP-39, pp. 1598 -1604, November 1991.
[10] High Frequency Structure Simulator - Version 5.5, Agilent Technologies, 2000.
[11] R. F. Munson, "Conformal Microstrip Antennas and Microstrip Phased Arrays,"
IEEE Trans. AntennasPropagat., Vol. AP-22, pp. 74 -78, January 1974.
[12] D. M. Pozar, "Input Impedance of Microstrip Antennas," Proc. IEEE, Vol. 80, pp. 74
- 81, January 1992.
[13] R. B. Waterhouse and D. Novak, "Design of Patch Antennas for Integration in OEICs
for Optical Fiber Picocellular Systems," International Topical Meeting Microwave
Photonics (MWP96), Kyoto Japan, pp. 89 - 92, Dec. 1996.
[14] D.-K. Park, R. Waterhouse, Y. Qian and T. Itoh, "Self-diplexed integrated antenna
transceiver for wireless applications", 2001 IEEE Antennas & Propagation
Symposium, Boston, USA, July 2001.
(15] D. M. Pozar and S. M. Voda, "A rigorous analysis of a microstripline-fed patch
antenna", IEEE Trans. Antennas & Propagation, vol. AP - 35, pp. 1343-1349, Dec.
1987.
[16] W. F. Richards, J.D. Ou, and S. A. Long, "A theoretical and experimental
investigation of annular, annular sector, and circular sector microstrip antenna," IEEE
Trans. AntennasPropagat., vol. 32, pp. 864-867, Aug. 1984.
[17] L. C. Shen and S. A. Long, "The resonant frequency of a circular disc, printed-circuit
antenna," IEEETrans. AntennasPropagat., vol. 25, pp. 595-596, July 1977.
[18] K. S. Yee, "Numerical solution of initial boundary value problems involving
Maxwell's equations in isotropic media," IEEE Trans. Antennas Propagat., vol. 14,
pp.302-307, 1966.
(19] Special Issue on Fiber-Optic Microcellular Radio Communication Systems and Their
Technologies, IECETrans. Commun., vol. E76B, September 1993.
[20] F. Ali and J.B. Horton, "Introduction to special issue on emerging commercial and
consumer circuits, systems, and their applications", IEEE Trans. Micro. Theory &
Tech., MTT-43, pp. 1633-1638, July 1995.
[21] J.T. Aberle, D.M. Pozar and C.R. Birtcher, "Evaluation of input impedance and radar
cross section of probe fed microstrip patch elements using an accurate feed model",
IEEE Trans. AntennasPropagat., AP-39, pp. 1691·1697, December 1991.
[22] J. Huang, A. Densmore and D.M. Pozar, "Microstrip Yagi array for MSAT vehicle
antenna application", Int. Mobile Sat. Conj; pp. 554-559, June 1990.
[23] D. M. Pozar and B. Kaufman, "Increasing the Bandwidth of a Microstrip Antenna by
Proximity Coupling," Electronic Letters, Vol. 23, pp. 1070 - 1072, September 1987.
[24] D. M. Pozar, "A Microstrip Antenna Aperture Coupled to a Microstrip Line,"
Electronic Letters,Vol. 21, pp. 49 - 50, January 1985.
[25] J. R. James and P. S. Hall, Handbook of Microstrip Antennas, Vols. 1 and 2, Peter
Peregrinus, London, UK, 1989.
[26] C. A. Balanis, Antenna Theory: Analysisand Design, 2nd Edition, Wiley, New York,
1997.
66
(27) 1. Huang, "A Technique for an Array to Generate Circular Polarization with Linearly
Polarized Elements", IEEE Trans. Antennas Propagat., AP-34, pp, 1113 - 1124,
September 1986.
(28) D. M. Pozar, "Scanning characteristics of infinite arrays of printed antenna
subarrays," IEEE Trans. Antennas Propagat., AP- 40, pp. 666 - 674, June 1992.

67
Chapter 3 Enhancing the Bandwidth of
Microstrip Patch Antennas

3.1 Introduction

Probably one of the most researched areas in the history of microstrip


patch technology has been how to improve the impedance bandwidth of this
antenna such that it can be utilized in many more applications. From when
the first patch was developed at the narrow impedance bandwidth observed,
researchers throughout the world have proposed and investigated procedures
that will overcome the inherent limitation of the microstrip patch: all with
varying degrees of success and all with other compromises that had to be
made to achieve their goal.

In this chapter we investigate in detail some of the procedures to


enhance the bandwidth of a microstrip patch. From the proper choice of
materials for a single layer case giving a few percent enhancement to multi-
layer, multi-element patches with bandwidths in excess of an octave. To be
consistent with the title of this book, only broadband and wideband antennas
based on microstrip patch technology are presented. Other broadband printed
antenna solutions, such as tapered slots [1] and quasi-Yagi antennas [2] are
not examined and the reader can investigates these antennas in several papers.
Also for the sake of consistency with the book title, several examples of
developed antennas from the bandwidth enhancement procedures to be
reviewed are given, as well as design hints and performance trends when
appropriate. For the methods where the impedance bandwidth is truly
expanded (beyond 50 % of the operation frequency), a more thorough
investigation of the radiation performance is given. As was mentioned in
Chapter 1, the radiation bandwidth of a microstrip patch antenna is
broadband, typically a lot wider than the impedance bandwidth, however, for
these extreme cases this no longer holds. In fact there will be cases presented
where the impedance bandwidth is larger than the radiation/gain bandwidth.

A general 'rule of thumb' on how to increase the bandwidth of a


microstrip patch antenna is appropriate to state here: to substantially improve
the bandwidth of a microstrip patch antenna, more resonant radiators are
necessary. Importantly these elements must be appropriately coupled and this
is one of the objectives of.this chapter, namely, to show the reader how to do
this.

R. B. Waterhouse, Microstrip Patch Antennas: A Designer’s Guide


© Springer Science+Business Media New York 2003
This chapter is organized as follows: Section 3.2 presents some
general procedures for increasing the bandwidth of a single layer microstrip
patch antenna. These are derived from the trends presented in Chapter 2.
Section 3.3 explores horizontally coupled parasitic patches and is the first
forte into the concept of coupled resonators. In Section 3.4 direct contact
stacked patches are thoroughly investigated. Here material selection, design
trends and several examples are given. Also in this section an attempt to
qualitatively measure the performance of a broadband patch antenna is
provided. Optimization procedures are applied to a stacked patch antenna to
determine the best choice of laminates to give the broadest bandwidth. In
Section 3.5 large slot excited patches are investigated and a design example is
given. Techniques to counter some of the issues associated with using a large
slot are also introduced and once again design examples are given. Section
3.6 investigates Aperture Stacked Patches (ASPs), probably the widest
bandwidth patch antennas to date. Here bandwidths in excess of an octave
can be achieved. Material selection and design procedures and design
examples are given in this section. Once again, methods to overcome
radiation issues associated with using a large slot are presented and examples
of the newly formed antennas given. A dual polarized example is also
provided. Finally Section 3.7 presents an ultra-wideband antenna based on
microstrip patch technology. The Ultra Broadband Printed (UBP) antenna
consists of a diplexer and two aperture stacked patch antennas . This antenna
can produce a 10 dB return loss bandwidth of almost 120 % with constant
gain of 6.5dBi. A relatively uniform radiation pattern over entire band was
also achieved.

An important point needs to be made: all the procedures described in


this chapter have been applied to conically shaped patches (rectangular,
circular, etc). Even broader bandwidths may be achievable using other
shapes, such as those discovered using genetic algorithms [3), albeit at the
cost of complexity and radiation performance.

3.2 Intuitive Procedures

3.2-1 Intrinsic Techniques

From the general performance trends presented in Figure 2.4.1 there


are two intrinsic procedures that can be used to enhance the bandwidth of a
conventional direct fed microstrip patch antenna: use a thick laminate as the
substrate for the antenna; and make sure the dielectric constant of the laminate
is low, close to one. Doing so can enhance the bandwidth of a direct contact
microstrip patch antenna to close to 10 %. A series of somewhat obvious

70
questions arise, how far can this envelope be pushed and what are the limiting
factors?

Figure 3.2.1 shows the impedance bandwidth of a probe fed circular


microstrip patch antenna mounted on Rohacell foam (s, = 1.07) of thickness
0.06 1..0• The 10 dB return loss bandwidth was measured as 8 %. The
dimensions of the antenna are given in the figure caption. The gain of the
antenna was measured as 8.0 dBi. There an important points that need to be
made here: even for this somewhat moderate improvement in bandwidth, an
accurate numerical analysis is required . This is necessary to accurately model
the complex current associated with the discontinuity between the feed and
the patch conductor. Also as the thickness of the substrate becomes larger, it
is imperative that the currents on the feed itself (for a probe fed case the z-
directed currents) are modeled.

Figure 3.2.1 Impedance Response of Probe-fed Patch Etched on Electrically Thick


Substrate (0.06 1..0)

The bandwidth presented in this case is pretty well the inherent limit
of a direct contact patch antenna, without doing any other modifications to the
patch (stacking, match structures, etc..). This can be attributed to the
inductive nature of the feeding mechanism for the antenna. As the thickness
of the substrate is increased, the inductance of the feedline becomes larger and
it shifts the impedance locus of the antenna to a more inductive state. This
can be graphically represented on a Smith chart, as shown in Figure 3.2.2. As
the thickness is increased, it becomes more and more difficult to make the
antenna resonate, thereby reducing the efficiency of the radiator. For a probe-
fed solution, as the thickness of the substrate is increased, the location of the
probe for a 50 .Q match at resonance must be moved closer to the edge of the
patch. In fact a similar trend must be applied to an edge-fed patch, namely to
71
achieve a 50 Q matched resonance on thicker material, the insert length, must
be reduced, if not removed. Another consequence is that the patch conductor
size becomes smaller. This is because the antenna is more inductive in nature
and so for resonance, the operating frequency is where a thinner antenna
would normally be in a capacitive state, namely at a frequency above its
natural first resonance.

Figure 3.2.2 Impedance Response of Probe-fedPatches

Using this intrinsic procedure to improve the bandwidth of the


microstrip patch is not advised for edge-fed solutions. This is because the
feed network starts to contribute to the radiated power, creating issues with
spurious radiation. As the substrate thickness is increased and the dielectric
constant of the material reduced, the feedline becomes wider. This together
with the fact that the microstrip patch becomes smaller leads the patch
contributing less and less to the overall radiated power. This is very evident
at high microwave frequencies. Probe-fed solutions too suffer from radiation
issues, although not as dramatic as the edge fed patches on low dielectric
constant, thick substrates. The cross-polarized radiation increases as the
thickness increases, because of the fields radiated by the probe feed, which
begins to radiate like a monopole as its length is increased, thereby increasing
the H-plane cross-polarization levels.

3.2-2 External matching structures

External matching networks were used to increase the inherent


narrow bandwidth of the microstrip patch and were introduced to microstrip
patch technology in the mid- to late 1980s [4]. These structures are typically
single or double stub matching circuits and bandwidths approaching 15 % are
72
achievable, although this will depend on the material being used for the
substrate of the patch and the type of feed mechanism. Here the input
impedance of the printed antenna is treated as a load impedance and the
designer users sections of transmission line and open/short stubs to match the
impedance response to 50 Q. By using such a technique removes the
necessity of matching the input impedance of the antenna directly to 50 Q.
Thus some of the difficulties encountered in the previous sub-section are no
longer a concern. An accurate analysis of the patch antenna is still required to
make this technique a success as an accurate representation of the load
impedance of the matching network is required.

One potential flaw of applying the matching network to an edge fed


patch solution is that the stubs and transformers associated with this feed can
radiate and if the edge fed patch mounted on low dielectric constant material,
these antennas will suffer from spurious radiation from the feed.

Using matching structures are/were quite common when developing


arrays of microstrip patches. Trimming stubs are placed near the radiating
element to allow the designer tweaking of the input impedance via modifying
the stub to account for any uncertainties in the properties of the material and
also mutual coupling effects that may have not been considered in the analysis
of the antenna. Such a technique is not a preferred option as it increases the
labor required to develop the antenna. Full-wave simulators have alleviated
this issue to a slight degree.

Matching structures do offer the designer an extra degree of freedom


and in this book we will give several examples where matching structures are
used. One such example is for cases when trying to generate circular
polarization from a single feed point on a microstrip patch antenna. It turns
out that minimal axial ratio for these antennas does not always correspond to a
minimum in return loss. So a matching circuit can be used to match the
impedance of the antenna whilst still producing an antenna with a good
circular polarization response.

3.2-3 Non-contact fed patches

As mentioned previously, the non-contact feed methods are inherently


more broadband than the direct contact feed procedures and this may be
attributed to the lack of a current discontinuity. Having said this, both
proximity and aperture coupled patches do have their limitations and it is
really only the aperture coupled patch that has shown a true broadband nature.
The reason for the large bandwidths that can be achieved with this style of

73
feeding a microstrip patch radiator are relatively straightforward: there is
more than one radiator/resonator in the broadband solutions. More will be
discussed and presented on this issue in subsequent sections, however it is for
this reason that has seen aperture coupled patches (and variants of this printed
antenna) become the most widely used microstrip antenna.

Since their formulation, proximity coupled patches on the hand have


not shown as much promise to provide broadband solutions. Difficulties in
achieving bandwidths in excess of 8 or so percent, without the aid of
matching circuits have limited the use of this microstrip patch. The issue
comes down to the coupling mechanism. If a thick laminate is used for the
antenna layer, then it .is difficult to couple power from the feed circuit to the
patch via the open circuit stub. To make the stub a better radiator and
therefore couple more power to the patch, the feed layer must be made thicker
and so spurious radiation from other feed-lines becomes an issue. Having
said this, using a fairly simple matching structure, such as a quarter-wave
transformer, and a thicker antenna layer can easily doubling the bandwidth
can be realized. Figures 3.2.3 and 3.2.4 show the return loss performance of
such a case. A thicker antenna layer (increased by a factor of 2) and a simple
impedance transformer to change the 100 Q input impedance of the proximity
coupled patch to 50 Q were used in conjunction to enhance the bandwidth
(see Figure 3.2.3 for a schematic diagram of the matching structure). The
measured 10 dB return loss bandwidth has been increased from 8 % to 14 %.

···................... ....
i~~~~~~~----l-
: :
- i .\ i
Input

50 n microstrip line
Patch (on coupled layer)

Figure 3.2.3 Matching Structure Usedto Enhance Bandwidth of Proximity-Coupled


Patch

An L-shape excitation method has recently been introduced where an


L-shape probe is electromagnetically coupled to a patch residing on a
dielectric layer above the feed [5]. This technique seems to have resolved
some of the issues associated with the proximity coupled patch at the
74
expensive of complexity and can probably be interpolated as a hybrid
proximity coupled/stacked patch antenna.
o
-5

-.
-10
~
L/- r--.
~15
:::'20
\ / \
o
..J
...c-25
1- Experiment / \
~
I
'G>-30
II:
-35

-40

-45
32 33 34 35 36 37 38 39 40
Frequency (GHz)

Figure 3.2.4 ReturnLoss Response of Proximity CoupledPatch with Matching


Structure

3.3 Horizontally Coupled Parasitic Patches

3.3-1 Introduction

In the early 1980s horizontally coupling parasitic patches to the


excited patch was proposed and investigated [6]. The procedure uses the
.philosophy that if the resonant frequency of the coupled element or elements
is slightly different to that of the driven patch, then the overall bandwidth of
the antenna could be enhanced. Figure 3.3.1 shows an example of a driven
probe-fed rectangular patch with two parasitic patches positioned either side
of the excited patch in the y-axis direction. The critical parameters here are
the lengths and widths of each patch for the control of resonant frequency and
bandwidth, and importantly the gap between the elements. The gaps tend to
control the coupling between the patches and therefore the tightness of the
resonant loop (or loops) in the impedance locus of the antenna.

Bandwidths in the order of 20% have been achieved using this


enhancement technique, although there are several shortcomings of using
parasitically coupled patches. Firstly, to achieve these reasonable
bandwidths, wide parasitic elements are required making the overall size of
the printed antenna configuration electrically large and therefore difficult to
develop an array without incurring grating lobe problems. Also the radiation
patterns of parasitically coupled patches tend to be somewhat distorted across
75
the useful impedance bandwidth, due to lack of symmetry of the generated
currents with respect to the center of the printed antenna.

1:d
f
Figure 3.3.1 Horizontal Parasitically Coupled Patches

Figure 3.3.2 shows a case where this technique can give comparable
bandwidth enhancement with that of a stacked patch configuration (which we
will give more detail on in the next section). From Figure 3.3.2, for moderate
bandwidths the horizontally coupled procedure can function quite well,
although the overall size of this antenna is close to 3 times as large as the
stacked version, making it difficult to use as an element of an array (as
mentioned before). Thus a very critical limitation of this procedure is the
large horizontal dimension.

0.7 .--- - - - - - - -- - -----,

f',
0 .6 '"
r" \
~
.... O.S
" "
~, \ VSWR. H
I '
/
/
t::: 0 .4
\... .. - - Con"YCDlionai /
~ '\, ' \ - - Gap c<>upled ' /
U t .. ... Stecked I' /
=
0
0.3
'':::
~
'ii
c:.: 0.2

0 .1

I 1.03 1.06 1.09 1.1 2 1.1 S

Nonnalized Freq uency

Figure 3.3.2 Comparison of Impedance Responses for: single layer. gap


coupled and stacked patches

76
3.3-2 Annular Ring Loaded Shorted Patch

3.3-2.1 Introduction

A small microstrip patch incorporating a single shorting pin located in


close proximity to its coaxial feed has been developed [7]. This small printed
antenna will be presented in detail in Chapter 6. The fundamental advantage
of the shorted patch is the significant reduction in antenna real estate required
compared to a conventional microstrip patch. However, this is achieved at the
cost of decreasing the impedance bandwidth and the gain [7]. To enhance the
bandwidth of the shorted circular patch antenna, a parasitic concentric ring
can be placed about the shorted circular patch on the same substrate.
Although the concept of loading a printed antenna with an annular ring is not
new [8], it is very important to differentiate the proposed configuration and
that presented previously. In [8], the conventional circular patch weakly
excites a higher order mode on the annular ring rather than the more dominant
mode, namely, the TM ll mode. Hence importantly the improvement in
bandwidth is not significant. The reason for this is the physical size of the
circular patch is too large to excite the dominant mode on the ring. Due to the
small area of the shorted patch, it is possible to excite the TM lI mode on the
ring and thereby utilize the resonances of both radiators to significantly
improve the bandwidth.

3.3-2.2 Antenna and Theory

Figure 3.3.3 illustrates a probe-fed circular microstrip patch antenna


that incorporates both a shorting pin and an annular ring. Here the patch of
radius R is mounted on a substrate of thickness d and with a dielectric
constant tr. The coaxial probe feed of radius ro and the shorting post of
radius ros are located at (xp,yp) and (xps, yps) from the center of the patch,
respectively. The annular ring is of outer radius, a and inner radius, b and is
placed concentrically about the circular patch.

To accurately predict the electrical characteristics of the modified


circular patch, the spectral domain electric field integral equation technique
was implemented [7]. The analysis considered here incorporates entire
domain basis functions on the circular patch and attachment modes to
accurately model the currents through both the feed and the shorting pin [7].
Entire domain basis functions are also used to model the currents on the
annular ring [9]. By using two special attachment modes to model the
shorting pin and the probe feed, the continuity of current between the pins and
77
the circular patch, the finite radius of the pins and the coupling between these
pins are accurately modeled.
y

Figure 3.3.3 Probe-fed circular microstrip patch antenna with a concentric annular ring
and a single shorting pin

3.3-2.3 Design and Experiment

The design procedure for this enhanced bandwidth printed antenna is


similar in principle to any antenna relying on two resonators. The driven
antenna, namely the shorted patch, was designed for 50 Q impedance match
at the desired center frequency without the presence of the parasitic element,
namely the annular ring. Using the cavity model, a ring was chosen such that
the dominant TM II mode resonates at a similar frequency. Consideration of
the physical size of both the shorted patch and the ring must taken into
account to ensure strong coupling between the two radiators. Fine
adjustments were then made to the probe position and other dimensions of the
printed antenna to achieve the desired input impedance behavior. Generally,
the trends that are typically observed for other forms of parasitic printed
antennas hold for this configuration. For example, to tighten the resonant
loop associated with the two radiators, the spacing between the elements must
be increased, such that the coupling is reduced. It is important to note that the
presence of the annular ring relaxes the tolerances of the position of shorting
post with respect to the coaxial feed to achieve 50 Q resonance.

78
The predicted and measured input impedance behavior of a shorted
patch with an annular ring incorporating 5 rom ROHACELL 71 HF is shown
in Figure 3.3.4 (refer to the caption for dimensions). The experimental results
were measured using a Wiltron 360B network analyzer. As seen from these
results very good agreement between experiment and theory was achieved.
The predicted and measured 10 dB return loss bandwidths were 6.6% and
6.8% respectively. Itshould be noted that to fabricate the antenna, the shorted
patch and ring conductors were etched on 0.254 rom thick Rogers 5880
Duroid which was adhered to one side of the foam and a brass plate was
affixed to the other to provide a ground plane. The electrically thin Duroid
layer was not considered when designing the antenna and may be the
contributing factor to the minor frequency shift between the measured and
analytic results (approximately 1%) as well as the slight closure of the
resonant loop. The co- and cross-polar patterns in the cardinal planes of the
shorted patch with a concentric ring were measured and the results were very
similar to a conventional microstrip patch antenna mounted on foam The
measured 3 dB beamwidth in the E- and H-planes were 84° and 75°,
respectively compared to the predicted values of 86° and 70°. The radiation
patterns are shown in Figure 3.3.5. The measured H-plane cross-polarization
level was always more than 20 dB below the co-polar level.

Figure 3.3.4 Predicted and Measured Impedance of the Annular Loaded Shorted
Microstrip Patch (Er = 1.13, tan 0 =0.001, d =5.0 mm, R = 11.8 mm, xp =8.0 mm, Yp =0, xps
=9.3 mm, Yps =0, TO =0.325 mm, ros =0.325 rnm, a =31.5 mm, b =17.0 mrn)

To put these results into perspective, it is useful to compare the


theoretical characteristics of the novel patch configuration to both the
conventional circular microstrip patch and a shorted patch fabricated using the
79
same substrate. Table 3.3.1 presents the impedance bandwidth, gain and
physical size of the three antennas. As can be seen from theseresults, the
shorted patch with an annular ring has enhanced bandwidth compared to the
other patches and is smaller than the conventional patch. Thus by applying
the horizontally parasitically coupling procedure to two small printed
antennas, a larger bandwidth antenna results than for a conventional patch of
the same overall dimensions.

, --
-2 .."j,."
i -4
:,. /
.,. ~'
~
If -6
.:,,',,
" - - E-planeExp .

: I - •• - ,. E-plane Theory
~ .. I
" I
'; -8 I
.. . . . . .. "-plane Exp.

~ I
I
- - - - " -plane Theory
-10

-12 L- ----.J

-72 -54 -36 -18 0 18 36 54 72


Theta (degrees)

Figure 3.3.5 Co-polar Radiation Patterns of ARL

Printed AntennaTvue 10dBReturn Loss Bandwidth Gain (dBD Area (cm2)


Conventional Circular Patch 4.2% 9.4 51.5
Shorted Circular Patch 2.1% 4.0 4.3
Proposed Printed Antenna 6.6% 8.4 31.2

Table 3.3.1 Comparison of the characteristicsof a conventional circular microstrip


patch, a shorted circular microstrip patch and the proposed printed antenna mounted on the
same substrate (d = 5 rom, Er = 1.13)

3.3-3 Coverlayer Annular Ring Loaded Shorted Patch

3.3-3.i introduction

The Annular Ring Loaded (ARL) shorted patch antenna presented


above has several key advantages over conventional microstrip patches, in
particular, smaller size and larger bandwidth. However, this ARL shorted
patch is still too large for many applications, including mobile communication
handset terminals. In this section we incorporate size reduction techniques
used in printed antenna technology to the ARL shorted patch. By doing so,
80
the size of the new ARL shorted patch is compatible to that needed for
handset terminals and the bandwidth achieved is more than required. Both
theoretical and experiment input impedance results of this ARL shorted patch
antenna are presented and the far-field radiation characteristics are also given.

3.3-3.2 Antenna and Results

Figure 3.3.6 illustrates a probe-fed circular shorted microstrip patch


antenna that incorporates an annular ring. Here the patch of radius R is
mounted on a substrate of thickness di and with a dielectric constant Erl. The
coaxial probe feed of radius r0 and the shorting post of radius ros are located
at (xp,Yp) and (xp s' Yps) from the center of the patch, respectively. The
annular ring of outer radius, a and inner radius, b is placed concentrically
about the circular patch conductor. A dielectric coverlayer of height d z and
with a dielectric constant Er 2 is situated above these printed conductors. To
accurately predict the electrical characteristics of this printed antenna, the
spectral domain electric field integral equation technique was again
implemented.

Figure 3.3.6 Annular Ring Loaded Shorted Patch with Coverlayer

The design procedure of an ARL shorted patch was given in the


previous sub-section. Once again, it is important to note that the presence of
the annular ring relaxes the position tolerances of the shorting post with
respect to the coaxial feed to achieve 50 .Q resonance. To reduce the size of a
printed antenna, it is well known that the dielectric constant of the substrate
should be increased (as was pointed out in Chapter 2). To do this without
compromising the achievable bandwidth, a substrate with a dielectric constant
of 2.2 was used. To further reduce the size of the antenna, a coverlayer of Erz
= 2.2 was incorporated. Once again, using a dielectric material here with too
high a dielectric constant tends to reduce the impedance bandwidth, as well as
the efficiency.
81
The predicted and measured input impedance behavior of the shorted
patch with an annular ring is shown in Figure 3.3.7 (refer to the figure
captions for dimensions). The experimental results were measured using a
Wi/tran 360B network analyzer. As seen from these results very good
agreement between experiment and theory was achieved. The predicted and
measured 10 dB return loss bandwidths were 9.9% and 10.1%, respectively.

Figure 3.3.7 Predicted and measured impedance of the annular loaded shorted
microstrip patch (Er 1 =Er2 =2.2, tan oJ =tan ~ =0.001, d J =6.096 mm, d2 =0.254 mm, R =
8.2 mm, xp =4.1 mm, Yp =0, xps =5.3 mm, Yps =0, r o =0.325 mm, ros =0.325 mm, a =25.8
mm, b = 9.6 mrn)

The co- and cross-polar patterns in the cardinal planes of the CARL
shorted patch were measured and the co-polar E- and H-plane results are
shown in Figure 3.3.8 with the antenna mounted on a 30 x 30 cm2 ground
plane. The measured 3 dB beamwidth in the E- and H-planes were 85° and
80°, respectively compared to the predicted values of 90° and 78°. The slight
asymmetry in the E-plane pattern can be attributed to the presence of the
shorting pin. The measured H-plane cross-polarization level was always more
than 20 dB below the co-polar level. The measured and calculated gains of
the printed antenna were 6.2 dBi and 6.4 dBi, respectively. It is pertinent to
compare these results with a conventional circular probe-fed rnicrostrip patch
designed using the same material. Here the 10 dB return loss bandwidth is
only 4.7 %, the diameter is approximately 6 em and the gain is 7.1 dBi. Thus
the CARL shorted patch offers a bandwidth enhanced variation of the single
layer microstrip patch.
82
o

180

Figure 3.3.8 Measured E-plane and H-plane Radiation Patterns of CARL Patch

3.4 Stacked Patches

3.4-1 Introduction

A fundamental limit of the direct contact patch antenna has been the
achievable impedance bandwidth. Here, 10 dB return loss bandwidths
approaching 8 % have been obtained for a single element configuration if
foam is used as the substrate, however this is by no means adequate for most
wireless systems. Vertically parasitically coupling another printed antenna to
the driven patch, or stacking, is one procedure that can alleviate this problem
to a degree. It is interesting to note that for early versions of the direct feed
stacked patch, only moderate bandwidths, less than 15 % were achieved [10].
It can be postulated that this barrier was due to the inductive nature of the feed
and so impedance control of the radiator when thick substrates were used was
very difficult. Also due to complex nature of the currents associated with the
discontinuity between the patch conductor and the probe feed, to accurately
model this antenna configuration required sophisticated, accurate numerical
tools. As was pointed out in the Chapter 1, not only do designers need
accurate analysis tools, the software needs to be fast, such that many iterations
can be processed. It wasn't until the late 1990s that such analysis tools were
available and that the true potential of the direct fed stacked patch antenna
could be realized.

83
In this section design trends for broadband direct contact feed stacked
patches are established. Here broadband is defined as a 10 dB return loss
bandwidth greater than 25 %. The objective of this section is to develop a
robust design philosophy for stacked patch designs including how to start the
design procedure and which dimensions of the printed antenna should be
varied to achieve the desired impedance response. The characteristics and
impedance trends of probe-fed stacked circular, rectangular and annular rings
will be presented. It will be shown that all three configurations behave
effectively in the same manner and varying one parameter on say the circular
patch geometry has the same effect when changing a similar parameter on the
other geometries . It will be shown that annular rings and rectangular stacked
patches do offer an extra degree of freedom compared to circular patches .
This gives easier impedance control at the expense of slightly reduced
bandwidth.

3.4-2 Geometries, Theory and Validation

Figure 3.4.1 shows schematic diagrams of the considered probe-fed


stacked circular, rectangular and annular ring patches. As can be seen from
this figure, the bottom patch of each configuration is fed by a coaxial feed
through the ground-plane. To accurately analysis each probe-fed stacked
patch arrangement rigorous, full-wave analyses were devised based on the
Spectral Domain Integral Equation approach (refer to [9, 11, 12] for details of
the formulation) . It should be noted that each of these formulations accurately
models the discontinuity associated with the probe-patch interconnect. The
computational time of each formulation varied according to the inherent
symmetries of the configurations, which is important when trying to design
antennas. However, it should be noted that most commercial .packages today
can analyze all three styles of printed antennas in a relatively modest
timeframe. To validate the incorporated analyses, broadband probe-fed
stacked rectangular, circular and ring patch antennas were designed and
developed.

Figure 3.4.1 Side-views of Probe-fed Stacked Patch Geometries: (a) Rectangular Patches; (b)
Circular Patches; (c) Annular Ring Patches
84
Stacked Rectangular Patches

Figure 3.4.2 shows the predicted and measured impedance behavior


of the antenna (refer to the figure caption for the relevant dimensions). As
can be seen from Figure 3.4.2 very good agreement between experiment and
theory was achieved. The predicted and measured 10 dB return loss
bandwidths of the antenna were 27 % and 26 %, respectively. The bandwidth
presented here is considerably larger than previous probe-fed patch results
and can be attributed to taking advantage of increasing the thickness of the
lower dielectric layer and accurately modeling its effect on the overall
impedance response, as stated before. The slight discrepancy between theory
and experiment is possibly due to the thin layer of dielectric material needed
to etch the top patch conductor (s, = 2.2, d = 0.127 mm). This is also the
reason for the slight shift in the frequency response, by less than 0.6 %.

Figure 3.4.2 Predicted and Measured Input Impedance of Probe-fed Rectangular


Stacked Patch (parameters : Erl = 2.2, d. = 1.524 mm, tan B1 = 0.001, ea = 1.07, d2 = 2.5 mm,
tan ~ = 0.001, L 1 = 13.5 mm, WI = 12.5 mm, ~ = 15.0 mm, W2 = 16.0 mm, xp = 5.4 mm, ro=
0.325 mm)

Stacked Circular Patches

A stacked circular patch antenna was designed for 1.9 GHz using the
full-wave analysis and the parameters are as follows: Erl =2.2, d, =3.048 mm,
R1 =31 mm, x, =27.5 mm, YP =0, ro =0.325 mm, Erz =1.15, dz =9 mm, R z =34
mm. The simulation and experimental impedance behavior are shown in
Figure 3.4.3. The theoretical 10 dB return loss bandwidth was 20% compared
to a measured bandwidth of 19%.

85
Figure 3.4.3 Input Impedance of Stacked Probe-fed Circular Patches

It can be seen from the Smith chart plot that the resonant impedance
loop for experimental result was slightly shifted from the predicted curve.
This can be attributed to the thin layer of material required for mounting the
upper patch on the foam that was not included in the simulation. Overall,
these results are acceptable for the application at hand (here a micocell base
station) since the required bandwidth and operating frequency have been met.
The far-field E-plane and H-plane for co-polar and cross-polar radiation
patterns were measured at 1.9 GHz. The measured results are very similar to a
conventional microstrip patch antenna. The plots of E-plane and H-plane co-
polar are shown in Figure 3.4.4 and Figure 3.4.5, respectively. The cross-polar
levels in the E- and H-plane were 18 dB and 14 dB lower respectively than
the co-polar levels.

The broadside gain of the stacked circular patch antenna was


measured with reference to a Scientific Atlanta standard hom antenna model
No.l2-1.7. The measured gain was 7.6 dBi compared to a theoretical value of
8.3 dBi. This difference can be due to the finite sized ground plane used in the
experiment. The analysis used to design the antenna assumes an infinite
ground plane and therefore ignores any diffraction due to the edges of the
ground plane. Here the dimensions of the ground plane were 110 x 110 mm'.

86
o

180

Figure 3.4.4 Measured far-field E-plane co-polar radiation pattern

180

Figure 3.4.5 Measured far-field H-plane co-polar radiation pattern

Stacked Annular Rings

The impedance variation of a single layer printed annular ring has


been thoroughly examined in the literature both experimentally and
theoretically [13, 14]. For this reason and for the sake of brevity an
investigation will not be included here. There are however, several features
of a probe-fed printed annular ring that are pertinent to the more complicated
stacked patch design examples presented here. In particular when operating

87
in the fundamental mode (TM Il ) the probe-fed printed ring has a high input
impedance at resonance. Thus there is a strong coupling between the feed and
the patch, that is, the radiator is overcoupled. This is why proximity coupling
a micros trip line to the ring is the preferred option, where by moving the
feedline away from the patch the coupling to the radiator can be reduced (and
hence the input impedance). The strong coupling between a probe feed and a
printed annular ring is extremely advantageous, however, when considering
stacked configurations. Typically, for any stacked patch configuration, the
lower resonator is deliberately over-coupled and the top patch is used to
effectively impedance match the entire configuration. Thus the high
impedance encountered for a probe-fed single layered ring is very suited to a
stacked version. ' Importantly too, the high impedance level is relatively
independent of substrate thickness. Thus thick dielectric layers below the
lower resonator can be used without the penalty of poorly behaved input
impedance due to the inductive nature of the feed. This is in direct contrast to
probe-fed circular and rectangular stacked patch configurations where the
lower substrate cannot be made too thick. Another advantage of the annular
ring over the circular patch version is the extra degree of freedom in design,
namely the inner radius. This can be varied independently of the outer radius
to provide fine tuning of the impedance variation of the antenna. The effect of
this extra degree of freedom is similar to that of varying the width of a
rectangular patch.

Figure 3.4.6 shows the predicted and measured input impedance plot
of a probe-fed stacked annular ring antenna (refer to the caption for the
relevant dimensions). As can be seen from this figure, very good agreement
between theory and experiment was achieved. The predicted and measured
10 dB return loss bandwidth was 21 % and 22 % respectively. The E- and H-
plane co-polar and cross-polar far-field radiation patterns were measured and
are very similar to a conventional patch radiation characteristics.
Comparisons of these measured values and the predicted results for the E-
plane and H-plane co-polar patterns are shown in Figure 3.4.7. The cross-
polarization levels were greater than 20 dB below these values in both planes.
The gain of the stacked configuration was measured as 8.2 dBi, compared to a
theoretical value of 8.5 dBi.

It should be noted that although it appears the rectangular stacked


patches are broader in bandwidth, which is consistent with what was
presented in Chapter 2, a comment needs to be made with respect to the cases
considered. It is important to compare "apples to apples " and for the three
cases presented the overall electrical thicknesses of the substrates were:
rectangular: 0.1 Ao; circular: 0.076 Ao; and ring: 0.089 Ao. Thus the relative

88
bandwidths of the three cases can readily be explained . It is now pertinent to
examine some design trends.

Figure 3.4.6 =2.2, tan 01 =0.001, dj =6.096 nun, aj =10.0 nun, bl =29.0 nun, Pp
Er l

=21.0 nun, ~ =0°, ro =0.325 nun, Er2 =1.07, tan 02 =0.001, dZ =8.0 nun , aZ =14.0 nun,
=
bZ 31.0 mm

180

Figure 3.4.7 Predicted and Measured Radiation Patterns of Stacked Probe-fed Annular
Rings

89
3.4-3 Design Trends

As mentioned previously the main purpose of this section is to


develop a design strategy for probe-fed stacked patch antennas. It should be
pointed out that the trends derived here could be applied to edge-fed
configurations due to the similarity of these feeding techniques as well as
cavity-backed configurations.

A. Substrates

A lot of research over the years has focused on what are the best
materials for the dielectric layers for microstrip patches. It has been shown
that thick, low dielectric constant laminates tend to give the largest bandwidth
responses and good surface wave efficiencies. Although it should be stated a
combination of high dielectric constant and low dielectric constant materials
can also yield good impedance behavior (we will investigate this further in
Chapters 4 and 6). Perhaps the easiest choice of materials to work with is a
combination of low dielectric constant material (say e, =2.2) and foam. This
tends to give the largest impedance bandwidth and also has the simplest
design procedure (We will examine this in a subsequent section in more
detail, when we optimize the substrate layers to give the best bandwidth
performance). Using two layers of foam does give better surface wave
efficiency, however at the expense of bandwidth reduction. We have found
that using the higher dielectric constant laminate for the substrate of the lower
patch gives enhanced results. This can be attributed to the coupling of the
modes on each patch. Unlike in [15] where the 'mutual coupling' terms are
the important factors in the design, here the resulting bandwidth is
predominantly governed by the current distribution on the lower patch. The
broadest bandwidths were achieved when the first order mode on the lower
patch is considerably greater in magnitude than the corresponding mode on
the top patch. In other words the top patch is 'loosely coupled' . To do so,
requires the lower dielectric layer to have a greater dielectric constant than the
upper layer. If the layers have the same dielectric constant or the upper layer
has a higher value, the modes are too strongly coupled yielding very tight
resonant loops and resulting in lower impedance bandwidths.

The thickness of each layer in a direct contact stacked patch


configuration plays an important role in the overall achievable bandwidth.
Consider the lower dielectric layer, d, in Figure 3.4.1. Obviously the thicker
the lower layer, the greater the bandwidth, however here is where the first
trade-off must be made. The lower patch must be over-coupled to achieve a
good impedance response. Thus, rather than the lower patch being designed
for minimum return loss in the desired band, the patch should be strongly
90
capacitive over this frequency range. An obvious way of doing so is to
position the feed near one of the radiating edges and operate the patch above
resonance, however this has limited success. An important trend is as
follows: the thicker the substrate, the less capacitive the impedance of the
patch can become. Thus if the lower substrate is too thick, when the
parasitically coupled patch is positioned onto the configuration, the overall
impedance trend will appear inductive and the impedance control and hence
achievable bandwidth will be limited. This effect is shown in Figure 3.4.8 for
two stacked circular patch configurations. The first Smith Chart shows the
impedance loci for a single layered patch and the effect of locating the
parasitically coupled element. The presence of the second element moves the
very capacitive impedance region of the single patch locus to near a matched
condition. The bandwidth of this stacked patch configuration is 28 %. The
second Smith Chart shows the case where an electrically thicker dielectric
layer is used below the lower patch and the resulting effect on the impedance
locus of the stacked configuration. Note the impedance after the first
resonance of the second single layer example is considerably less capacitive
than the first one. To achieve flexible impedance control of the antenna a thin
layer is required for the lower material. Thus there is a compromise between
impedance control and achievable bandwidth.

Stacked patch

Single layer

(a)

91
(b)

Figure 3.4.8 Effect of d l on Stacked Patch Design (frequen cy range: 6 - 11 GHz): (a)
parame ters: Erl =2.2, d l =1.524 mm, tan 0 1 =0.001, er2 =1.07, d2 =2.5 mm, tan ~ =0.001 , R 1
=7.0 rnm, R2 =7.4 mm, xp =5.4 rnm, ra =0.325 mm; (b) paramet ers: Erl =2.2, d l =2.524 rom,
tan O. =0.001, e,z =1.07, d2 =2.5 rom, Ian ~ =0.001, R. =7.0 rom, R2 =7.6 rnm, xp =5.5
mm, ra = 0.325 rom)

Consider now the thickness of the second layer, d2• d2 essentially


controls the tightness of the resonant loop, or the interaction between the two
patches: the further the second patch is away from the lower patch, the tighter
the loop. Thus if a lower return loss limit is required (for example a 15 dB
return loss bandwidth), the thickness of d2 should be increased. Of course this
comes at the expense of a reduced frequency range of operation. The height
of d2 is very much dependent on that of d l : the greater d., the less freedom
available for the selection of d2• Typically if d l is too electrically thick (for
example the value used in the second Smith chart in Figure 3.4.8), d2 must be
electrically thick as well and therefore the overall bandwidth is not improved.

B. Starting Point

We have found through numerous simulations that to achieve


maximum bandwidths, d l should be approximately 0.04 Ao and d2 should be

92
approximately 0.06 1.0 . These values were obtained using the combination of
the dielectric layers suggested previously.

The dimension of the patches is obviously dependent on the geometry


of conductors used, however a reasonable starting point is to make the outer
dimensions of both conductors approximately Ai2 where Ag is a guided
wavelength in the lower dielectric material. Any of the typically used
approximations for a single layered patch [16] could also be utilized, however
at this stage in the design it is not that critical. As mentioned previously, the
lower patch must be over-eoupled, so the feed should be positioned for an
input impedance of approximately 250 Q at resonance. This resonance
should be at a frequency slightly less than the lower edge of the desired band.
Importantly, the impedance should have a negative reactance in the required
bandwidth, with the locus outside the VSWR 2.5: 1 circle. Doing so will
ensure that when the top patch is positioned, the overall impedance behavior
will not be too inductive. Once these steps have been carried out a resonant
loop for the stacked configuration should appear in the input impedance
response of the antenna. Unfortunately it will not typically be located around
the center of the Smith chart. The next subsection will outline how the loop
can be shifted to a better location.

C. Patch Conductor Dimensions

Once the resonant loop has been formed, adjusting the dimensions of
the patches can control its location on the Smith chart. In this section, only
circular patches are presented, however the same trends are achieved when
varying the patch conductor sizes of rectangular patches (LI> L z) and annular
rings (bj , b z). Figure 3.4.9a shows the effect of varying R) on the impedance
locus while keeping the other parameters of the stacked patch constant. Note
that an arc can be drawn through the center of the resonant loops and the
position of the local minimum of this arc is dependent on the thickness of d).
As R) is decreased the center of the resonant loop shifts from a high real
component to a lower impedance. Figure 3.4.9b shows the effect of varying
Rz on the impedance behavior of the stacked patch configuration, once again
keeping all other parameters constant. Here increasing R z increases the real
component of the input impedance. Thus changing R) and Rz have contrary
effects on the overall behavior. This is very important when designing the
stacked configuration for a particular frequency band. For example, if the
resonant loop occurs at the desired frequency band, however the real part of
the impedance at the center of the resonant loop is low, then Rz can be
increased and R) decreased proportionally to shift the locus without affecting
the desired bandwidth. Similar loci to that shown in Figure 3.4.9 have been

93
achieved for square and annular ring patches when varying the outer
dimensions and so for the sake of brevity, these will not be presented here.

Adjusting the probe position has only a second order effect on the
overall impedance behavior of the stacked configuration. A trend similar to
that shown in Figure 3.4.9b can achieved when increasing xp, however the
relative movement in the center of the resonant loop is by no means as
extreme. Thus changing the position of the probe really only provides a fme
tuninz mechanism.

(a) (b)
Figure 3.4.9 Effects of Patch Dimensions on Impedance Locus (frequency range: 6 -11
GHz): parameters: Er. = 2.2, d) = 1.524 mm, tan SJ = 0.001, ea = 1.07, d2 = 2.5 mm, tan ~ =
0.001, xp = 5.4 mm, TO = 0.325 mm; (a) R J variation (R2 = 7.4 mm); (b) R2 variation (R J = 7.0
mm)

D. Advantages ofRectangular and Annular Ring Patches

Rectangular patches and annular rings do have an extra degree of


freedom in their geometries compared to circular patches. These equate to
better impedance control in the design of broadband stacked patches. The
inductive nature of the impedance locus experienced when a thick lower
dielectric layer is used can be countered by decreasing the width of the lower
rectangular patch. This is equivalent to making the impedance of the single
layered patch more capacitive and thereby giving enhanced impedance
control, similar (although to a lesser degree) to using a thinner dielectric layer
as stated in Section 3.4-3A. The enhanced control can also be achieved in an
annular ring configuration by increasing at. the inner radius of the lower
patch. However, as mentioned before the price for achieving better

94
impedance control is reduced bandwidth, here typically by a few percent.
Simplistically this can be related to the size of the antenna. If we reduce the
volume of an antenna, for example reducing the width of a patch, the
bandwidth will decrease. This is indeed the case for the rectangular patches
where the width is reduced and for the annular rings where the inner radius is
increased. Varying the width of the top rectangular patch and the inner radius
of the annular ring has minor effects on the impedance loci of these stacked
patches.

3.4-4 Figure of Merit

3.4-4.1 Introduction

Having now developed a design strategy for direct contact stacked


patches, perhaps it is appropriate to 'step back' and attempt to qualify what
has been done. Throughout the history of microstrip patch antennas, there
have been several investigations on how to design these antennas and general
'rules of thumb' have been developed for the conductor dimensions.
Although the designs above yield good bandwidth and radiation properties,
however, a generic approach to the choice of material has yet to be developed.
Fundamental questions have remained unanswered including the appropriate
dielectric material selection for a particular bandwidth and efficiency
performance. Although the RF characteristics of a microstrip patch antenna is
dependent on the feeding mechanism, the dielectric material used to support
the radiator primarily governs the fundamental performance of this printed
antenna.

In this section we derive a figure of merit important for the


fundamental design of microstrip patch antennas. The quantity gives the
designer an understanding and limit of both the radiation efficiency and
achievable impedance bandwidth of the printed antenna. Importantly, the
quantity can be determined without the need for a full-wave numerically
intensive analysis and can be applied not only to single layer patches, but
stacked geometries as well. The figure of merit is valid for probe-fed, edge-
fed and proximity coupled patches. It also holds for aperture coupled patches
as long as the slot is not too large. In this section, the theory behind this
figure of merit will be derived for a circular patch geometry, although it can
easily be applied to other conductor shapes, such as rectangular or triangular.
Results for circular microstrip patches using various materials are provided,
including broadband stacked patches and stacked patches utilizing high
dielectric constant and low dielectric constant material combinations (which
will be considered in more detail in Chapters 4 and 6). From the results given

95
it will be evident that the derived figure of merit provides a good basis for the
design of micros trip patch antennas.

3.4-4.2 Theory

As mentioned previously in Chapter 2 and here, although specific


details such as patch shape and feeding method influence patch antenna
behavior, the major factor influencing the performance of a microstrip
antenna is the substrate upon which it resides.

Two important characteristics of a patch antenna are its quality factor


(Q) and surface wave efficiency (llsw). The quantity, llsw/Q can be interpreted
as a figure of merit for any particular dielectric material. Consider a circular
patch mounted on a substrate of thickness d and a dielectric constant of e, as
shown in Figure 3.4.10. To evaluate this figure of merit for any substrate, we
shall use an assumed current distribution corresponding to the TM II mode of a
circular magnetic wall cavity. This current distribution is given by:

-
J(p,rp)=a - -{Xli')
a
P - 1 P
Ji'(XII'-)cosrp-a?-ll(XII'-)sinrp
a p a
(3.1)

The total complex power emanated by this mode is given by:

~ot = -4n j pfJI (fJ)FI 2 (P) + G2 (P)F22(P)}tp (3.2)


o

Where 0 1 and G2 are the Tm and Tc Green's functions for the substrate
respectively and F I and F2 are the Fourier transforms of the currents given in
Equation 3.1 (please refer to [17] for more details of the formulation). The
power trapped in the surface wave modes is given by:

NSW( [ /16w+S ]]
Psw =~ - Re 4~ /16lf1 (P)F/ (P) +G2 (fJ)F/ (fJ)}pIp
(3.3)

where Nswis the number of surface waves.

From Equations (3.2) and (3.3) the power in the space wave can be
determined and therefore the surface wave efficiency can be evaluated.
96
The impedance of the mode can be defined as:

Z- -Prot
- 1 -- - R +]'X (3.4)
2//1
2

where I = P.d;.

It can be shown using simple circuit theory that the Q of the .mode is
given by:

Q= k o aX(ko)
(3.5)
2R(ko) ako

where the derivative, i)X/i)ko can be computed numerically and ko is the free
space wave number.

Figure 3.4.10 Single Layer Circular Patch Antenna

Thus the figure of merit for the circular microstrip can be computed
relatively easily as a function of the substrate parameters. It is important to
note that for this calculation the current coefficient of the mode is not required
and therefore a rigorous analysis is not needed.

To extend this calculation to a stacked patch geometry (as shown in


Figure 3.4.11), an important assumption must be made. This assumption is
similar to that given in the derived theory for a single layered geometry,
namely that there is only one dominant mode per patch. For a conventional

97
stacked patch configuration (such as a probe or edge-fed) this assumption is
valid even for relatively broadband cases. Thus for the stacked configuration,
the impedance is written in terms of a simple two port network from which
the Q can be readily determined. The surface wave efficiency can be
determined from Equations (3.2) and (3.3) for the multi-layered case and thus
the figure of merit can be calculated.

Figure 3.4.11 Stacked Circular Patch Antenna

A very important part of the figure of merit calculation is the


determination ofaXiako, in particular for the stacked cases. When a stacked
patch is designed for broad impedance bandwidth, a 'resonant loop' is formed
in the impedance response due to the interaction between the two patches. If
care is not taken where the derivative of the reactance of the patch is
calculated, negative values can result giving incomprehensible results for the
figure of merit. We have found that using a cubic spline to interpolate the
impedance points and the derivative of the reactance at three frequency
points, one near the design frequency (fo) and the other two at fo ± O.3fo
consistent results were achieved for the Q of the antenna.

3.4-4.3 Results

Table 3.4.1 summarizes the many cases examined. Firstly single


layered circular patches were considered. Although these trends relating the
material properties of the substrate to the RF performance of the antenna are
well known [18], it is important to confirm the validity of the proposed figure
of merit and also to provide a reference for other patch configurations. As can
be seen from Equations (3.2) and (3.3) the size of the patch does playa role in
the evaluation of the surface wave efficiency and the Q. For the single
layered cases, the radius of the patch R was assumed to be Agl4 (where Ag is
98
the guided wavelength taking into consideration the dielectric properties of
the substrate [19]). Consider the effect of the dielectric constant e, on the
performance of the antenna. As can be seen from Table 3.4.1 as the dielectric
constant of the substrate is increased the figure of merit decreases (cases 1 and
2). This is due to both the decrease in surface wave efficiency as well as the
increase in the Q. Now consider the effect of increasing the thickness of the
substrate (case 1 and case 3 in Table 3.4.1). As is evident from Table 3.4.1,
the figure of merit increases, as the decrease in Q is greater than the reduction
in the surface wave efficiency. Case 4 shows the results when using a high
dielectric constant substrate (e, = 10.4) and a relatively thick material (d :::::
0.03 Ao). For this case even though the Q has decreased, the 'Ilsw is also
relatively low (66 %) thereby resulting in a relatively low figure of merit.

Case Dielectric Properties Fiaure of Merit


1 d =0.0086 Ao; e, =2.5 0.011
2 d =0.0086 Ao, e, = 10.4 0.004
3 d =0.0267 Ao, e, =2.5 0.034
4. d =0.0324 Ao, e, =10.4 0.016
5 d t =0.02 Ao, Ert =2.2, d2 =0.06 Ao, fa = 1.07 0.164
6 d t =0.0324 Ao, Erl = 10.4,d2 =0.07 Ao, fa =1.07 0.350

Table 3.4.1 Summary of Figure of Merit Results

Figure 3.4.12 summarizes the general trends established using the


Figure of Merit approach on single layered circular microstrip patch antennas.
As can be seen from these graphs, that show the variations in hsw/Q as a
function dielectric constant and patch radius, the trends are consistent with
those given in Figures 2.4.1.

O. o.
o. Increasing lOr

mm
O. Decreasing Radius o. ~ d= 5.0
TJs.jQ 0.0
O.
."

• o.
~ 2.5 mm
0.01
~= 1.2 mm
o0 5 10 15 20 25 30 35 40 45 50 05 10 15 20 25 30 35 40 45 50 55

E, Patch Radius (mm)

Figure 3.4.12 Figure of Merit for Single LayeredPatches

99
Case 5 in Table 3.4.1 presents the probe-fed circular stacked patch
configuration designed in the previous section. Here a combination of RT
Duriod 5880 (e, = 2.2) and Foam (e, = 1.07) was used to produce an antenna
with approximately 23 % bandwidth. As can be seen from the figure of merit
for this antenna, it is significantly better than for the single layered cases. As
for the single layered examples in Table 3.4.1, the dimensions of the stacked
patches were determined taking into consideration the properties of the
dielectric materials [19].

Case 6 presents the results for a stacked circular patch configuration


utilizing a high dielectric constant for the lower layer and a low dielectric
constant material for the upper layer, which will be presented in Chapter 4.
This combination of material is appropriate for integrating patches with
MMICs and OBICs. The dimensions of the patches were chosen ignoring the
other dielectric layer, that is, R1 =Agl4 (where s, = 10.4) and R2 =Agl4 (where
e, = 1.07). As can be seen from Table 1 the results for the hi-lo stacked patch
are in fact better than for the previously considered stacked patch antenna.
This has been confmned experimentally, where the measured 10 dB return
loss bandwidth of the hi-lo stacked patch was approximately 30 % and the
calculated surface wave efficiency was comparable to that for case 5.

3.4-5 Material Optimization for Direct Contact Stacked Patches

3.4-5.1 Introduction

The results of the previous section are preliminary investigations into


examining the overall effect of the dielectric constant and thickness of the
materials for both single layered and stacked patch configurations on the
impedance bandwidth and radiation efficiency of the antenna. Although the
results presented for single layered configurations followed well-established
trends the stacked patch configurations confirmed the somewhat surprising
results, namely, that it was not necessary to use low dielectric constant
materials for both laminates to yield good impedance bandwidth and
efficiency. These findings are contradictory to previously derived 'rules of
thumb' and logically lead to the following question: 'what composition of
materials can give the optimum impedance bandwidth and antenna
efficiency?'

In this section, we explore material combinations to obtain the


optimum impedance and efficiency performance for a probe-fed stacked
microstrip patch antenna. To search for these optimal solutions manually
would require many hours of a skilled antenna engineer's time and would

100
likely be colored by the rules of thumb discovered previously. Instead we
employ numerical optimization to automatically synthesize high performance
antenna geometries under specified conditions and within given constraints.
Using this approach, the dielectric material values can be set and optimal
antennas using these materials can be synthesized. By comparing the
performance of the solutions resulting froin different material combinations,
design trends suggesting the best material combinations for efficient
broadband patch antennas may be identified.

3.4-5.2 Optimization Approach

Numerical techniques have been used previously in the investigation


of optimum microstrip patch antennas. Recently an investigation was
conducted on the impact of the conductor shape of a microstrip patch antenna
on the bandwidth of the antenna and how it can be optimized to yield
performance compared to more conventional, canonical patch geometries [3].
Here a genetic algorithm was utilized to search for conductor shapes that
could improve the impedance response. Bandwidths on the order of 20% were
achieved. Conversely, the approach used in this section maintains the same
basic geometry, the direct contact probe fed circular stacked patch antenna,
but varies other parameters of this geometry.

The optimization procedure used here begins with a random set of


antenna parameters. It would be possible to begin with a known solution, but
so as not to bias the solution we begin with purely random parameters. These
are then checked to see if they fall within loose constraints, failure results in a
penalty being applied to the cost function. If the parameters are within
constraints, the antenna response is modeled and the results are assessed using
a cost function. The result of this cost function is compared to the cost
function of the previous iteration to determine if the solutions have
converged. If the normalized change in cost is smaller than a specified
tolerance, then the solution is deemed converged and the optimization is
halted, otherwise, the current parameters and cost are passed to the
optimization routine where a new set of parameters is generated. This new set
of parameters is checked against the constraints and the cycle continues until
the convergence criterion is met. The Simplex method is used as the
optimization routine due to its stability and reasonable rate of convergence.

Antenna model, parameters and constraints

The patch geometry is presented in Figure 3.4.1. This particular


geometry was chosen due to the availability of a fast, accurate simulation tool
based on the work of [11]. There are thus eight variables for optimization R\,
101
R 2, erh er2, 010 d2, ro and rp• As mentioned previously, we wish to examine the
dependence of antenna bandwidth and surface wave efficiency on the choice
of dielectric substrate layers by choosing erl and er2 and recording the
optimum solution as a function of this choice. Hence, six variables remain for
each optimization. Constraints must be set to ensure the synthesized patch
can be realized. A further constraint must be placed on the thicknesses d,
and d2 as common laminates are only available with certain thicknesses. To
account for this we initially allow a continuous choice of d 1 and d2 • On
finding an optimum, these thicknesses are rounded to an available thickness
and then optimization resumes with d, and d2 held at these values . The
optimizer can then adjust the remaining four parameters to compensate for the
thickness rounding. A practical solution near the optimum is thus achieved.

The costfunction

The cost function is a quantified measure of what we actually want


from the optimization, that is, we would like a patch antenna operating at a
specified frequency with as broad as possible impedance bandwidth, but
which maintains acceptable (80 %) surface wave efficiency. A good radiation
pattern can be "assumed due to the canonical geometry of the structure. The
challenge with defining a cost function is how to best translate this request
into a mathematical expression indicating the worth of a solution . Here, the
cost function is calculated by simulating the antenna performance at 10
frequency points equally spaced from 0.75 Ac to 1.25 Ac (where Ac is the
wavelength of the center of the desired band) and then performing weighted
sum of three distinct quantities representing mean return loss, bandwidth and
in-band surface wave.

To optimize the bandwidth, there must first be a matched impedance


near the frequency of operation. A completely random case is likely to result
in an impedance bandwidth of exactly zero due to the absence of any matched
points. Nearby solutions are also likely to have zero bandwidth and thus the
bandwidth alone supplies insufficient information for the optimization routine
to improve the patch. The optimizer must thus try initially to form an
impedance-matched point. Numerically, this is achieved by minimizing a
weighted average of the return loss of all frequency points. The weights also
favor well-matched points closer to the specified operating frequency to
encourage the band to be centered near this frequency.

Maximum impedance bandwidth

Once a matched point is achieved, the bandwidth may be optimized.


At each iteration, an attempt is made to calculate the 10dB return loss
102
bandwidth. If it is zero, the cost is also set to zero otherwise the cost becomes
the reciprocal of the bandwidth value.

Constrained in-band surface wave efficiency

The final component of the cost function attempts to ensure radiation


efficient solutions are found. To do this, the surface wave efficiency is
calculated using the full wave model. Frequency points within the 10dB
return loss band, with surface wave efficiency 80 % incur a significant
penalty.

Component weights

Careful attention must be paid to the weighting of each of the


components of the cost function such that the distinct goals that they represent
are given the desired priority. The surface wave efficiency weight is simply
set to unity as the large penalty will certainly dominate all other cost
components. The bandwidth weighting is made 10 times larger than the
weight for the average return loss so that bandwidth optimization dominates
once a matched resonance is found. In the absence of a matched resonance,
only the return loss cost is non-zero and thus may be set arbitrarily to unity.

3.4-5.3 Optimization Trends

Having developed a suitable automatic algorithm, it is possible to


begin to explore the design trends and identify the optimal material
combinations for broadband, high efficiency probe-fed stacked patch
antennas. Before presenting this investigation, it will be useful to summarize
and validate some previously reported design trends, observations and 'rules
of thumb'.

Firstly, the dielectric constant of the upper layer must be close to


unity to ensure high surface wave efficiency levels, greater than 80%.
Alternatively, if the upper and lower dielectrics are the same and greater than
2.0, impedance bandwidths on the order of 45-50% can be readily achieved;
however, the surface wave efficiencies of these configurations are very poor,
less than 40 % across the entire 10 dB return loss bandwidth. For cases when
the top layer has a higher dielectric constant than the bottom layer, only
narrow impedance bandwidths can be achieved « 10 %) as it is extremely
difficult to obtain a weakly coupled mutual resonance between the patches.
This finding can be attributed to that the excited currents on each radiator are
too strongly coupled. Cases where both dielectric constants are set to
approximately one yield highly efficient antennas as there is no excited
103
surface wave, however, as has been increasing the bandwidth with this
material combination is difficult. Once again this can be attributed to the
difficulty in arranging weakly coupled mutual resonances.

Keeping these rules of thumb' in mind, cases were run for different
Erl values leaving Er2 constant at 1.07, the value for hard foam. Values of Erl
simulated were: 2.2, 5.0, 7.5 and from 10 to 40 in steps of 5. All other
parameters in the probe-fed stacked patch configuration were set as variables.
Figure 3.4.13 presents the maximum impedance bandwidth achieved by the
optimizer for the stacked patch configuration as a function of the dielectric
constant of the lower layer. The open circles represent the optimum
bandwidth achieved when the procedure was constrained to produce solutions
exhibiting a minimum of 80 % in band surface wave efficiency, the filled
circles represent the bandwidths of optimum solutions with this constraint
relaxed.

40

l35
130
Qll
E
:>
.~ 25 -E-E-- ---
8
Bandwidlh (SWE > 80%) 0
20 Min. SWE (SWE > 80%) 0
Ban<lwidlh(relax SWE) V
Min. SWE (relax SWE) 0
15 0
o 5 10 15 20 25 30 35 40
Layer 1 Dielectric constant (l,)

Figure 3.4.13 Optimum Bandwidth and MinimumIn-band SurfaceWave Efficiencyas


Functionof E,.1

Also shown in Figure 3.4.13 is the actual minimum surface wave


efficiency within the 10 dB return loss band exhibited by each of these
optimal bandwidth solutions. Solutions where a minimum in band surface
efficiency 80 % was enforced are represented by open squares as where the
solutions achieved with this condition relaxed are represented by filled
squares. Several important trends are evident.

Good impedance bandwidths (> 25 %) and high efficiency (> 80 %)


can be achieved for stacked patches with Erl < 15. For Erl > 15, the minimum
surface wave efficiency within the 10 dB return loss starts to degrade. In fact,
for these cases it was necessary to relax the 80 % constraint on surface wave
104
efficiency to obtain reasonably broadband solutions at all. Thus for Erl > 20,
although good impedance bandwidth can still be obtained, the minimum
surface wave efficiency is significantly reduced, compromising the radiation
performance of the antenna.

Figure 3.4.14 presents the return loss and surface wave efficiency
frequency response for the optimal case with Erl =15. Note that the bands in
which high efficiency and good impedance bandwidths are achieved overlap,
although the maximum efficiency tends to be near the upper frequency edge
of the 10 dB return loss bandwidth. Similar frequency responses were
calculated for the range of values of Erl but are not shown for brevity. It is
interesting to note, however, that as Erl is reduced, the overlap between the
high efficiency band of frequencies and good impedance matching becomes
better and the efficiency is maintained over the matched impedance
bandwidth declining only slightly with reducing frequency. This implies that
the coupling between the 2 radiators improves the surface wave efficiency.
100

95

90
-5

85
l
~
e
80 s'"
75
'"'">
~
70
8
'"
't:
-15 cil
85

60

55
12 1~ lA 1~ 1~ 1.7 1.8
Frequency (GHz)

Figure 3.4.14 Return Loss Response and Surface Wave Efficiency for Optimized Patch
=
with Erl 15

A simplified explanation of this phenomenon is as follows: the


surface wave generated from the lower the patch and its associated dielectric
layers which is normally trapped with the dielectric layers couples to the
radiating fields associated with the upper, resonant patch. This explanation is
also consistent with the findings in [20], where an oversized resonant ring was
used to couple to the surface wave mode, thereby reducing its effect on the
efficiency of the antenna.

105
The trend in Figure 3.4.13 shows that despite this coupling of the
surface wave mode to the upper patch, as the dielectric constant of the lower
layer is increased the efficiency starts to degrade. This can be explained by
having a closer look at the parameter values used to give the solutions in
Figure 3.4.13. As Erl is increased, good impedance bandwidth and efficiency
occurs when the height d, is decreased. Although intuitively, using well-
established trends for single layer patches (see Chapter 2), the lowering of the
height should to an extent counter the increase of Erl and therefore the overall
contribution to the surface wave may decrease, there is an important effect
that needs to be pointed out. As d, decreases, obviously the height of the
probe decreases and so the effect of the discontinuity associated with the
probe feed becomes less pronounced. It has been shown in the past that
surface waves couple well to discontinuities and that the presence of possible
discontinuities can determine whether the power is trapped within a surface
wave or radiated [21]. By reducing the height of d, the discontinuity
associated with the feed pin decreases and therefore the amount of power
coupled to the surface wave decreases.

From Figure 3.4.13 there appears to be an optimum solution for a


probe-fed stacked patch antenna where the impedance bandwidth and the
minimum surface wave efficiency are both high. This occurs when Erl = 5.
For this case the theoretical 10 dB return loss approaches 40 % and the
minimum surface wave efficiency is approximately 90 %. This return loss
bandwidth is significantly larger than previously reported probe-fed stacked
patch solutions reported in the previous sections, approaching values obtained
for large slot aperture coupled patches configuration (which we will examine
in the next section) and small slot-sized aperture coupled stacked patches
[22]. The probe-fed stacked patch solution has the advantage of reduced
backward directed radiation compared to these aperture solutions.

Due to the unavailability of commercial microwave laminates with a


dielectric constant of 5.0, an optimized probe-fed stacked patch solution was
sought utilizing standard PCB (FR4, e, = 4.5) as the bottom layer. This
material can be somewhat lossy for frequencies above 2 GHz so an
investigation centered at approximately 1.4 GHz was conducted. To ensure a
solution could be realized, the thickness of the bottom layer d l was limited to
multiples of 1.59 mm and the upper foam layer d2 to multiple combinations of
3, 5 or 6 mm thick laminates.

Figure 3.4.15 shows the predicted input impedance response of the


optimized stacked patch (refer to the Figure caption for parameters). The 10
dB return .loss bandwidth is approximately 37 %, approaching that achieved

106
for the optimum results presented in the previous section. Figure 3.4.15 also
presents the impedance locus for the optimized dimensions predicted by
Ensemble 6.1. For this simulation, the circular patch was divided into 36
segments and thus it is possible that discretization errors may account for the
slight discrepancy between the simulated results of Ensemble and the full
wave analysis used during optimization.

Figure 3.4.15 Impedance Response of Optimized Stacked Patch

The proposed antenna was fabricated using standard etching


techniques. A thin layer of RT Duriod 5880 (thickness of 0.254 mm and a
dielectric constant of 2.2) was used to etch the conductor of the top patch
located above the layers of foam in the stacked patch arrangement. To realize
the feeding pin, a wire with a thickness of 6 mm and a height of 12 mm was
soldered to the conductor of the bottom patch and to an N-type connector
located below the ground plane of the antenna. The measured impedance
response of this realized antenna is also presented in Figure 3.4.15. As seen
from this figure, there is reasonably good agreement between measurement
and predicted results. The measured 10 dB return loss is 33 %. The slight
reduction in bandwidth (37 % predicted) and center frequency (actually
centered at 1.323 GHz) can be attributed to the following factors: (1) the thin
layer of material used to etch the top patch is not included in the predicted
results; (2) the uncertainty in the dielectric constants of FR4 and the hard
foam; and (3) thin air-gaps between the constructed layers.

A capacitive shift is also evident in the measured response when


compared to the theoretical predictions, which may contribute to the reduction
in the observed bandwidth. This shift is due to manufacturing difficulties
associated with connecting the electrically thick probe feed to conventionally
107
styled connectors. The diameter of the hole in the ground plane to
accommodate for the pin connector needs to be large enough such that a 50 .Q
transition between the guided medium (a coaxial line) and the bottom of the
patch antenna is maintained. From Figure 3.4.15 it is apparent that the hole in
the ground plane is slightly undersized.

The measured co-polar E- and H-plane radiation patterns of the


optimized stacked patch antenna at 1.4 GHz are shown in Figure 3.4.16. The
radiation patterns are very similar to the computed patterns (not shown here
for the sake for brevity) and are typical for a stacked patch configuration. The
ground plane for the antenna was very small, extending only 2 em beyond the
edge of the top patches. For this reason, there appears to be a slight distortion
in the patterns, particularly for the E-plane case, in the backward directed
plane.
o Eplane-
Hplane-

.. ..

7/'
~ ,

"

./
i

......:: ~-,:. • ,.:~./·.:.':J20


~••:.... I'

-150 •... •.•:: •. ·.: :L~'·'·: ·':"'~'· '· ·:~·;;;·· · · · 5dB/diV


180

Figure 3.4.16 Measured Co-polarE- and H-planeRadiation Patterns at 1.4 GHz

The measured cross-polarized levels are less than -20 dB in the E-


plane and less than -12 dB in the H-plane below the co-polar fields. These
values compare to theoretical levels of less than -40 dB and -14 dB,
respectively. Once again, the finite ground plane contributes to the higher
than predicted levels. It is interesting to note that the predicted and measured
cross-polar levels are not as low for the case when a higher dielectric constant
material is used for the lower layer presented in the previous section. One
source for these comparatively high levels could be the relatively thick (6
mm) feed pin used to obtain the optimally efficient and broadband solution.
As was presented in [23], the probe does contribute to the cross-polarized
field, and the thicker the probe, the higher the cross-polar content. From this
there is another apparent compromise that must be made in the choice of the
108
optimum solution of a stacked patch antenna: using thicker probes can yield
better bandwidths; however, the cross-polarized levels are also increased.

The theoretical and measured gain of the stacked antenna was 9.0 dBi
and 8.5 dBi at 1.4GHz, respectively. The discrepancy can be resolved by
considering the large backlobe in the H-plane and the accuracy of the antenna
range (± 0.5dB). These results verify that efficiency has not been
compromised to attain large bandwidth.

3.4-6 Cavity backed stacked patches

Cavity backed configurations can be incorporated to enhance system


integration, reduce backward directed radiation and increase the antenna
efficiency by suppressing the excitation of surface waves. A cavity backed
stacked patch antenna can have similar RF characteristics as a probe-fed
stacked circular patch antenna, if the wall of the cavity is not too close to the
edge of the patches (a 'rule of thumb': further than 0.2 1..0).

Figure 3.4.17 shows a schematic diagram of a cavity backed circular


stacked patch antenna. A full-wave spectral domain integral equation solution
based on a Greens functions technique was used to analyze the printed
antenna [24]. Each layer of the antenna can be accounted for in the analysis.
In designing this antenna, four layers were used including the layer required
to etch the upper patch and a small air-gap that exits between this dielectric
layer and the top of the cavity (refer to Figure 3.4.17).

Figure 3.4.17 Cavitybacked stacked patch antenna(parameters: £'-. = 2.2, d l = 3.048 rom,
R1 =31.3 mm, xp =28 rom, yp =0, ro= 0.325 rom, fa = 1.07, dz = 10.0 rom, £'-3 = 2.2, d3 = 0.254
rom, Rz =34 rom, R3 =50 rom)

109
The simulated and experimental results of a cavity backed stacked
patch antenna designed for operation at 1.9 GHz are shown in Figure 3.4.18.
The parameters obtained from the simulation for this design are as follows: Erl
=2.2, dl =3.048 rnm, R, =31.3 rnm, xp =28 rnm, YP =0, ro =0.325 rnm, Ea =2.2,
d2 =0.254 rnm, R2 =34 mm, R3 =50 rnm. Foam with a thickness of 10 mm was
used between the lower patch and upper patch. It can be seen from this figure
that the theoretical and experimental impedance behavior are in very good
agreement. The 10 dB return loss bandwidths of 23% was obtained, the
largest ever reported for a cavity backed probe-fed stacked microstrip patch
antenna.

Figure 3.4.18 Input Impedance of Cavity Backed Stacked Patch Antenna

The measured gain of antenna was 7.35 dBi. The far-field radiation
patterns were measured at 1.9 GHz for E-plane and H-plane co-polar and
cross-polar levels. The plots of E-plane and H-plane co-polar are shown in
Figure 3.4.19 and Figure 3.4.20, respectively. The cross-polarization levels
were similar to a stacked circular patch reported in Section 3.4-3. Figure
3.4.21 shows a photograph of the cavity-backed antenna used in a printed
Antenna Remote Unit CARU) for optically distributed mobile communications
compliant with the 1.9 GHz peN specifications [25].

tOO

Figure 3.4.19 Measured Far-Field E-plane Co-polar Radiation Pattern of Stacked Patch
Antenna

110
'10

Figure 3.4.20 Measured Far-FieldE-planeCo-polarRadiation Patternof Stacked Patch


Antenna

Figure 3.4.21 1.9 GHz PCN ARU

3.5 Large Slot Excited Patches

3.5-1 Introduction

A simple and common way of enhancing the bandwidth of an


aperture-coupled patch without increasing the complexity of the antenna by
means of stacking is to increase the size of the slot [26]. By having the slot
close to resonance allows an impedance response similar to that presented in
Figure 3.4.18, as there are now two strongly coupled radiating structures,
namely the slot and the patch. In fact having a relatively large slot is a natural
progression to achieve large bandwidth from an aperture-coupled patch. As
was shown in Chapter 2, to increase the bandwidth of a microstrip patch
antenna, thick dielectric material is required. In an aperture-coupled patch
environment, to ensure power is coupled to the patch located on a thick
dielectric layer, the size of the slot must be increased. Bandwidths in excess
of 40 % have been achieved using this relatively simple technique [26].
Large slot excited patches have all the salient feature of the aperture coupled
111
patches presented in chapter 2, in particular because they have a non-contact
excitation mechanism they do not suffer from the problems associated with
the current discontinuity of either the probe or edge fed patches.

3.5·2 Design Example

Figure 3.5.1 shows a schematic diagram of a large slot excited


rectangular patch antenna. As for the case of a conventional aperture coupled
patch, a microstrip feed line on the bottom substrate is coupled through an
aperture (typically a narrow slot) in the ground plane to the microstrip patch
on the top substrate. To analyze the printed structure shown in Figure 3.5.1,
an analysis based on the spectral domain moment method solution was used
using entire domain basis functions on the patch element and piece-wise
sinusoidal basis functions in the aperture. To take into consideration the
effect of air gaps between the layers as well as a cover layer, a generalized
multilayer spectral domain Green's function [27] was incorporated into the
model.

U " ' I - - - Antenna Substrate (t;.. tan 0•• dJ

Aperture (SL' S.)

#-olll --- - Feed Substrate (E". tan li,. d,)

Figure 3.5.1 Large Slot Aperture Coupled PatchAntenna

Figure 3.5.2 shows the predicted and measured impedance of an


aperture coupled patch designed at 1.9 GHz for the application of a mobile
communication base station (refer to the figure caption for the relevant
dimensions). Very good agreement between experiment and theory was
achieved as is evident from this figure. The predicted and measured
bandwidths are 27 % and 26 % respectively.

112
Figure 3.5.2 Input Impedance of LargeSlot Excited Patch

The co-polar and cross-polar E- and H-plane radiation patterns were


measured and the co-polar patterns are shown in Figure 3.5.3. As is evident
from this figure, scalping of the radiation pattern, particularly in the E-plane
has occurred. This results from using a large slot in conjunction with a small
ground-plane (here the ground-plane extends on 2 em beyond the edge of the
patch). As a reasonable portion of the power radiated from a slot is directed
towards endfire, this power can be easily diffracted off a truncated ground-
plane, resulting in a deformation of the radiation pattern. Possible ways to
alleviate this problem are to ensure the ground-plane extends a relatively large
distance with respect to the center of the patch and slot, in the order of a
couple of wavelengths or so. Also absorbing material can be used to coat the
edges of the ground-plane.

Figure 3.5.3 Far-field Radiation Patterns

113
As can be seen from Figure 3.5.3, the backward directed radiation
level is reasonably high with a front-to-back ratio (FIB) of 12 dB, compared
to a theoretical value of 15 dB. This sort of level can be an issue, in particular
for base stations for wireless systems. Typical base stations/radio hubs
provide a sectoral coverage area to increase the system capacity, for example
for LMDS (Local Multipoint Distribution Services) the 3600 azimuth plane is
usually split into four 900 sectors, whereas for a mobile base station it is
usually three 1200 sectors. Imperative for these sectoral systems is to
minimize the back radiation from each antenna to ensure interference from
adjoining sub cells is also minimized. Because a slot in a ground-plane
radiates equally in both the upper and lower half spaces (ignoring the affect of
the dielectric materials surrounding the ground-plane), the level of backward
radiation can become somewhat high. This leads to less power available in
the power budget of a link and more importantly for sectored wireless
communication systems, it can lead to increased levels of interference. The
predicted and measured gain of aperture coupled patch antenna was 9.0 dBi
and 8.8 dBi, respectively.

3.5-3 Methods to reduce backward directed radiation

3.5-3.i introduction

A commonly used method to improve the radiation characteristics of


a large slot excited patch antenna is to place a shielding plane behind the
antenna. This approach eliminates any unwanted radiation in the back region.
However, the incorporation of a shielding plane allows the propagation of
parallel plate waveguide modes (this is shown graphically in Chapter 7).
Power transferred into these modes can seriously degrade the efficiency of the
antenna. Another common procedure is to enclose the aperture in a cavity; a
cavity backed solution. The use of a cavity eliminates the directly radiated
backward fields (not the diffracted ones) at the expense of greater
manufacturing complexity, thereby increasing cost.

Another way to improve the front-to-back ratio is to place a


microstrip antenna element behind the aperture as a reflector [28], as shown in
Figure 3.5.4. The action of this element is akin to that in a Yagi-Uda dipole
array. However, the incorporation of a reflector in an aperture coupled patch
antenna is much easier to achieve. Proximity coupling between the feedline
and the reflecting element is negligible due to the thick foam substrate used,
allowing use of the reciprocity method of analysis. Also, the directive patch
elements are shielded from the reflector by the ground plane. Therefore, only
interactions between the reflector and the aperture need to be modeled,
114
resulting in a simple analysis. For aperture coupled patch designs with a
front-to-back ratio of lOdB or greater, the introduction of a reflecting element
has a negligible effect on the input impedance of the antenna. Therefore, a
reflector element can readily be incorporated into existing designs.

Figure 3.5.4 Geometry of Aperture Coupled Microstrip Antenna with Reflector Element

3.5-3.2 Cavity backed solutions

In Section 3.5-2 it was shown that the use an aperture coupled


microstrip patch can provide good return loss performance. Although this
configuration yields a broad impedance bandwidth, the front-to-back ratio
was quite poor. A common technique to reduce this effect is to incorporate a
metallic back-plate, however, problems can arise. If the aperture coupled
patch is configured in an array, the mutual coupling between elements can
significantly increase due to the excitation and propagation of a parallel plate
mode between the back plate and the ground plane of the aperture coupled
patch. If this is indeed the case, the input impedance variation of a single
aperture coupled element will differ significantly from that of a patch in the
array environment.

Probably the most robust solution to the aforementioned problem is to


enclose the back plane of the slot in a metallic cavity [29]. A schematic of
this configuration is shown in Figure 3.5.5. By doing so, there will be no
back radiation and thus no mutual coupling via the propagation of a parallel
plate mode. Another advantage of such a configuration over a traditional
aperture coupled patch arrangement is the reduction in weight as no large
backing plate is required, if an array of elements is to be assembled.

115
patches

metallic cavity slot

Figure 3.5.5 Cavity Backed Aperture Coupled Patch

A cavity backed aperture coupled microstrip patch was designed for


1.9 GHz with an impedance bandwidth of 20 %. The analysis is similar to
that mentioned in the previous section with the spectral domain representation
of the Green's function for the magnetic field produced by the magnetic
current modified to account for the presence of a cavity. A metal cavity of
size 10 x 10 x 2 em' was used. The predicted and measured input impedance
results are presented in Figure 3.5.6. The dimensions of this antenna are the
same as those even in Figure 3.5.2 with only a slight reduction in the length of
the open circuit microstrip line to move the resonant loop towards the center
of the Smith chart. The reason for requiring only a slight modification to the
design is due to the reasonably large sized metallic cavity used. The predicted
and measured gain of the cavity backed aperture coupled antenna was 9.3 dBi
and 9.0 dBi, respectively. The far-field co-polar E- and H-plane radiation
patterns are shown in Figure 3.5.7. The measured FBR is 20dB. Another
example of a cavity backed slot coupled patch antenna will be presented in
Section 3.6.

"
o

·10
..;r
'l\~~/I
~20 I
, ~
.;
::l
,
I

.3-30
~
E
::> ~! I-4-Theory
0-40 I -
a: ~ 1-'-'Expertment
(
· 50
I

-60
1.2 1.4 U U 2 2.2 2.4
Frequency (GHz)

Figure 3.5.6 Return Loss Response of Cavity Backed ACP

116
''''
Figure 3.5.7 Measured Radiation Patternsof CavityBackedACP

3.5-3.3 Reflector patch solutions

Theory and Design

The analysis of the an aperture coupled patch incorporating a reflector


patch structure (see Figure 3.5.4) was performed using a spectral domain
moment method model incorporating a multilayer Green's function [27]. Use
of this Green's function allows for generality and freedom in the design. The
radiated far fields were computed from the stationary phase evaluation of the
Green's function for the field components of the elements.

The design process involves examining the relative magnitude and


phase of the radiated far fields due to the aperture and reflector. A typical
plot of these quantities at backfire (8 = 180°) versus reflector length is shown
in Figure 3.5.8.. The resonance of the reflector is readily seen at a length of
11.6 mm, with its radiated field being much stronger than that of the aperture.
However, as with a Yagi-Uda array, the reflector is operated above resonance
to achieve the desired effect. At a length of 16mm the fields are of
approximately equal magnitude and opposite phase, resulting in a cancellation
of the total field and a marked improvement in front-to-hack ratio. An
interesting observation from Figure 3.5.8 is that as the element length is
increased past resonance, the relative magnitude and phase of the far fields
remain fairly constant. This implies that the cancellation effect is fairly
wideband, and that the reflector can be used with wideband aperture coupled
patch elements.

The design is also quite straightforward in that the relative phase is


controlled primarily by the resonant length, while the relative magnitude of
117
the fields can be controlled by adjusting the reflector spacing and width.
Increasing the reflector spacing will decrease the magnitude of its radiated far
field. The same effect can be achieved by decreasing the reflector width,
thereby providing an extra degree of freedom in the design.
2.5 ...----r--,---,--""T""-.---,.--,---, 360
I I I I
I I I ,
2 _ .:.-=l=-- ~~ ~ ___ I -- Magnitude
320
I I ' I , ---...- ..- Phase
I I "' ; I "C
CII " ) . I , I t
"C 1.5 - - -,--- -,-- --1 -- . . - --+- - -+ -- -1- -- 280 ;-
E :: I : I : :
III
CD
6,
---i- ---i--- :- -\j---:---:~::..~-~~ _
I I " t I CL
240
~ 1
CD
~
-;i7--~ - --~ ~.- ~ ---~ ---~ --
I I , , I I I

0.5 - - - :- - 200
I "",/ I I I .....--.. ..... I I
.....)........ I I I I "'t-.__ ..1._ .._.
o w..J- 160
4 6 8 10 12 14 16 18 20
Reflector Length (mm)

Figure 3.5.8 Relative magnitude and phase of reflector and aperture far fields:
Aperture length =12.5 mm, aperture width = 1 mm, reflector width = 1 mm, reflector spacing =
5.635mm; fr = 9050Hz

Design Example

In order to validate the theory, a case consisting of two experimental


designs was fabricated and tested. Directive patches were left out of this
example to show more clearly the effect of the reflector on the radiation
pattern. The first radiator was a microstripline fed aperture antenna etched on
0.025" thick Rogers Duriod 5880 with a design frequency of 9.5 GHz. The
second was exactly the same as the first, except that a microstrip dipole was
incorporated as a reflector element. It is interesting that the addition of the
reflector provided only a reactive shift to the input impedance of the antenna.
This reactive shift was easily tuned out by adjusting the microstripline stub
length.

Computed and measured radiation patterns for the design


incorporating a reflector are shown in Figure 3.5.9. The reflector was
designed to provide a 20dB decrease in the radiation level at backfire. It can
be seen that the radiation level increases away from backfire, particularly in
the E-plane. This is due to the relative magnitude of the slot and reflector
radiation patterns, which differ quite greatly near ±90 degrees. However, a
distinct change is seen from the bidirectional aperture pattern, with a
corresponding increase in front-to-back ratio. Defining front-to-back ratio as
the ratio of total radiated power in the two half-spaces, an increase of 7.4 dB
118
was computed with the addition of the reflector element. This in tum resulted
in a 2.2 dB increase in directivity. A similar improvement in front-to-back
ratio can be achieved by using a reflector with aperture coupled patch
antennas.

180 180

(a) (b)
Figure 3.5.9 Radiation patterns of microstripline fed aperture antenna with reflector
element at 9.5 GHz: (a) E-plane; (b) H-plane

Another interesting result of the freedom and simplicity inherent in


the design process is that the reflector can be designed for maximum
reduction in the radiation level at any angle in the back half-space. This can
be achieved by simply adjusting its length, width, and spacing from the
aperture. This has implications for use in arrays scanned off broadside, where
the maximum reflection can be designed for the desired scan angle. Another
design example and performance evaluation of using the reflector patch for a
slot excited printed antenna will be provided in Section 3.5-5.

3.5-3.4 Dual polarization methods

Introduction

There are many applications that require dual polarization antennas.


A high priority is typically placed on the degree of symmetry that can be
obtained within the antenna. Such symmetry is extremely important in
achieving a high degree of isolation between the two polarizations; in theory
total isolation over all frequencies can be achieved if perfect symmetry is
maintained in the antenna, and in practice isolation as high as 35 dB can be
obtained. However, any asymmetry inherent in the design will limit the level
of isolation that can be achieved.

119
In this section three configurations of microstrip antennas based on
the aperture coupling feeding technique will be discussed. Each configuration
is capable of dual linear polarization, and is also capable of wideband
operation. Advantages and disadvantages of implementing each configuration
will be presented, with the main focus being on the potential of each
configuration to meet all of the performance specifications. Attention will
also be paid to ease of design, ease of fabrication, and overall material cost.

Dual OffsetApertures

By utilizing the fact that the aperture may be offset in the direction
perpendicular to resonance without suffering a drop in coupling, dual offset
apertures can be included in the design of an aperture coupled patch in order
to achieve dual linear polarization [30J. The geometry of this configuration
is shown in Figure 3.5.10. Here we have two perpendicular apertures feeding
a square patch, with each aperture feeding a separate polarization. This
geometry is attractive in that each aperture is coupled to a single feedline,
minimizing the design complexity of the feeding network. However, there are
two major drawbacks in implementing this geometry for this application. The
first drawback is the limitation of aperture size that is placed on the design -
by referring to the geometry in Figure 3.5.10 it is obvious that the size of each
aperture is restricted to less than half the patch length. This limitation on
aperture size forces the aperture to operate well below resonance, rendering it
extremely difficult to achieve broad bandwidth.

1/ Apertures
Pnrt 1 Coupling
"
II
U
= jf,

/'
Mlcrostrlp Patch

Port 2

Figure 3.5.10 Geometry of dual linear polarized patch antenna using offset apertures

120
The second, and most critical, drawback of this approach is the
inherent asymmetry introduced into the antenna. As mentioned previously,
this asymmetry decreases the amount of isolation that can be achieved
between the two polarizations. Practical limits on polarization isolation using
this technique are in the range of 20 to 25 dB. Also, the degradation of
polarization isolation is especially apparent when using thick substrates to
increase bandwidth [31], causing this option to be especially unattractive.

CrossedSlot with a Balanced Feed

A higher degree of symmetry can be maintained by using a centered


crossed slot to feed the patch element. Also, the use of a crossed slot removes
the length restriction that was placed on the apertures in the previous
configuration. Unfortunately, the introduction of the crossed slot facilitates
the need for a balanced feed network, as using single feedlines on each arm of
the crossed slot introduces asymmetry and greatly decreases polarization
isolation to ievels less than 10 dB. The geometry of an aperture coupled
patch using a crossed slot for dual polarization is shown in Figure 3.5.11.
Each arm of the crossed slot is fed with a balanced network, effectively
decoupling the two polarizations. The disadvantage in this network, however,
lies in the fact that an air bridge is needed as the microstrip lines cross. The
incorporation of the air bridge introduces asymmetry into the feed network,
which in tum can produce coupling between the two orthogonal arms of the
crossed slot and limit the amount of polarization isolation that can be
achieved. The air bridge also increases the difficulty in fabrication of the
antenna, therebv>y imcreasmz
. costs.

-,
Mlcrostrlp Patch

II Pnrt 1

,,;:¥
Crossed Slot,
V I)
In Ground Plane "-
""-- ,--/ Air Bridge

Port 2

Figure 3.5.11 Geometry Crossed Slot ACP with Balanced Feed Network
121
Dual Layer Crossed Slot with a Balanced Feed

The only drawback of the crossed slot configuration presented in


the previous section is the need for construction of an air bridge in the feed
network. This can be circumnavigated by implementing the feed network in a
dual layer configuration [32] as depicted in Figure 3.5.12. Here the feed
network for each polarization is placed on a separate layer on opposite sides
of the ground plane, eliminating the need for an air bridge. This configuration
takes advantage of the fact that if the same substrates are used for each feed
network, the coupling to the aperture will be the same for each. The input
impedance for the two ports will differ slightly, however, since the section of
feed network above the ground plane can electromagnetically couple to the
patch, akin to the mechanism of the proximity coupled antenna. In this
application, where a thick foam substrate will be used for the patch, this
proximity coupling will have a minimal effect on the input impedance as the
coupling of the microstrip feedlines to the aperture will be much stronger.

I~

Port 1

Lower Feed Network Upper Feed Network


PlItch
llIelectrlc Suoerst rate
_ _ c
. :::
L ==
====:6==
Ground Plane
wtlh Crossed -+ E========~y""'~ Uooet' Fc>edNetwor1<
Slot

Figure 3.5.12 Geometry of CrossedSlot Fed Patch Antenna with Dual Layer Balanced
Feed Network

By using this type of configuration, the maximum amount of


symmetry is preserved in the design, thereby allowing for the highest possible
isolation between the two polarizations. The two feed networks are also
separated by the ground plane, thereby eliminating any cross coupling that
could occur between the microstriplines in the two networks. Published
122
results have achieved better than 35 dB isolation over a 20 percent bandwidth
[32]. An example of this method for achieving dual-polarization from a slot
coupled patch will be given in the next section.

3.6 Aperture Stacked Patches

3.6-1 Introduction

Despite the promising bandwidth enhancements shown so far in this


chapter, the primary barrier to implementing microstrip patch antennas in
many more applications is still their limited bandwidth. In this section we
present a microstrip antenna that utilizes a resonant aperture with stacked
patches, and the result is the near doubling of the achievable bandwidth
reported in the previous sections. The geometry of this ASP (aperture-stacked
patch) microstrip antenna is shown in Figure 3.6.1, and differs from the
typical aperture coupled stacked patch antenna in that a larger aperture and
thicker substrates are used. Also, the geometry allows for a general
multilayered configuration having N dielectric layers, with the lower patch
being placed directly above layer N] and the upper patch directly above layer
N z• The term "aperture coupled" could be used to describe this antenna,
however, it is avoided in this text because the aperture is used primarily as a
resonator and not as a mechanism for coupling the patches to the microstrip
feedline.

Ground Plano

~:" ;.:'lII~""':':~~ Aperlure(SL,SW)

Figure 3.6.1 Geometry of Aperture Stacked Patch

123
From this enhanced bandwidth configuration, bandwidths (defined as
having a VSWR < 2:1) from 50 to 70% have been realized. This is the widest
instantaneous bandwidth yet achieved for a microstrip patch antenna. Section
3.6-2 provides a comparison of different methods for bandwidth
enhancement. Techniques for impedance matching are presented in Section
3.6-3. The impedance behavior is very sensitive to several physical
parameters of the antenna, and the effect of varying these parameters is
analyzed in Section 3.6-4. Variations on the design, such as incorporation of
additional patches, are discussed in Section 3.6-5. Details of an experimental
design are presented in Section 3.6-6. An example of an ASP designed at
millimeter-wave frequencies, where material selection is somewhat limited is
given in Section 3.6-7. Section 3.6-8 investigates techniques to reduce the
backward directed radiation of an ASP microstrip antenna, namely using
cavity backing and also a reflector patch.

3.6-2 Comparison of wideband aperture coupled microstrip antennas

Stacking patches and exciting a patch via a large slot bandwidth


enhancement techniques share a common trait: the wideband characteristics
are the result of coupled resonances. The term "mutual resonance" is used
here to describe the action of these coupled resonances, which in general
produces a loop in the impedance locus when plotted on a Smith chart. The
operation of each configuration is essentially the same - a resonance which is
overcoupled to the microstrip feedline is combined with a low-Q resonance,
and the results are impedance loci such as those in Figure 3.6.2. In this
section theoretical results from examples utilizing these techniques are
compared to those of the ASP antenna; in all three examples the antenna
substrates are comprised entirely of foam (e, = 1.07). The theoretical analysis
is based on the spectral domain integral equation technique, with boundary
conditions enforced using a Galerkin moment method (refer to [27, 33] for
more details).

Figure 3.6.2a shows the impedance locus of a wideband microstrip


patch using a thick foam substrate and near-resonant aperture. In this case the
overcoupled resonance is that of the aperture, and the low-Q resonance is that
of the patch. The coupled action of these two resonances produces an
impedance locus that exhibits a tight loop around the center of the Smith
chart, resulting in a 1.5:1 VSWR bandwidth of 21% for this example. The
advantage of this technique is simplicity of design since only a single patch is
used, however, the drawback is the level of back radiation that results from
using a large aperture. The front-to-back ratio was better than 12 dB over the
band.

124
At the expense of increased design complexity, back radiation levels
may be reduced through the use of stacked patches. The impedance locus for
a typical aperture coupled stacked patch design is shown in Figure 3.6.2b.
Here the overcoupled resonance of the lower patch is combined with a low-Q
mutual resonance between the patches, and an impedance locus similar to that
of Figure 3.6.2a results. The antenna substrates consist of foam with a total
thickness equal to that of the previous example. A 1.5:1 VSWR bandwidth of
20% is achieved, and the front-to-back ratio is improved to better than 18 dB
over the band.

(a) (b)

(c)
Figure 3.6.2 Impedance Loci of Wideband Aperture Coupled Microstrip Antennas:
(a) Broadband Patch using Thick Substrate and Near-Resonant Aperture;
(b) Aperture Coupled Stacked Patch; (c) ASP antenna

The use of a resonant aperture with stacked patches results in the ASP
antenna, with a typical impedance locus shown in Figure 3.6.2c. This locus
differs from those of Figures 2a and 2b, with two tight loops near the center of
the Smith chart. The locus is produced by the interaction of the individual

125
resonances - a low-Q mutual resonance produced by the interaction of the two
patches, and a mutual resonance produced by the interaction of the aperture
with the lower patch. The latter resonance is overcoupled in the sense that the
interaction between the aperture and patch is strong. In the absence of the
upper patch, the impedance locus resulting from this overcoupled resonance is
similar to Figure 3.6.2a, however exhibiting a larger loop that typically cannot
be matched to a VSWR<2:1. The mutual resonance between the aperture and
lower patch is associated with the lower frequency loop in the impedance
locus and the mutual resonance of the two patches produces the upper
frequency loop. These mutual resonances can also couple to each other,
changing the overall .behavior of the impedance locus. The addition of
another mutual resonance results in a dramatic increase in bandwidth
compared to the previous two configurations, with a 1.5:1 VSWR bandwidth
of 44% being achieved. The required antenna substrate thickness and
aperture size are also increased and this in tum produces higher back radiation
levels, with a front-to-back ratio of better than 10 dB over the band for this
example.

It was previously speculated that this type of wideband performance


could not be achieved on foam substrates [15]. Although it was achieved in
this example, it is generally much more difficult to obtain than if using
substrates with higher permittivity. The use of only foam substrates provides
a significant improvement in surface wave efficiency. However, there exists a
tradeoff of reduced bandwidth - contrary to what is usually observed in a
single layer geometry.

3.6-3 Impedance matching techniques

In the aperture coupled microstrip antenna the most common method


of controlling the coupling to the microstrip feedline is to vary the size of the
aperture. However, due to the fact that the aperture is used as a radiator in the
ASP microstrip antenna its size cannot be varied independently, and the
coupling to the feedline must be controlled in another manner. Two
techniques to achieve this are shown in Figure 3.6.3.

Use of a radiating aperture results in a high level of coupling which


must be reduced to properly impedance match the antenna. The wide
centered feedline of Figure 3.6.3a achieves this through the fact that the
coupling decreases with feedline width. This technique is advantageous in its
simplicity, however it generally requires the modeling of currents on the
feedline, resulting in a complex analysis. As an alternative, dual offset
feedlines, shown in Figure 3.6.3b, can be used. In this configuration, the
impedance delivered to each feedline at the aperture is nominally lOOQ. The
126
feedlines (Zo=100Q) are joined by a reactive power combiner. This
configuration does not sufferfrom the increased cross polarization introduced
by a single offset feedline, and is easily analyzed with a slight modification to
the reciprocity method of [33]. Of course, if the area occupied by the feed
network is not an issue, a multisection matching transformer can be
incorporated in the design.

low Impedance
line 2d « 1000 line
~ o~ / '

../'
50nllne
50nllne ~

(a) (b)
Figure 3.6.3 Impedance Matching Techniques for ASP antenna : (a) wide centered
feedline; (b) dual offset feedlines

3.6-4 Parameter study

The complicated operation of the mutual resonances in the ASP


antenna introduces a great deal of complexity in the design process. Not only
must the coupling between the individual resonators be weak enough to
produce tight loops in the impedance locus, but these loops must also be close
together in the locus so that they can be matched to an acceptable return loss
specification. The best way to gain insight in how to achieve this is through
varying several critical physical parameters of the antenna, and to note the
overall effect on the impedance locus. In this section we will examine an
ASP antenna with two substrate layers and square patches, and look at the
effect of varying the size of each patch, the aperture length, and the thickness
of each substrate layer. Attention will be focused on the size of the loops and
their relative position to each other in the impedance locus, rather than on
impedance matching. Impedance matching can be accomplished by the
techniques given in Section 3.6.3, and is independent of the parameters under
observation.

The dimensions of the prototype design are given in Figure 3.6.4, and
in each of the examples to follow, all parameters except for one are held
constant to the values given. To have an extremely rigorous theoretical model,
the currents on the feedline should be modeled for this wide feedline case.
127
However, the model chosen to generate the data was based on the reciprocity
method, as it is acceptable to show the trends as the individual parameters are
varied while being computationally efficient.

Figure 3.6.5 shows the effect of varying the aperture length on the
input impedance locus. It is seen that this has a pronounced effect on the
lower frequency loop while having a minor effect on the other loop. This
observation supports the previous statement that one loop is due to the
interaction of the aperture and lower patch, while the second is due mainly to
the interaction of the two patches. Lengthening the aperture has the effect of
increasing the coupling between the aperture and lower patch (therefore
increasing the size of the lower frequency loop), while at the same time
bringing the two loops closer together.

Figure 3.6.4 InputImpedance of Prototype ASPdesign: WjC4.75mm. L.tub=3.4mm,


SL=lOmm, SW=O.8mm; Patches: PLt=PWt=9.lmm. Nt=l, P~=PW2=10mm, N2=2;
Dielectric layers: N =3, £rt=2.33, tan 0r-0.0012, dr-1.6IJUll,£rt=2.2, tan 01=0.0009.
dt=3.175mm. £1'2=1.07, tan 02=0.0009, d2=3mm, £'3=2.2, tan 03=0.0009, d3=O.l27mm

The effect of altering the size of the lower patch is shown in Figure
3.6.6. This is a critical parameter as the action of the lower patch contributes
to both mutual resonances, and therefore varying its dimensions has an effect
on both loops in the impedance locus. It is very interesting to compare the
plots of Figure 3.6.6 with those of Figure 3.6.7, which shows the effect of
varying the upper patch dimension. Note the similarities between Figures
3.6.6a and 3.6.7c, and Figures 3.6.6c and 3.6.7a. These similarities show that

128
it is not the absolute dimensions of each patch that govern the impedance
behavior, but the relative size of each patch to the other that is important.

(a) (b)

(c)
Figure 3.6.5 Input Impedance as Function of Aperture Length: (a) SL=8mm. (b)
SL=10mm; (c) SL=12mm

When the two patches are overcoupled (the situation evident in


Figures 3.6.6a and 3.6.7c), this mutual resonance dominates the impedance
behavior. Conversely, when the resonances of the patches are undercoupled
(Figures 3.6.6c and 3.6.7a), it is the mutual resonance between the lower
patch and aperture that dominates. Achieving the proper balance between the
two mutual resonances maximizes the bandwidth, and this situation is seen in
Figures 3.6.5b, 3.6.6b, and 3.6.7b.

In Figure 3.6.8 the lower substrate thickness is varied. The main


effect of this parameter is on the coupling between the aperture and lower
patch. With increasing thickness the coupling is reduced, and thus the size of

129
the corresponding loop in the impedance locus is smaller. The same effect
occurs when using a small aperture, and is evident by the similarities in
Figures 3.6.5a and 3.6.8c. In terms of design criteria, this signifies that when
aperture size is increased, the lower substrate thickness must also be increased
to maintain the same impedance behavior.

(a) (b)

(c)

Figure 3.6.6 InputImpedance as Function of LowerPatchSize: (a).PL1= PW1=8mm; (b)


PL1= PW1=9mm; (c) PL.= PW.=lOmm

Figure 3.6.9 shows the effect of upper substrate thickness on the


impedance behavior. This affects the size of the upper frequency loop, as the
coupling between the two patches is reduced with increased substrate
thickness. This parameter also has an effect on the interaction between the
two mutual resonances. At a thickness of 2mm (Figure 3.6.9a), the two
mutual resonances are essentially decoupled, however at 4mm (Figure 3.6.9c),
a high degree of coupling exists between the mutual resonances, decreasing
the size of the lower frequency loop.

130
(a) (b)

(c)

Figure 3.6.7 Input impedance as a function of upper patch size:. (a) P~= PW2=9mm;
(b) P~= PW2=1 Omm; (c) P~= PW2=12mm

(a) (b)

131
(c)
Figure 3.6.8 Input Impedance as Function of Layer #1 Dielectric Thickness: (a)
d t=2.5mm; (b) d 1=3mm; (a) d J=4mm

(a) (b)

(c)

Figure 3.6.9 Input Impedance as Function of Layer#2 Dielectric Thickness: (a)


d2=2mm; (b) d2=3mm; (c) d2=4mm

132
3.6-5 Discussion

It was stated in Section 3.6-2 that the lower frequency loop in the
impedance locus was associated with the mutual resonance of the aperture and
lower patch while the upper frequency loop was due mainly to the interaction
of the two patches. While this statement was supported by the data presented
in Section 3.6-4 for the configuration under examination, it also raises two
important questions. First, is this true for any ASP configuration? And
second, can the incorporation of more than two stacked patches in the
multilayer geometry of the ASP produce a third mutual resonance and
enhance the bandwidth even further?

To try to provide some answers for these questions, the currents


generated on the patches are examined in detail. Although five entire domain
modes were used in the current expansion for each patch, the first mode
dominates the impedance and radiation characteristics. Of most importance is
the ratio of the dominant mode currents on each patch, and this quantity is
shown in Figure 3.6.10. The magnitude reaches a peak at approximately 9.9
GHz. This quite interestingly resembles the current behavior of a single
microstrip patch near resonance, hence the use of the term "mutual
resonance" to describe the interaction of the two patches. Of more interest is
the phase difference of the currents on the two patches, shown in Figure
3.6.lOb. Comparing this behavior with the impedance locus of Figure 3.6.4, it
is seen that near the top end of the frequency band the phase difference
between the currents approaches 180 degrees. As this happens the antenna
becomes a very poor radiator and also represents an upper limit to the
bandwidth that can be obtained with the ASP, or a stacked patch configuration
in general. Therefore, to be an efficient radiator over the entire band, the
mutual resonance between the aperture and lower patch must be lower in
frequency than the stacked patch resonance. This also suggests that additional
patches, while possibly providing a loading effect that can assist in impedance
matching, will not enhance the bandwidth characteristics of the ASP as
additional mutual resonances between the patches will exhibit the same
behavior.

3.6-6 Design Considerations and Results

Using the data provided in Section 3.6-4 as a guide, an ASP


antenna was designed, fabricated, and tested. Referring to the geometry
shown in Figure 3.6.1, the dimensions of the antenna were:

133
Feedparameters - Wr= 0.5mm, doff = 5.4mm, LSlUb= 5.8mm, Err = 2.2, dr =
0.635mm, tan or= 0.0009.

Aperture - SL=16mm, SW=lmm.

Antenna substrate - N=4,Erl=1.07, d 1=1.2mm, tan 01=0.OOOQ;Er2=2.2,


d2=3.175mm, tan 02=0.0009; Er3=1.07, d3=3.1mm, tan
03=0.0009; Er4=2.53, d4=0.508mm, tan 04=0.003.

Patch elements - PL1=1O.8mm, W1=20mm, N 1=2; P~=11.2mm,


PW2=20mm, N2=3.

Linearly polarized rectangular patches were used for additional


bandwidth. For dual-linear or circular polarization, square patches and a
crossed slot could also be used in a similar configuration with only a 5-10
percent decrease in bandwidth.

'.,....,---r--r---,--r--.-....--r--,-, 360 ~r--r---,.--r-r--r--,-......-r--,


~- I- - 1- -1--4- -1- -1-" -
340
0.9 L_L _ _ 1__1_ _ ' _ .1 _ .1 _ .1 _
320 I I I I I I I I
0.11
..- 300 -r-r-r,- -, - , - , -, -
_ 0.7 oF 2110 -~-~-I- -I- ~ ~ -~-~-~-

j; 0.11 .£ 260 _L_'-_I_ _ I__I_ _ .1 _ .l _ .l _

- 0.5 E' 240


l'lI 220
-r-r-r,-,-,
I I I I

- ~ - ~ - ~ ~ - ~ -~-
I
'-'-1-
I J

--1--1-
I

0.4
200 _L _'-_L~_J_.1_.l _ _ .I_
180 - - - - - - --- - - - - - -
G.3 I I I I J I I I

0.2 IiJ.ul.w.LLL.l..U!.t.w.lu.ul.u.uL.u..L!.u.uJ.u.u.llwJ 160 Iir..uJI.w.Ll.lJ..lJ.l.WLLu.u.Lr..wL.w.!.lJwlu.u.L.J.liJ


4 5 8 7 II 0 10 11 12 13 14 4 5 6 7 II 9 10 11 12 13 14
Frequency(GHz) Frequency(GHz)

(~ ~)
Figure 3.6.10 Ratio of DominantModePatch Currents, 12111: (a) Magnitude; (b) Phase

The antenna substrate consisted of four layers of varying thickness.


Each layer serves an individual purpose. Previous experience has shown that
when using a dielectric slab for the bottom layer, even small air gaps between
the slab and ground plane can significantly affect the input impedance. The
bottom layer in this design, a 1.2mm thick foam substrate, serves as a buffer
between the ground plane and the thick RT Duroid 5880 slab, minimizing the
effect of any unintentional air gaps present in the experimental model. The
combination of the next two layers, a thick RT Duroid 5880 layer followed by
a thick foam layer, provides good bandwidth characteristics. Because foam is
used for the third layer, a fourth dielectric layer was needed to facilitate
etching of the upper patch.

Impedance matching was achieved by the use of dual offset feedJines,


as discussed in Section 3.6.3. The computed and measured return loss for the
134
antenna is shown in Figure 3.6.11. Good agreement is achieved, with a
computed bandwidth (VSWR<2:1) from 5.05-10.1 GHz (67%) and a
measured bandwidth from 5.07-10.38 GHz (69%). Measured and computed
gain for the antenna is shown in Figure 3.6.12. Again good agreement is
achieved with a computed bandwidth (Gain> 6 dBi) from 5-10.5 GHz (71%)
and a measured bandwidth covering the octave from 5.2-10.4 GHz (67%).
The gain drops off quite rapidly near the band edges; at the lower band edge
this is primarily due to the sharp increase in return loss. However, at the
upper band edge the drop in gain is due to an increased level of back
radiation, as the patches do not radiate efficiently.
O~""""J;~-,--..,----,---.-..,.--r----.-...,
I I I I I I I
__ _ I L _ _ J _ _ L __ __
·5
~ ~

I I I I I I 1
V - '
iii ·10
I I I
- -r - - r - - , - - , - - ~
I
- r--
I 1 1 "'- , I I I
~ ·15 -- ~ -- ~ I - - I - 1 -7i1 ---1-- -~--

--t---i- --i---~ - Vt i--==~:=:u~e~


III
III
.3 ·20
c
~ -25 - - +--~- - I- - - ~ -- - ~ -- -~ --~--
I I I I I I I
li
a: ·30
I
- - "'I - -
I I
-I - - - ,- - -
I
I - -
I I
I - - -, - - -,- - -
I
I - -
I I f I I I I I
~5 - - ~ -- ~ -- - I- - - ~ - - ~ - ~-- - ~ -- ~ --
I 1 I I I I I I
-40 ~ ~ '_'_'_ ~ ~.w

3 4 5 6 7 6 9 10 11 12
Frequency (GHz)

Figure 3.6.11 Computed and MeasuredReturnLoss of ASP Antenna

iii
~
c
ii
CJ

o ""'-'-'u-Wu-Wu..J..w...o...41I

3 4 5 6 7 6 9 10 11 12
Frequency (GHz)

Figure 3.6.12 Computed and MeasuredGain of ASP Antenna

Two other important factors in assessing the overall performance of


the antenna are front to back ratio and surface wave efficiency. The ASP
exhibits a lower front to back ratio than a typical aperture coupled patch
antenna, due to the use of a resonant aperture. The computed front to back
135
ratio ranged from 8 to 14 dB over most of the band, and dropped to 6 dB at
the upper band edge. An improvement on these values can be achieved by
placing a ground plane behind the aperture, or by using a third microstrip
element behind the aperture as a reflector (this will be addressed in the next
section).

In some ASP configurations, a sizable amount of power can be lost to


surface waves due to the thick substrates that are used. For this configuration,
surface wave efficiency ranged from 82 to 90 percent over the band. By
etching the patches on thin dielectric material and using thick foam substrates,
the surface wave efficiency can be improved to over 95 percent. This type of
configuration exhibits reduced bandwidth, but may be more attractive for
phased array applications.

3.6-7 Another design example

3.6-7.1 Introduction

There are a variety of wireless communication applications aiming to


utilize millimeter-wave (mm-wave) frequencies for their radio links. Point-
to-Multipoint (PMP) systems such as Local Multipoint Distribution Service
(LMDS) [34] and broadband wireless access networks (with data rates> 1
Gb/s) [35] have been proposed to provide simultaneous services to a
substantial number of customers via a single centralized mm-wave radio
architecture, or hub. Other communication links requiring mm-wave
technologies are classified as Point-to-Point (PTP) systems. One example of
this type of system is the present day Personal Communication System (PCS)
network, where mm-wave radio links are used between the Base Station
Controllers (BSC) and Base Transceiver Stations (BTS). Associated with all
these applications is the need for unobtrusive, low cost antennas that can be
easily integrated with mm-wave circuitry and also photonic technologies in
future mm-wave radio communication systems employing optical fiber
distribution networks [36 - 40].

Although the mm-wave radio applications outlined here do not


require a large frequency spectrum, using conventional microstrip patch
design techniques would require different designs for each system application.
This in turn would lead to large costs associated with the design phases of
each radio network. The development of a mm-wave printed antenna that is
fundamentally broadband would enable the various radio service applications
to share some equipment hardware thereby reducing the number of hub sites
required to deliver these services and the overall cost of the radio network. In
this section we present a millimeter-wave microstrip patch antenna that can
136
efficiently operate over the entire Ka-band. The antenna is an ASP microstrip
antenna.

3.6-7.2 Millimeter-wave ASP

The mm-wave version of the ASP is the same schematically as that


shown in Figure 3.6.1. As is evident from Figure 3.6.1, there are multiple
layers of dielectric material on the antenna side of the ground-plane. These
consist of the laminates required for the stacked patch geometry as well as
small air gaps that are present as a result of the fabrication procedure to
implement the antenna. The effect of these layers are more pronounced at
mm-wave frequencies than in the lower microwave spectrum and therefore
must be considered when designing an ASP at these frequencies.

To design a Ka-band ASP, the procedure outlined in Section 3.6.4


was followed. To give good surface wave efficiency as well as a broadband
impedance response, a combination of s, = 2.2 (RT/Duroid 5880) and e, =
1.07 (Rohacell foam) was used. It should be pointed out that the choice of
commercially available thicknesses of these dielectric materials is somewhat
limited for the Ka-band frequency range. Fortunately the ASP has many
degrees of freedom and these limitations can be alleviated somewhat by
changing other dimensions of the printed antenna. To operate over the entire
Ka-band, a bandwidth of 41 % is required. As was shown in the previous
section, ASP antennas can have bandwidths in excess of an octave therefore
obtaining the required bandwidth of 41 % is not too onerous. In fact, the
smaller bandwidth allows for the relative dimension of the slot to be smaller
than that of the patches, as compared to the previous solutions, thereby
yielding an improved front-to-back ratio as will be discussed later.

The predicted and measured return loss of the Ka-band ASP are
shown in Figure 3.6.13. The measurements were carried out using a 40 GHz
Wiltron 360 B vector network analyzer. As can be seen from this figure,
theory and experiment are in relatively good agreement. The dimensions for
the antenna are given in the figure caption. The discrepancies in the results
can be attributed to uncertainties in the dielectric constants of the materials
used (± 0.02), alignment errors of the layers and the uncertainty associated
with the presence of the small air gaps. Despite these, a 10 dB return loss
bandwidth covering the Ka-band was achieved for the fabricated mm-wave
antenna. The broadband nature of these printed antennas ensures that the
effects of design and fabrication uncertainties are minimal. The tightened
loops in the experimental impedance loci and hence the lower return loss over
a smaller bandwidth in Figure 3.6.13, are most likely due to an

137
underestimation in the thicknesses of the air gaps, which were assumed to be
0.1 mm. It should be noted that no modifications/reiterations in the design of
the ASP were undertaken as the specification was met in the first fabricated
antenna.

Figure 3.6.13 Measured and Predicted Impedance Loci of Ka-band ASP (Antenna
Parameters: Dielectric Layers : Er • = 1.0, d. = 0.1 rom, Ea = 2.2, d2 = 0.787 rom, Er3 = 1.07, d3 =
1.1 rom, Er4 = 2.2, d4 = 0.254 rom, Patches: P LI = PW1 = 2.5 rom, N = 2; Pu = P W2 = 2.4 rom, N
= 3, SL= 2.5 rom, Sw = 0.2 rom . Feed Parameters: Erf= 2.2, d r = 0.254 rom, Wr= 0.8 rom, L SlUb
=0.65 rom)

The E- and H-plane co- and cross-polar radiation patterns of the mm-
wave antenna were measured across the Ka-band. A typical measured
response for the co-polar E- and H-plane patterns are shown in Figure 3.6.14,
which were measured at 38 GHz. The evident ripple in the E-plane is typical
for an aperture-eoupled antenna with a finite ground-plane (for this antenna 5
x 3 em). Note also there is an apparent asymmetry in the E-plane pattern.
This is due to the presence of the K-type connector used to couple power to
and from the microstrip feed-line. The flange for the connector is relatively
large in comparison to the printed antenna and its shielding effect quite
pronounced for angles near the appropriate endfire (here e --7 + 90 0 ) . The
gain of the antenna was measured across the Ka-band and 'was greater than 6
dBi over this frequency range. The cross-polarization levels were more than
20 dB below the co-polar patterns in each plane. The front-to-back ratio was
greater than 12 dB across 26 - 40 GHz, approximately 4 dB greater than that
reported in Section 3.6.5 due to the reduced bandwidth requirement here and
hence reduced slot size of the antenna. The surface wave efficiency of the
mm-wave antenna was calculated as greater than 85 % across the Ka-band.

138
110

Figure 3.6.14 Radiation Patternsof mm-wave ASP Antenna

3.6-8 Backward directed radiation reduction techniques

3.6-8.1 Introduction

In this section we investigate techniques to reduce the backward


directed radiation for an ASP microstrip antenna. The applications for the
two resulting antennas are different, although they do share some common
goals, in particular reducing the unwanted backward directed radiation is a
must. Both examples are conducted at millimeter-wave frequencies and so
some comparisons between the two can be made.

3.6-8.2 Reflector Patch Solution

The simple technique of using a microstrip patch antenna element


behind the microstrip line fed aperture acting as a reflector to substantially
reduce the back radiation from an aperture-coupled antenna was proposed and
experimentally verified in Section 3.5. Please note that the proposed reflector
patch and the conventional radiating/directive patch share the same ground-
plane. It should also be noted that the reflector element principal of operation
is not the same mechanism used in [41, 42] for reflectarray styled printed
antennas. Reflectarrays, similar to classical reflectors, rely on far-field effects
to obtain the required characteristics. For the printed reflector element, it is a
near field interaction between the excitation slot and the reflector patch that
gives the desired far-field cancellation of the radiated fields.

While the reflector element was successfully demonstrated at


microwave frequencies over moderate bandwidths in Section 3.5, it has yet to
be determined whether this technique is suited to broadband printed antennas
or whether it can be used at mm-wave frequencies where the choice of
139
material for the back/reflector patch is somewhat limited. In this section, we
investigate the performance of a broadband mm-wave printed antenna with
reduced back radiation.

Antenna Configuration and Reflector Element Design

A schematic diagram of an ASP antenna with a microstrip reflector


element is shown Figure 3.6.15. Here the printed antenna consists of an
aperture in the ground plane fed by a microstrip line and a stacked patch
configuration. A reflector patch of length L, and width Wr is located on the
feed side of the ground-plane of the antenna. The reflector microstrip patch is
separated from the feed-line by a low dielectric constant laminate (typically
foam) of thickness d, to ensure the surface wave losses associated with this
structure are minimized. A thin dielectric layer is used to etch the conductor
of the reflector patch as depicted in Figure 3.6.15. The dielectric material
separating the feed-line and the reflector can be various thicknesses, however
it is typically electrically thick (greater than a.1Ao where Ao is the wavelength
corresponding to the center frequency within the operation band of the
antenna) and therefore direct coupling between this patch and the feed-line is
negligible.

Figure'3.6.15 Aperture Stacked PatchAntenna with Reflector Patch Element

To analyze the antenna configuration of Figure 3.6.15, the full-wave


spectral domain moment method model incorporating a multilayered Green's
function was utilized [15]. The analysis is very suited to this form of antenna
at mm-wave frequencies. as small air-gaps associated with the fabrication of
the multilayered antenna can be taken into account. As the design process
involves investigating the relative magnitude and phase of the radiated far-
fields due to the aperture and reflector, the radiated far-fields were computed
140
from the stationary phase evaluation of the Green's function for the field
components of the elements.

As mentioned in Section 3.5, the design procedure for an aperture


coupled microstrip antenna with a reflector element is reasonably
straightforward. If the spacer thickness is relatively large, then the impedance
response of the entire antenna is similar to that for an ASP without the
reflector element. To minimize the backward directed radiation there are
several parameters that can be varied. The length of the reflector patch
primarily controls the relative phase between the slot and the reflector patch,
while the relative magnitudes are controlled by the width of the patch and the
spacing between the slot and the reflector. Intuitively this latter variation
makes sense because as the distance between the slot and the reflector patch is
increased, it is expected that the power coupled to this patch will decrease and
hence the change in relative magnitude. The form of redundancy in the
design procedure provided by the width and spacing relationship is very
important since there are only a limited number of foam thicknesses
compatible with the Ka-band frequency range and therefore varying the width
of the reflector patch will give the finer phase control needed to provide good
field cancellation. Figures 3.6.16, 3.6.17 and 3.6.18 show the effect of
varying these parameters on the FIB ratio performance for a series of ASP
antennas incorporating reflector elements, based on the ASP with no reflector
element presented in Section 3.6-7.2. As can be seen from Figure 3.6.16, the
longer the reflector patch length, the broader band of frequencies where the
FIB ratio has been significantly improved, until a point where at the higher
frequencies the performance starts to degrade. Figures 3.6.17 and 3.6.18
show evidence of the previously discussed flexibility in the design procedure.
From these figures it is evident that increasing the spacer thickness has a
similar effect on the FIB ratio as decreasing the width of the reflector patch.

-' .
40
·r·' :--'~'"
. ~

'. ~~...~ :....... .......


" I~ ~
'.

~ , ,, .'
35 ~-
.~-,- "~.....~

. .,
lil30
~25
o
~20
,,
,.
1/ ,.' ,,~
/ , ,, .' "
.'
Increase LR
\.
-,
~
u 15 ~
~ ' .' V
.,~
<, '\
0 10
";' -r ,
.' .' no_reflector
-,
C5 ---- 6.5mm
e0 /
IL
•..•..·· 6.1mm
- ·- ·6.9mm
"-
-5
i -"-" 5.5mm
7.5mm
-10
22 24 26 28 30 32 34 36 38 40 42 44
Frequency (GHz)
Figure 3.6.16 Effect of Length of Reflector Patch on Front-to-back Ratio

141
45

40
.... ......
' . ....... .. .
;J;'.
m35
:2- ~ ..., , .... ......... ...........
..../" ......
030
i ..,:>\ IncreasedR ....... ....... ..'
....,
~25 , .......'
u
~20 .".."'1
.~;;-
'. . ~~,
~15
c
o
.
.t10 f.it
~,.
.......V - no_reflector -, ......
V ····1.1mm
....····1.3mm '\.

"
5
- •• • O.9mm
o
~ ~ U ~ 30 ~ M ~ ~ ~ ~ ~
Frequency (GHz)

Figure 3.6.17 Effect of Spacer Thickness for Reflector Patch on Front-to-beck Ratio

r---.-r---.-r--.-.--.-.--.-.----,
40
..
..... ,~' :,.............

~30 1--t---"l~'_;jL-I---+-~t--'-t"""""'"'i--o.m--+---j
/,'11...;
o : ~#},
~ 25 F--++t-:L~,.___-b-"o.q...-+--1--+--P~+--f

~20 I-+~I-*---::~--+--t--+-+--+---"\~--j
X
~c·15
..
"
.'/
.l,~'· . / - r:
no_re IIec t or
~.
'.
E10 ~"""':"--1I---H"" O.3mm H--I--t--'<:I--;
IL
5 ....
~. ........ 0.2mm
- .-. O.4mm
·"···1.Omm

Figure 3.6.18 Effect of Width of Reflector Patch on Front-to-back Ratio

For the proposed mm-wave antenna, we chose the following


parameters for the reflector element: Lr = 6.5 mm, Wr =0.3 mm and d, = 1.0
mm (the thinnest foam layer available at the time of fabrication). Although d,
can be made smaller, the advantage of choosing a value greater than 0.1 Ao is
that the reflector patch will have a minimal effect on the input impedance
response of the entire antenna. For the chosen parameters, as can be seen
from Figure 3.6.16, the theoretical FIB ratio has been significantly enhanced
with the minimum ratio being 28 dB. The simulated input impedance
response of the ASP antenna with a reflector element of these dimensions is
similar for the ASP antenna without the reflector element and both are shown
in Figure 3.6.19 and the 10 dB return loss bandwidth ranges from 26 to 41
GHz.

142
Figure 3.6.19 Simulated Input Impedance of ASP with and without Reflector Element

It is interesting to note that despite the large electrical thickness of the


proposed mm-wave printed antenna configuration (including the reflector
element), its surface wave efficiency is still above 85 % across the entire Ka-
band. This is contrast with what is usually perceived for a printed antenna of
a total thickness of greater than 0.2 An. The reason for its high efficiency is
quite simple: there are multiple radiators within this structure and these
elements can couple power from the potentially detrimental surface waves.
This was shown in Section 3.3, where a stacked patch antenna utilizing high
and low dielectric constant materials had very good efficiency despite the
material selection. It appears that the trends for degradation in efficiency as
the material thickness increases (see Chapter 2) are only valid for an
environment where there is only one radiating patch element. In the cases
examined here there are several radiators, namely, the slot, the two directive
patches and the reflector patch.

3.6-8.3 Cavity Backed Solution

Although the ASP can be easily integrated with mm-wave and


photonic devices in its present form (Figure 3.6.1), one potential difficulty is
that these active devices need some form of heat-sinking and support
structure. To implement this for the ASP is not typically straightforward as
the performance of the antenna may be compromised if a metal plate for
mechanical support is located in close proximity to the electrically large slot
or if the ground-plane in which the slot resides is electrically thick [43]. One
possible solution to the design problem is to incorporate a cavity-backed
structure, where a cavity is located under the slot of the antenna. This not
143
only provides a good support structure for the antenna and the active devices,
but will also improve the front-to-back ratio of the antenna itself.

Figure 3.6.20 shows a cross-sectional view of the cavity-backed ASP.


Here an air-filled rectangular prism of dimensions Ae x Be X De, resides below
the slot of the ASP. As can be seen in Figure 3.6.20, the microstrip feed-line
is located on the same side of the ground-plane as the patches. To
theoretically investigate this printed antenna, the analysis presented in [15]
was modified to take into consideration both the electric sidewalls and the
bottom of the cavity. In the analysis it is assumed that there is no direct
coupling between the microstrip feed-line and the patch antennas. For a
cavity-backed version of an ASP antenna this assumption is quite valid as the
laminates used for the patches are electrically thick, typically greater than
0.07 x,
patches
feed network

I--::::r----~---"ro:­
~i
metallic cavity

Figure 3.6.20 Cavity-backed mm-wave ASP

There are some subtle differences in the design procedure of the


cavity-backed ASP compared to the conventional ASP. Firstly, the cavity
dimensions must be sufficiently larger than the slot such that the bandwidth is
not adversely affected by its presence. Also the thickness of the feed layer
must be taken into account when determining the impedance characteristic of
the ASP as a function of the height of the laminates. Aside than these
modifications, the same design procedures in Section 3.6.4 were adopted for
the Ka-band cavity-backed ASP. A similar material selection to that used in
the conventional ASP was incorporated for the substrate of the lower patch,
with the exception that a thinner layer of RT/Duroid 5880 (d = 0.508 rom)
was used. The required cavity (5 x 5 x 3 rom in size) was machined into a
brass block with dimensions 20 x 20 x 10 mm, and a K-type connector was
mounted on the side of this block to couple power to and from the rnicrostrip-
line.

144
Figure 3.6.21 shows the predicted and measured return loss of the designed
Ka-band cavity-backed ASP (refer to the figure captions for the dimensions of
the antenna). As was the case for the ASP, there is relatively good agreement
between theory and experiment with the bandwidth requirements clearly met.
Figure 3.6.22 shows the measured E- and H-plane co-polar radiation patterns
.of the cavity-backed antenna at 32 GHz. Once again the ripple and
asymmetry in the E-plane is evident in the plot. These features can be
attributed to both the finite ground-plane, which for this antenna is even
smaller than for the fundamental ASP of Section 3.6-7 hence leading to the
large ripple, and the K-type connector. For the cavity-backed ASP this
connector is located closer to the ASP and therefore its effect on the radiation
pattern is more dramatic than in Figure 3.6.14. The gain of the antenna was
measured as greater than 6 dBi across the 26 - 40 GHz frequency range and
as before the cross-polarization levels in each plane were at least 20 dB below
the co-polar patterns. The front-to-back ratio of the cavity backed antenna
was measured as greater than 20 dB across the Ka-band.

."
~
-5
~\
-m -: , , ,..--- '.
-10 ,,~~~

-
~
\, ,, ,
<,
-15
/ ,
'tJ ~

I~
,
(I)
(I) ·20
r\ /,l
.. ,
I
0

"
..J ~ I
f
c ·25
::J
l) -30
I
, I
I
\....,f
a: II

I
v
I--theory
·35 ,- - - - experiment I
-40 I

-45
25 27.5 30 32.5 35 37.5 40
Frequency (GHz)

Figure 3.6.21 Measured and Predicted Return Loss of Ka-band Cavity-backed ASP
(Antenna Parameters: Dielectric Layers: Ert = 2.2, d t = 0.125 nun, E,z = 1.0,d2 = 0.1 nun, Er3 =
2.2, d3 = 0.508nun, E,4 = 1.07, d, = I.l nun, E,4 = 2.2, d4 = 0.125 nun, Patches: PLI = PWt = 2.5
nun, N =3; Pu = PW2 = 2.3 nun, N = 4, SL = 2.45 nun, Sw = 0.2 nun. Feed Parameters: En =
2.2, de = 0.125 nun, We = 0.8 mm, L.,Ub = 0.75 nun, Ae = Be = 5 nun, De = 3 nun, Ere•• = 1.0)

145
3.6-9 Dual Polarized ASP

3.6-9.1 Introduction

There are a variety of applications requiring broadband, efficient,


dual-polarized antennas including polarimetric Synthetic Aperture Radar
(SAR) and multi-service wireless communication base stations. In addition to
the above-mentioned electrical/radiation properties, the environment in which
these applications/systems are utilized necessitate the need for the antennas to
be conformal, low in profile and easily implemented into arrays. There have
been reported several variations of the microstrip patch antenna that yield a
broadband, dual-polarized radiator (see for examples, [44 - 45]). These
techniques incorporate a twin slot/aperture excited patch to give good
polarization control. VSWR < 2:1 bandwidths in the order of 25 % and
isolation between the two polarizations of greater than 35 dB have been
achieved. Although these configurations can be used in some wireless and
SAR applications, a truly broadband dual-polarized printed antenna solution
has yet to be found.

Recently a broadband microstrip patch antenna was developed that


had an impedance bandwidth in excess of 50 % [15]. The printed antenna
utilized an aperture stacked patch (ASP) geometry and had constant gain
(greater than 7 dBi) across the entire matched impedance bandwidth. A
consequence of the ASP configuration is the relatively poor front-to-back
radiation ratio (sometimes less than 10 dB) resulting in reduced radiation
efficiency for SAR applications or sector interference for cellular systems.
Fortunately a simple technique was introduced in Section 3.6-8 to overcome
this problem.

In this section we present a broadband, dual-polarized printed antenna


suitable for polarimetric SAR and multi-service wireless communication base
stations. The antenna utilizes the ASP structure presented in Section 3.6-7
and separate feeding layers (similar to the arrangement summarized in Section
3.5-3.4). To reduce backward directed radiation in both polarizations, a
printed-eross back patch was incorporated. The overall antenna has an
impedance bandwidth of greater than 50 %, a port-to-port isolation of greater
than 39 dB and a front-to-back ratio of more than 20 dB across the entire
impedance bandwidth for both polarizations.

3.6-9.2 Dual Polarized ASP Design

The geometry of the dual polarized ASP antenna is shown in Figure


3.6.22. It consists of four layers of RT Duroid 5880 laminate (Er3. Er4. Ern. Er8),
146
one layer of FR4 dielectric material (Crl) and three air/foam layers (Cr2' crS, Cr7)'
To achieve a high degree of isolation between the two polarization ports over
all frequencies, perfect symmetry is required for the design of the antenna. By
using a centered crossed slot to feed the patch element, a high degree of
symmetry can be maintained, however this requires a balanced feed network.
Utilizing a single feedline on each polarization of the crossed slot introduces
asymmetry and degrades the isolation between the two ports.

~
c r8. h8
Patch 2 (W2. L2) - - -.L....",,- ~
cr7. h,

~
c r6. h6
~
Patch I (W I. LI) crs. hs

~
Feed 2 (W2F. cr4. h,
L2SruD)
~
cr3. h3
Er2. h2

~
ErI.hl
Reflector (W I REP Ll REF
Reflecto r (W2REP L2RE!F)c:====::t::

Figure 3.6.22 Geometry of Dual Polarized ASP Antenna

In the dual polarized configuration each feed network is placed on a


separate layer on opposite sides of the ground plane of the antenna. The input
impedance for the two ports will differ slightly as the upper feed network can
couple to the patch elements. However this is typically negligible as
electrically thick layers are used between the slot and the patches of the ASP
and so direct coupling from the feed-line to the patches should be minimal.
Having said that, the height of two feeding layers must be chosen so that the
mutual coupling between the slot and patch elements remains the same for
both polarization ports.

As shown in the previous sub-section, ASP antennas normally


produce a relatively low front to back ratio (FBR, typically 10-14 dB) due to
the resonant aperture. To increase the FBR of the antenna, a microstrip
antenna element can be placed behind antenna as a reflector as shown in
Section 3.6-8. For the application considered here a cross element was
chosen to ensure minimal back radiation in both polarizations was achieved.
147
To reduce the cost of the antenna and to make it more rigid, an FR4 substrate
was used for the microstrip reflector. The spacing between reflector and
feedline was selected in order to achieve a FBR of greater than 20 dB without
affecting the input impedance behavior of both polarization networks. Figure
3.6.23 shows the photograph of dual polarized ASP antenna and how it was
constructed. Throughout this work, Ensemble 6.0 was used to predict the
performance of the dual polarized ASP antenna.

Figure 3.6.23 Photograph of Dual Polarized ASP Antenna

Figure 3.6.24 (a) and 3.6.24 (b) show the measured and calculated
return loss for Port 1 and Port 2 of the dual polarized ASP antenna,
respectively. As can be seen from these plots, there is good agreement
between theoretical and experimental results. Bandwidths of greater than 52%
(VSWR < 2:1) for both polarizations were obtained. The lower edge of the
matched band is slighter lower for Port 2 than Port 1. This can be attributed
to the height of upper feed substrate and so the coupling between slot and the
patch element for this polarization is stronger than between slot and the patch
element for Port 1 excitation. The shift is less than 1 %. Figure 3.6.25 shows
measured and calculated isolation between two ports. An isolation of more
than 39 dB over the bandwidth has been achieved. As mentioned before, the
two feed networks are separated by the ground plane, thereby eliminating any
cross coupling between the microstrip-lines in the two networks.

A full radiation characterization of the antenna was conducted and the


results at the center frequency (2 GHz) are shown in Figures 3.6.26 and
3.6.27. The measured E-plane radiation patterns of both ports are shown in
Figures 3.6.26 (a) and (b). The cross-polarization level in this plane is less
148
than -19 dB in the entire upper half space. The measured H-plane radiation
patterns for both ports are shown in Figures 3.6.27 (a) and (b). The H-plane
cross-polarization level is less than -24 dB in the upper half space. The gain
of the antenna was greater than 7 dBi. An FBR of approximately 20 dB was
obtained for the antenna over the impedance bandwidth.
0 0
-.5
-5
·10
·15 · 10
~ ·20 Iii
Eo · 15
·25
;;; ~
-30 ·20
-35
·25
-40
-45 -30
1 1.5 2 2.5 3 1 1.5 2 2.5 3
Frequency (GHI) Frequency (GHI)

(a) (b)
Figure 3.6.24 Measured and Calculated RL of Dual polarized ASP Antenna : (a) Port I,
(b) Port 2 (------- calculated)
· 30

· 35

·40

en
:Eo
·45
I
I

i}j ·50 ,
I

,
I
I
·55 I
I
I
I
I
·60 \ I
I ,I'
·65
1.5 2 2.5 3
Frequency (GHz)

Figure 3.6.25 Measured and Calculated Isolation between Port 1 and Port 2 (------ calculated)

Figure 3.6.26 E-plane Radiation Patterns of Dual Polarized ASP Antenna: (a) port I, (b)
Port 2 (---- X-pol.)
149
Figure 3.6.27 H-plane Radiation Patterns of DualPolarized ASP Antenna: (a) Port I, (b)
Port2 (----X-pol.)

3.7 Ultra-wldeband ASPs

3.7-1 Introduction

A wireless application recervmg much attention recently, is the


development of software defined radio (SDR) systems. Essentially these
systems do not need conventional circuits to implement radio transmitters or
receivers at the base stations, as these processes can be done substantively in
software. The appeal of SDR is that the modulation/demodulation processes
are specified by software in the digital processor, thereby allowing different
modulation formats and protocols to be accommodated for by simply
changing the software. Utilizing such a scheme permits for systems that can
readily contain different protocols and therefore accommodate separate
wireless interfaces, such as 30, Bluetooth and WLAN with a common base
station. Imperative for this development are base station antennas that can
provide the necessary air-interface and operate over very large bandwidths,
more than likely greater than an octave.

In this section we introduce a technique by which the bandwidth of


the ASP antenna may be effectively doubled with minimal impact on its gain
characteristics. The method used to extend the bandwidth of the ASP is to
divide the required band into several sub-bands and have suitably designed
antennas for each of these sub-bands. In essence a log periodic antenna is a
series version of this with antennas covering different bands cascaded one
after the other. A more robust method of providing extremely broad
bandwidth is to implement a parallel version of the subdivision where the
signal is first split into the sub-bands using a multiplexer and then each of the
split signals is fed to a specially designed antenna. The arrangement used in
this investigation consists of a pair of ASP antennas, designed to cover
150
adjacent bands attached to a diplexer. If the whole unit can be considered a
single antenna, then its bandwidth is twice that of one of the ASPs. The key to
such a technique is the design of the diplexer to ensure the correct band of
frequencies is delivered to the appropriate patch.

3.7-2 ASPs: design and results

3.7-2.1 Lower Frequency ASP

Based upon the parameter and characteristic investigation in Section


3.6-4, two ASP antennas in the frequency range of I-20Hz (antenna 1) and 2-
40Hz (antenna 2) were designed. These antennas consist of five layers, the
first layer is the feed layer below the aperture and all other layers are above
the ground-plane . The second is a foam layer, the third a thick RT Duroid
5880 layer e, = 2.2), the fourth a foam layer and finally the fifth is a relatively
thin RT Duroid 5880 layer. It was found through many simulations that this
combination of the materials gave very good impedance bandwidth.
Rectangular patches were used for additional bandwidth. For dual or circular
polarization, square patches can be used which decrease bandwidth by 5 to 10
%. The dimensions of the two antennas are given in the Tables 3.7.1 and 2.

Antenna 1: 1 - 2 GHz
Feed width, Wr 1.4 nun Feed thickness, hr 1.58 nun
Stub length, 1, 26 nun Laver 1 thickness, hi 19 nun
Offset, Lorr 30 nun Layer2 thickness, h2 3.175 nun
Aperturelength, L. 84 nun Layer 3 thickness, h3 9 nun
Aperturewidth, W. 10 nun Layer4 thickness, !4 1.58 nun
Patch 1 length,L1 64 nun Feed, En 2.2
Patch 1 width,WI 150 nun Layer 1 Erl 1.07
Patch 2 length, k 60 nun Laver 2, Er2 2.2
Patch 2 width,W2 150 nun Laver 3, Er3 1.07
Layer4, Er4 2.2

Table 3.7.1 Dimensions of Lower Frequency ASP

Antenna 2: 1 - 2 GHz
Feed width, Wr 1.4 nun Feed thickness, hr 1.58nun
Stub length, 1, 14 nun Layer I thickness, hi 8 nun
Offset, Lorr 15 nun Laver 2 thickness, h2 3.175 nun
Aperturelength, L. 41 nun Laver 3 thickness, h3 5 nun
Aperturewidth, W. 5 nun Laver4 thickness, !4 1.58 nun
Patch 1 length,L1 32 nun Feed,Err 2.2
Patch I width, WI 65 nun Laver 1 Erl 1.07
Patch 2 length, k 30 nun Layer2 ,Er2 2.2
Patch 2 width,W2 65 nun Laver3, Er3 1.07
Layer4, Er4 2.2

Table 3.7.2 Dimensions of Upper Frequency ASP

151
The calculated and measured Return Loss (RL) for the lower
frequency ASP are shown in Figure 3.7.1. The calculated bandwidth (RL < -
10 dB) is 1-2.3 GHz (75 %) and the measured impedance bandwidth (RL < -
10 dB) is 1-2.1 GHz (71 %). There is a discrepancy between the predicted and
measured results, with a more important impact at the higher frequency band
of the ASP. One possible cause of this narrowing of the acceptable return loss
bandwidth could be due to the construction of the multi-layered antenna. As
can be seen from Table 3.7.1, the first and third layers are required to be
19mm and 9mm of foam respectively. These layers were constructed using
IOmm, 5mm and 4mm sheets of Rohacell. Thus air-gaps could have been
present in the resulting ASP, reducing the overall bandwidth. However
despite this, there is a good agreement between measured and computed
return loss, so further iteration is not required.

10

iii
~ 15
oJ
a: 20

25

30 '----~---~---'------'
0.5 1.5 2 2.5

Frequency (GHz)

Figure 3.7.1 Computed and Measured ReturnLoss for ASP 1

The E and H-plane radiation patterns (predicted and measured) for the
lower frequency ASP at 1.5 GHz are shown in Figure 3.7.2. The Front to
Back Ratio (FBR) is a bit higher than predicted which, is mainly due to the
use of a small ground-plane (the ground-plane of the antenna). The size of
ground-plane was 1.4 Ao x 0.7 Ao (at the center frequency). Using a larger
ground-plane would provide better FBR, however that increases the weight
and the cost of the ASP antenna.

Predicted and measured gain of the lower frequency ASP is presented


in Figure 3.7.3. A gain of 8 ± 1.5dBi in the frequency range of 1-2GHz was
achieved. The measured input impedance bandwidth for this ASP was 1-
2.1GHz, however the measured 3dB gain bandwidth limits the total
bandwidth to 69 %. From Figure 3.7.3 it can be seen that the gain starts to
drop sharply beyond 2GHz even though the return loss is less than -10dB.
152
This is due to the aperture: as the frequency increases, the slot becomes the
more dominant radiation mechanism and hence the gain drops quickly, due to
the slot radiating in both the front and back half spaces.

EXP -
T>£O - •

(a) (b)
Figure 3.7.2 Radiation Patterns of ASPl at 1.5 GHz: (a) E-plane; (b) H-plane

The Front to Back Ratio (FBR) of the ASP was measured to be


greater than 8dB for frequencies less than 2GHz. The co-polar and cross-polar
radiation patterns are very similar to the ASP presented in Section 3.6 and so
for the sake of brevity. they are not presented here.

TliEO-
10


f •
j

I

Figure 3.7.3 Computed and Measured Gain of ASP 1

3.7-2.2 Upper Frequency ASP

Theoretical and experimental return loss response for the upper


frequency ASP is shown in Figure 3.7.4. Once again there is a good
agreement between the calculated and measured results. The calculated
bandwidth (RL < -10 dB) is 2-4.25GHz (72 %) and measured impedance
bandwidth is 2 - 4.5 GHz (79 %). It should be noted that the SMA connector
and the section of transmission line between the connector to the slot was not
de-embedded from all the measured return loss results presented here. Hence
153
there are fme discrepancies between the predicted and measured values of the
return loss.

10 (

r-. i
1/
: : \ 0

00 ;

:: \,
25 II
I
0 0

,0

.f
0'
30

Frequency (GHz)

Figure 3.7.4 Computed and Measured Return Lossof ASP2

The radiation patterns (E and H-plane) at the center frequency of the


band for the upper frequency ASP are shown in Figure 3.7.5. Similar to the
lower frequency ASP the measured FBR is lower than the calculated FBR,
due to the small ground-plane. The ground-plane of antenna has dimensions
of 10cm x 5cm.
EXP - EXP -
T~EO THEO - 0

(a) (b)
Figure 3.7.5 Computed and Measured Radiation Patterns of ASP2 at 3 GHz: (a) E-
plane; (b) H-plane

The predicted and measured gain of the upper frequency ASP are
shown in Figure 3.7.6. This ASP has a measured gain of 8 ± 1.5dBi in the
frequency range of 2-4.1GHz. This gain response reduces the bandwidth of
the second antenna from 79 % to 70 % as the 3dB gain bandwidth is 2 -
4.15GHz. From Figure 3.7.8 it can be seen that the gain drops very rapidly at
4.10Hz. Once again the"radiation patterns of this ASP were very similar to
typical ASP antennas that are presented in Section 3.6. The FBR for this ASP
antenna over the 2-40Hz was no worse than 8dB.

154
El!J'-
THEO -
10

I u • u
'-flHol

Figure 3.7.6 Computed and Measured Gain of ASP 2

3.7-3 Broadband Diplexer Design and Results

Consider the diplexer that is shown in Figure 3.7.9. It consists of a


low pass (LP) and high pass (HP) filters which are connected in shunt. A
signal launched into Port 1 will propagate to either Port 2 or Port 3 depending
on its frequency.

The requirements for the filters compnsmg the diplexer for this
application are as follows. The filters must have pass bands at least equal to
the broadband antennas to which they are attached. The return loss within the
band of each filter must be as low as possible to ensure that the total RL of the
UBP antenna is less than -10dB across the entire band of operation. For this
application one filter must pass 1-2GHz, and reject 2-4GHz and the other
filter must pass 2-4GHz while rejecting 1-2GHz. It is imperative that each
filter has a sharp roll-off to ensure array-like distortions of the radiation
pattern of the UBP does not occur. For this case the interaction is at 2GHz.

Further, since the filters will be imperfect and are of limited order,
they will have finite impedance characteristics in their stop band. The result
of this is that when the two filters are connected, they will effectively load one
another and influence the respective band performances. In order to transfer
all the power to the designated port, the input admittance of each filter must
be made complementary. In normalized form this condition can be
represented as in [46]:

(3.6)

Equation 3.6 can be expressed in terms of real and imaginary components :

(3.7)

155
(3.8)

A final requirement, once all others have been met, is that filters must be
composed of realizable components.

There are many configuration options for designing diplexers


including LPIHP, LPlband pass (BP), BPlband stop (BS). The proposed
diplexer for this application was constructed using a LPIHP arrangement as it
is adequate for a diplexer. Maximally flat filters were considered for this
diplexer due to their good return loss characteristics.

To produce a minimum susceptance network using the singly


terminated method the first elements of the LP and HP filters were a series
inductor and a series capacitors respectively. If two or more minimum
susceptance networks are connected in parallel, the total network susceptance
is also a minimum.

The first step in the filter synthesis is to determine the number of


element required to produce the required insertion loss at a particular
frequency in the filter stop band. The greater the number of reactive elements
the sharper frequency response is, however this increases the insertion loss of
the filter. To calculate the required isolation, some initial tests were carried
out to see what level of mutual coupling between the two ASP antennas
would occur, when they are placed next to each other. To minimize the effect
of mutual coupling it was found that at least 13 dB loss at 1. 1roc (roc is cut-off
frequency) was required. A 15 element maximally flat filters will provide
adequate attenuation at the required frequency. Having discussed the number
of elements, the LP filter required here, was designed using simple filter
synthesis (refer to [46] for details).

Microstrip technology can be readily used to implement a low pass


filter as series inductors and parallel capacitors can easily be realized. A thin
microstrip line will present an inductor which connected to a thicker
microstrip line produces a capacitance [47]. The width of the capacitance
section of line should be less than a quarter wavelength of the highest
frequency in order not to support a transverse resonance.

A high-pass filter requires series capacitors that can be difficult to


realize in rnicrostrip. For a 20Hz HP filter required here lumped element
capacitors can be used but they must be chosen so that their resonant
frequency is much higher than the operating frequency of the filter. Shunt
inductors can be realized using high impedance transmission lines terminated
with short-circuits.
156
Using lumped elements has its disadvantages, as it is difficult to
purchase the exact values of capacitors required. The design parameters for
the high pass filter after its synthesis needed to be reconsidered and
redesigned to produce the required performance given the available
capacitors. For instance, the first capacitor the value calculated from the
synthesis was 1.15 pF and nearest available capacitor was 1.2 pF. The value
of the first inductor then needed to be altered so the performance of high pass
filter remained constant. EESOF was used to simulate the performance of the
low-pass and high pass filters and hence, the diplexer.

When the realized filters were connected in parallel there was a dip
and then a sharp rise at the cross over frequency in the simulated results,
which implied that the imaginary part of the admittances were not canceling
each other. In order to reduce this distortion, the value of the first capacitor
for the high pass filter was increased and the value of the first inductor of the
low-pass filter decreased. The values of the other elements needed to be
altered in order to minimize the ripple in the pass band as well as reduce the
susceptance.

The final parameters and lumped element values for the diplexer are
shown in Table 3.7.3. It is necessary to point out that the width of microstrip
lines were calculated using equations in [47]. The width of the required
inductors were 0.33 mm which corresponds to 100 0), and the width of the
capacitors were 3.3 rom which equates to 25 0) using a substrate with a
dielectric constant of 3.05 and a height of 0.508mm.

Low Pass Filter High Pass Filter


C 1 = 12.8 mm L =1O.7mm C = 0.9 of L 1 =3.60mm
C2= 12.6mm ~= 12.9mm C2= 0.8 of J2=3.70mm
C~= 11.4 mm ~= 12.1 mm C~ =0.90F 4=4.10mm
C4=9 .63 mm L4= 10.58 mm C4= 1.0 pF .04=4.48 mm
Cs= 7.43 mm Ls= 8.58 mm Cs = 1.2 of L, = 6.28 mm
C.. = 4.9 mm L.. = 6.2 mm C.. = 1.5 of L,,= 9.55 mm
C7 = 2.13 mm L,=3.5 mm C7 =3.3 pF L,= 12.8 mm
Ls = 0.71 mm c, = 8.2 pF

Table 3.7.3 Parameters for Maximally FIat Diplexer

A diplexer comprising of the maximally flat filters was constructed.


Figure 3.7.7 shows the experimental insertion loss of this diplexer that is very
similar to the theoretical results (also shown here). It can be noticed that at 3.8
GHz the insertion loss starts increasing dramatically. This loss is due to the
157
last capacitor of the high pass filter. The value of this capacitor is much larger
than the other capacitors (8.2 pF), and it has a resonance at 3 GHz
(specification of self resonance of C17 DLI capacitors). If a smaller value is
chosen it will change the performance of the diplexer at other frequencies
within the specification. Using this design limits the operating frequency to 1-
3.8GHz instead of l-4GHz.
EXPS21 _ . -
EXPS31 - .
THEOS21-
THEOS31-

iD
0 1--- - - _ . .-..-...--............-=---4
-" '- .
~ 10
C
o
.. 20
eu
:J
c
1lI 30

~
40

2.5 3 3.5 4.5

Frequency (GHz)

Figure 3.7.7 Predicted and Measured Insertion Losses of Diplexer

The theoretical and measured return loss of the maximally flat


diplexer is shown in Figure 3.7.8. In the high pass region the measured
reflection is higher than the calculated value. In the simulation the capacitors
were considered as ideal lumped element or as a black box that contained the
S-parameters of that element, however in practice these values are not
accurately known (± 5 % tolerance). Also the size of the capacitors are
electrically large for the higher frequencies. Despite, these concerns, the
overall results are well within specification for the entire band, limited to a
maximum of 3.8GHz.
EXP - ·
THEO-

10
-"

U U 2 U 3 U 4 U

Frequency (GHz)

Figure 3.7.8 Predicted and Measured Return Losses of Diplexer

158
3.7-4 UBP Antenna Results

Figure 3.7.9 shows a schematic representation of an ultra broadband


printed antenna where a diplexer is used to divide the operating frequency and
two ASP antennas are connected to the output ports of the diplexer. To
produce an ultra broadband printed antenna it is required to investigate the
characteristic of the two concepts previously discussed (ASPs and the
diplexer) when they are connected in series. The characteristics investigated
are the input impedance, the radiation pattern in both E-and H-planes and the
gain.
f----------------------------------------.··-------------------------- -----------.-- - -. - - -- -- - ~

al,~ I al
---.... : --...
I
i
_1_ rn~~~ 1
bll~ i bl i
i i
l !
Figure 3.7.9 Block Diagram of Ultra-wideband Printed Antenna

The influence of each ASP antenna on the input impedance of


diplexer and hence, the input impedance of the UBP antenna needs to be
investigated. It can be predicted that the input impedance of the UBP antenna
is dependent on the return loss of the diplexer, the insertion loss of the
diplexer and the individual return losses of each antenna. If we set our goal
for a return loss of better than -lOdB for the UBP, the return loss of the
individual antennas and the diplexer needs to be at least less than -13dB,
assuming an insertion loss of no worse than -1.5dB.

One of the main objectives of the UBP is to achieve a relatively


constant gain over the entire band. To reduce the overall size of the UBP
antenna, two ASP antennas were placed next to each other. Figure 3.7.10
shows a photograph of the UBP antenna. The question arises whether the two
ASP antennas will interfere causing the pattern to be distorted. Another
concern is whether the insertion loss and mutual coupling at the cross over
frequency of the diplexer will be an issue when the input impedance and the
gain of ultra broadband antenna is measured. These issues will be
investigated in the return loss and radiation pattern measurements.

Theoretical and measured results for the return loss of the UBP
antenna are shown in Figure 3.7.11. As it can be seen from Figure 3.7.11 the
measured return loss is not as low as the predicted results. At higher
frequencies (> 4GHz) the measured return loss approaches OdB, while the
159
predicted results approaches -5dB. This is mainly due to the difference in the
simulated and measured results of the diplexer alone. Referring to Figure
3.7.8 the measured return loss for the diplexer at higher frequencies (> 40Hz)
is around -5dB, while the simulated results are around -13dB. These
discrepancies were discussed in earlier. The fine discrepancies apparent in
Figure 3.7.11 between the experimental and theoretical results are due to no
de-embedding of the SMA conductors used in the UBP, as mentioned
previously.

Figure 3.7.9 Photograph of Ultra-wideband Printed Antenna

EXP- ' "


THEO--
o

iii 10
:2-
..J
II:
15

20

25

3Q L-_ "'--_ -'--_ -'----'---'-_ ..i-J'----'-_-'-_ --'-_ --J


0.5 1.5 4 4.5 5

Figure 3.7.10 Measured and Predicted Return Lossof UBP

The radiation performance of the .proposed UBP was investigated


across the entire band of operation, namely 140Hz. There are 3 bands of
interest which provide valuable information on the radiation performance of
the UBP: (1) the lower frequency band where the lower frequency ASP

160
should be dominant; (2) the cross over' frequency where the mutual coupling
between the two radiators may be detrimental; and (3) the upper frequency
band where the high frequency ASP should be dominant.

Figure 3.7.11 shows the E-plane radiation pattern for ultra broadband
antenna at 1.50Hz, 2 OHz and 30Hz. At 1.5 OHz, the minimal effect of the
second sub-antenna (240Hz) is observed as in the E-plane the second sub-
antenna will not affect the performance of the UBP antenna because it is
located in the H-plane. The measured pattern is very similar to typical ASP
antennas. The ripples in the radiation patterns are due to diffraction of
aperture radiation at the edge of ground-plane. These ripples are within 2-3dB
which is in an acceptable range for most printed antennas.

(a) (b)

(c)
Figure 3.7.11 Measured E-planeRadiation PatternsofUBP Antenna: (a) 1.5 GHz; (b) 2
GHz; (c) 3 GHz

Figure 3.7.12 shows the H-plane radiation pattern of the UBP at


1.50Hz, 2 OHz and 30Hz. There is minimal effect on the radiation pattern
due to the electrically small size of the second sub-antenna. At 2 OHz the side
lobes present indicates that the two sub-band antennas form an array which
produces a narrower beam, hence higher gain. However the insertion loss of

161
the diplexer at 2GHz (4dB) compensates for this variation in beamwidth
thereby yielding a constant gain antenna. For frequencies just beyond the
cross-over frequency (2GHz) less distortion in the H-plane pattern may have
been achieved, if the diplexer consisted of Chebyshev filters, with steeper
roll-offs than the used maximally flat filters. However, Chebyshev filters
would have compromised the return loss of the UBP. There is a distortion in
the right side of the radiation pattern of the UBP antenna at 3 GHz, due to the
electrically large size of the first sub-antenna.

(a) (b)

(c)
Figure 3.7.12 Measured H-plane Radiation Patterns ofUBP: (a) 1.5GHz; (b) 2.0 GHz
and (c) 3.0 GHz

Figure 3.7.13 shows the measured gain of the UBP. As can be seen
from this figure at least 6 dBi of gain has been achieved from 1 - 3.7 GHz.

3.8 Summary

The objective of this chapter has been to investigate methods of


enhancing the bandwidth of a microstrip patch antenna. A variety of methods

162
have been presented and explored, each with their own merits and degree of
complexity.
10 , - - - - - , - - - - - - . - - - r - - - - , - - - r- - r -- -,-- - - ,

O L---'~ _ __'__ _ . 1 __ _ ' __ ___l._ _ _ ' _ _ _ J . . . . . I __ __'

0.5 1.5 2 2.5 3 3.5 4 4 .5


Frequency (GHz)

Figure 3.7.13 Measured Gain of UBP Antenna

In Section 3.3 several horizontally coupled parasitic patches were


considered. In Section 3.4 we presented a procedure on how to design direct
contact stacked patches with bandwidths in excess of 25 %. To achieve this
degree of bandwidth a combination of low dielectric constant material and
foam for the lower and upper dielectric layers respectively should be used.
We have also discussed how the parameters affect the impedance locus of the
stacked configuration and which parameters should be varied to obtain the
desired response. Examples of various stacked patches (including cavity
backed versions) were given highlighting the ease of design of these antennas.
Also in Section 3.4 an investigation into the optimum choice of laminate
material for a direct contact stacked patch was given.

In Section 3.5 large slot excited patches were investigated and several
design examples presented. Techniques to overcome the backward directed
radiation were examined. In Section 3.6 design criteria for the ASP (aperture-
stacked patch) microstrip antenna, which utilizes a resonant aperture with
stacked patches, were presented. The characteristic action of the resonators in
this antenna produces a greatly enhanced bandwidth over that exhibited by
other aperture coupled microstrip elements. Impedance matching techniques
were discussed, and the effects of several key physical parameters of the
antenna were examined. Results of this parameter study provide a good
design guide for this antenna. A linearly polarized experimental design was
presented, from which an octave bandwidth was realized. A wideband dual-
163
polarized printed antenna based on a dual feed ASP configuration was also
presented. This microstrip antenna has a VSWR < 2:1 bandwidth of greater
than 50 % and isolation between the excitation ports of greater than 39 dB.

Finally an ultra broadband printed antenna consisting of two ASPs


and a broadband diplexer has been presented in Section 3.7. The bandwidth of
the UBP antenna is significantly greater than other reported microstrip patch
antennas. A bandwidth of almost 120% and gain of 6.5dBi across this band
was achieved. The radiation property of this antenna was measured at selected
frequencies, which was uniform and similar to the individual ASP antennas
performance. The UBP can be easily integrated into linear arrays, commonly
used for wireless communication base stations.

The decision of which of all the techniques presented to use can only
really be made taking into consideration other aspects of the antenna terminal.
A general 'rule of thumb' that applies here is to make the antenna as simple as
possible whilst stilI satisfying the performance goals. For this reason , the
progression of this chapter is probably the best plan of attack, when
considering an enhanced bandwidth problem. Further examples illustrating
this will be presented in subsequent chapters.

3.9 Bibliography
[1] T. Chio and D. H. Schaubert, "Effects of slotline cavity on dual-polarized tapered slot
antenna arrays", 1999 Antennas & Propagation International Symposium. Orlando
USA, pp. 130 -133, July 1999.
[2] Y. Qian, W. R. Deal, N. Kaneda, and T. Itoh, "A uniplanar quasi-yagi antenna with
wide bandwidth and low mutual coupling characteristics," 1999 IEEE AP-S
International Symposium, Orlando, FL, USA, pp. 924 - 927, July 1999.
[3] 1 M. Johnson and Y. Rahmat-Sarnii, "Genetic Algorithms and Method of Moments
(GAlMOM) for the design of integrated antennas," IEEE Trans. AntennasPropagat.,
Vol. AP-47, pp. 1606-1614, October 1999.
[4] H. F. Pues and A. R. Van de Capelle, "An Impedance Matching Technique for
Increasing the Bandwidth of Microstrip Antennas," IEEE Trans. AntennasPropagat.,
Vol. AP-37, pp. 1345 -1354, November 1989.
[5] l-S. Kuo and K. - L. Wong, "A dual-frequency L-shaped patch antenna," Microwave
& OpticalTechnology Letters, vol. 27, pp. 177-179, Nov. 2000.
[6] G. Kumar and K. C. Gupta, "Non radiating edges and four edges gap-coupled
multiple resonator broad-band microstrip antennas", IEEE Trans. Antennas
Propagat., Vol. AP-33, pp. 173-178, February 1985.
[7] R. B. Waterhouse, "Small microstrip patch antenna," Electronics Letters, vol. 31, pp.
604 - 605, April 1995.
[8] W. C. Chew, "A broadband Annular Ring Microstrip Antenna," IEEE Trans.
AntennasPropagat., vol. AP-30, pp. 918 - 922, May 1982.
[9] D. M. Kokotoff, 1 T. Aberle and R. B. Waterhouse, "Rigorous analysis of probe-fed
printed annular rings," IEEE Transactions Antennas & Propagation, vol. 47, pp. 384
- 388, Feb. 1999.
164
[10J S. A. Long and D. M. Walton, "A dual-frequency stacked circular-disc antenna,"
IEEE Trans. Antennas Propagat., Vol. AP-27, pp. 270-273, March 1979.
[IIJ 1. T. Aberle, D. M. Pozar and 1. Manges, "Phased arrays of probe-fed stacked
microstrip patches", IEEETrans. Ant. & Prop., vol. AP-42, pp. 920-927, July 1994.
[12J 1. T. Aberle and D. M. Pozar, "Analysis of infinite arrays of probe fed rectangular
microstrip patches using a rigorous feed model", Proc. lnst. Elec. Eng., Pt. H, vol.
136, pp. 110-119, April 1989.
[13J 1. S. Dahelle and K. F. Lee, "Characteristics of annular ring microstrip antenna",
Electron. Lett., Vol. 18, pp. 1051-1052, November 1982.
[14J Z. Nie, W.e. Chew and Y.T. Lo, "Analysis of the annular-ring-loaded circular disk
microstrip antenna," IEEE Trans. Antennas Propagat., vol. AP-38, no. 6, pp. 806-
813, June 1990.
[15J S. D. Targonski, R. B. Waterhouse, and D. M. Pozar, "A wideband aperture coupled
stacked patch antenna using thick substrates," Electronics Letters, vol. 32, pp. 1941 -
1942, October 1996.
[16J C. A. Balanis, AntennaTheory: Analysis and Design, Wiley, New York, 1997.
[17J J. T. Aberle and D. M. Pozar, "Accurate and versatile solutions for probe-fed
microstrip patch antennas and arrays", Electromagn., vol. 11, pp. 1-19, Jan. 1991.
[18J D. M. Pozar, "Microstrip antennas", Proc. IEEE, vol. 80, pp. 79-91, Jan. 1992
[19J J. R. James and P. S. Hall, Handbook of Microstrip Antennas, London, UK: Peter
Perergrinus, 1989.
[20J D. M. Kokotoff, R. B. Waterhouse, C. R. Birtcher and 1. T. Aberle, "Annular ring
coupled circular patch with enhanced performance,' Electronics Letters, vol. 33, pp.
2000 - 2001, Nov. 1997.
[21] R. B. Waterhouse, "The use of shorting posts to improve the scanning range of probe-
fed microstrip patch phased arrays," IEEE Transactions Antennas & Propagation,
vol. 44, pp. 302 - 309, March 1996.
[22] F. Croq and D. M. Pozar, "Millimeter wave design of wide-band aperture-coupled
stacked microstrip antennas", IEEE Transactions Antennas & Propagation, vol. AP-
39, pp. 1770-1776, December 1991.
[23] R. B. Waterhouse and N.V. Shuley, "Scan performance of infinite arrays of
microstrip patch elements loaded with varactor diodes," IEEE Trans. Antennas
Propagat., vol. AP-41, pp.l273 - 1280, Sept. 1993.
[24] 1. T. Aberle and F. Zavosh, "Analysis of Probe-fed Cicrular Micostrip Patches
Backed by Circular Cavities," Electromagnetics, Vol. 14, pp. 239 - 258, 1994.
[25] M. Lye, R. B. Waterhouse, D. Novak, F. Zavosh and J. T. Aberle, "Design and
development of a printed antenna remote unit for optically distributed mobile
communications," IEEEMicrowave and Guided Wave Letters, vol. 8, pp. 432 - 434,
December 1998.
[26] J.F. Zurcher, "The SSFIP: A global concept for high performance broadband
planar antennas", Electronics Letters, vol. 24, pp. 1433-1435, Nov. 1988.
[27] N.K. Das and D.M. Pozar, "A generalized spectral domain Green's function for
multilayer dielectric substrates with applications to multilayer transmission lines,"
IEEE Transactions on Microwave Theory and Techniques, vol. MTT-35, pp. 326-
335, Mar. 1987.
[28] S. D. Targonski and R. B. Waterhouse, "Reflector elements for aperture and aperture
coupled microstrip antennas," IEEE Antennas & Propagation Symposium, Montreal
Canada, pp. 1840 - 1843, July 1997.
[29] R. B. Waterhouse, M. Lye and S. D. Targonski, "Design of printed antennas for
mobile base station applications," Third Asia-Pacific Conference on
Communications, APCC '97, Sydney Australia, pp. 242 - 246, Dec. 1997.

165
[30] S.D. Targonski and D.M. Pozar, "Design of wideband circularly polarized
aperture coupled microstrip antennas" , IEEE Transactions on Antennas and
Propagation, Vol. 41, pp. 214-220 , February 1993.
[31] F. Croq and A. Papiemik, "Wideband aperture coupled microstrip antenna",
Electronics Letters, vol. 26, pp. 1293-1294, August 1990.
[32] 1.R. Sanford and A. Tengs , "A Two Substrate Dual Polarized Aperture Coupled
Patch ", IEEEAntennasand Propagation Symposium Digest, pp. 1544-1547,1996.
[33] D.M. Pozar, "A reciprocity method of analysis for printed slots and slot
coupled microstrip antennas ", IEEE Transactions on Antennas Propagat., Vol. AP-
34, pp. 1439-1446, December 1986.
[34] D. A. Gray, "Optimal cell deployment for LMDS systems at 28 GHz," Proc. Wireless
Broadband Conf., Washington DC, July 1996.
[35] H. Ogawa, D. Polifko, and S. Banba , "Millimeter-wave fiber optics systems for
personal radio communication," IEEE Trans. Microwave Theory & Techniques, vol.
40, pp. 2285 - 2292, Dec. 1992.
[36] Z. Ahmed, D. Novak , R. B. Waterhouse, and H. F. Liu, "37 GHz fiber-wireless
system for distribution of broadband signals," IEEE Trans. Microwave Theory &
Techniques, vol. 45, pp. 1431-1435, Aug. 1997.
[37] L. D. Westbrook and D. G. Moodie, "Simultaneous bi-directional analog fiber-optic
transmission using an electroabsorption modulator," Electronics Letters, vol. 32, pp.
1806 - 1807, Sept. 1996.
[38] K. Li, 1. X. Ge, T. Matsui, and M. Izutsu, "Millimeter-wave sub-carrier optical
modulation, photodetection and integration with antenna for optic fiber link system,"
Proc. 1999 IEEE MTT-S Int. Microwave Symp.• Anaheim, CA, USA, pp. 1015 -
1018, June 1999.
[39] A. Stohr, K. Kitayama , and D. Jager, "Full-duplex fiber-optic RF subcarrier
transmission using a dual-function modulator/photodetector," IEEE Trans.
Microwave Theory & Techniques, vol. 47, pp. 1338 -1341, July 1999.
[40] D. Novak , G. Smith, A. Nirmalathas, C. Lim and R. B. Waterhouse, (invited) "Fiber-
optic networks for millimeter-wave wireless communications," Proc. 1998 Asia
Pacific Microwave Conference, Yokohama, Japan, pp. 309-316, Dec. 1998.
[41] D. M. Pozar, S. D. Targonski and H. D. Syrigos, "Design of millimeter-wave
microstrip reflectarrays,' IEEE Trans. Antennas Propagat. , vol. 45, pp. 287 - 296,
February 1997.
[42] W. Menzel, D. Pilz and R. Leberer, "A 77 GHz FM/CW radar front-end with a low-
profile low-loss printed antenna," IEEE Trans. Microwave Theory Techn. , vol. 47,
pp. 2337 - 2241, December 1999.
[43] P. R. Haddad and D. M. Pozar, "Analysis of two aperture-coupled cavity-backed
antennas," IEEETrans. Antennas & Propagat., vol. 45, pp. 1717 -1726, Dec. 1997.
[44] 1.R. Sanford and A. Tengs, "A Two Substrate Dual Polarized Aperture Coupled
Patch",IEEETrans. Ant.& Prop. Symp. Digest, pp. 1544-1547, 1996.
[45] S.C Gao, L.W. u, P. Gardner, and P.S. Hall, "Wideband Dual-Polarized Microstrip
Patch Antenna", Electronics Letters, Vol. 37, pp. 1213-14, Sept. 2001.
[46] G. L. Matthaei, L. Young and E. M. T. Jones, Microwave Filters. Impedance-
Matching Networks and Coupling Structures, Artech House , Dedham, 1980.
[47] D. M. Pozar, Microwave Engineering - Second Edition, John Wiley and Sons lnc.,
New York, 1998.

166
Chapter 4 Improving the Efficiency of
Microstrip Patch Antennas

4.1 Introduction

There are many applications where simple, highly efficient printed


antennas mounted on high dielectric constant substrates are required. One
such case is in a mobile communications base station terminal for a
microcellular system. Ideally, the antenna is mounted on the same substrate
used for the active microwave components and perhaps photonic devices if an
optically distributed network is incorporated. Such substrates have a high
dielectric constant (e, > 10) and so, as was discussed in Chapter 2, a
conventional microstrip patch antenna would not be a suitable choice as it will
be inefficient. As mentioned previously, the source of inefficiency is due to
the excitation of surface/leaky waves. Surface wave effects can also manifest
themselves in other detrimental ways, including increasing the cross-
polarization levels and scalping in the radiation patterns of a printed antenna.
These are typically experienced when the surface wave diffracts off the finite
edges of the ground-plane, in a similar vain to the field radiated by a slot
(refer to Chapter 3).

In this chapter we explore microstrip patch antennas and procedures


that have improved efficiency compared to the conventional microstrip
radiator. In particular these case studies minimize the excitation and the
effects of the typically excited surface waves. In Section 4.2 a brief
introduction to surface waves is presented. In Section 4.3 several microstrip
patch antennas are discussed that minimize the excitation of the TMo surface
wave and as a result are very efficient, despite being mounted on electrically
thick, high dielectric constant material. A version that renders itself to
integration with active devices is presented as well. The main shortcomings
of this procedure are resolved by a simple technique in Section 4.4. The
printed antenna presented in this section, not only has high efficiency, but also
a broadband impedance response. Finally, in Section 4.5 high impedance
ground planes, based on Photonic Bandgap structures are presented and
designed. These relatively new structures minimize the excitation of surface
waves over a band of frequencies, thereby reducing the detrimental effects of
surface wave excitation.

R. B. Waterhouse, Microstrip Patch Antennas: A Designer’s Guide


© Springer Science+Business Media New York 2003
4.2 Surface Waves

Before launching into techniques that minimize the effects of surface


waves, it is important to define what these waves are. Surface waves are
modes of propagation supported by a grounded substrate. In antennas, surface
waves spread out in a cylindrical fashion around the excitation point, with
field amplitudes decreasing with distance (r), or more correctly 1/-../r [1].
Surface waves are reflected within the grounded dielectric substrate i.e. when
they meet the ground plane and at the dielectric-to-air boundary. The fields
remain trapped within the dielectric and take up part of the energy of the
signal, thus decreasing the desired signal amplitude and contributing to
deterioration in the antenna efficiency. On an infinitely large ground plane
the surface waves would be evident only as a reduction in radiation efficiency.
However in reality, the ground plane of a printed antenna is always finite in
size. Surface waves/currents propagate until they reach an edge or corner.
When surface waves reach these boundaries they are reflected and diffracted
by the edges as shown in Figure 4.2.1. The diffracted waves provide an
additional contribution to radiation, typically degrading the antenna pattern by
producing ripples in the radiation pattern, raising sidelobe and cross-
polarization levels [1].

Figure 4.2.1 Consequences of Surface Waves

The two types of surface wave modes possible in a microstrip patch


antenna structure are transverse magnetic (TM) and transverse electric (TE)
modes. In TM surface waves, the electric field forms loops that extend
vertically out of the surface and the magnetic field is parallel to the surface,
while in TE surface waves, the electric field is parallel to the surface, and the
magnetic field forms vertical loops out of the surface. Surface wave
excitation occurs on all substrate-based antennas because the lowest TMo
surface wave mode has a zero frequency cut-off, where the cut-off
frequencies for the different surface wave modes are expressed by:

168
n
fc = -4h--'j~;==&flo=~-r&=j.Jr=-=1 (4.1)

where n =0, 2, 4, ... for TM modes, n = 1, 3, 5, ... for TE modes, and


h is the dielectric thickness.

By decreasing the dielectric thickness and permittivity, less energy is


coupled into the surface waves, thus reducing the magnitude of the surface
waves. However a low substrate thickness results in a reduction in antenna
bandwidth and antenna efficiency, while using a low permittivity substrate
increases antenna size.

The quantity used to measure the impact of the excited surface wave
on the antenna performance is the surface wave efficiency defined as:

T)sw =1 - PswIPrad (4.2)

where Psw is the power trapped in the surface wave and Prad is the total
radiated power from the antenna.

4.3 Patches that do not excite TMo surface waves

4.3-1 Introduction

There has been reported a technique to improve the surface wave


efficiency of a printed antenna [2]. This method relies on the principle that if
the overall dimension of the printed antenna is large enough, then the
excitation of the TMo surface wave mode will be greatly reduced. Two novel
circular printed antennas were proposed that weakly excited a surface wave
mode. The resulting configurations improved the surface wave efficiency of
the radiating element considerably, however they were larger and somewhat
more difficult to fabricate than a conventional microstrip patch antenna. One
of the antennas had a core hollowed out below the patch conductor and the
other utilized a shorted annular ring. In this section we present two simple to
fabricate alternatives that rely on the same principal. In fact the second
technique whilst enhancing the efficiency also enhances the bandwidth due to
the 'mutual resonance' created.

169
The principal of operation of these patches is based on the following:
there is some critical radius of the patch that minimizes the excitation of the
TMo surface wave mode:
~TMo Rcrit =X' In =1.8412,5.3314, .... (4.3)

where ~TMo is the propagation of the surface wave mode. Thus if the
radius of the patch is greater than Relit. the TM o surface wave mode will not be
excited.

Before we explore the highly efficient patches, a benchmark radiator


should be considered. Here, the conventional microstrip patch is circular with
a radius of 4.74 mm and is printed on a 1.905 mm substrate with a relative
dielectric constant of 10.4 and a loss tangent of 0.002 was designed and
developed . The patch is fed by a probe with a radius of 0.45 mm placed 1.1
mm from the center of the patch. The antenna is resonant at 5.08 GHz with
an input impedance of 53 n. Figure 4.3.1 shows the measured and predicted
(using a full-wave in-house simulator) return loss of the antenna. As expected
the bandwidth is narrow (1.9 %) and the surface wave efficiency is somewhat
low (66 %) and the gain is 3.8 dBi.
- 0 'r-;u:;;;;g;~~~,-----r-:-=~~~-t
I I ' I
I I ,

~o --- ~--- ~- - -~ - ~-
I I I I I t I

~ ·20 -- - ~ - - -~ - - -~-- I - ~-- -~---~ -- -


II) I I I I ;" I I
~ 40 I, ~, I ' I
- - ', -Theory I
L
,

.... - - -
E I I ,
Q::J ~ o
IL IL IL I
L I
L I
L _
a:: I I I I I I
I I I I I I
I I I I I I
·50 -- -r - - -r --- r- - - ---r---r - - -r - - -
I I I t t I
I I I I I I
. 60 L..-o...-J'----'----'-----'-----'-~--'-~__L.~--'-~....J_~_J
U U U 5 ~ U U U ~
Frequency (GHz)

Figure 4.3.1 Benchmark Circular Micostrip PatchAntenna

As mentioned before, the two designs considered are based on the


concepts given by Equation 4.3. By properly adjusting the radius of the
circular patch, a critical dimension can be found that will greatly reduce the
excitation of the dominant TMo surface wave mode. Unfortunately, this
radius is much larger than the resonant dimension of conventional microstrip
patch antenna. As in [2], the resonant frequency of the element must be
170
increased to take advantage of the surface wave efficiency, as the critical
radius is a function of the substrate material and the frequency considered.
For the substrate and frequency of the benchmark patch, the radius has now
grown to 16.75 rom.

4.3-2 Some simple cases

4.3-2.1 Patch loaded with short circuit

The first design considered uses a probe and a shorting pin oppositely
placed to one another to increase the resonant frequency. This idea is similar
to the concept introduced in [3] (and also discussed in more detail in Chapter
6), which reduced the resonant frequency of a microstrip patch by placing a
shorting pin in close proximity to the probe feed. By positioning the shorting
pin away from the feed, one would expect the opposite affect on the resonant
frequency, simply because in this case the feed is seeing an inductive load.
The element and its dimensions are described in Figure 4.3.2.

Figure 4.3.2 CircularPatch Loaded withShorting Pin (R =16.25 rom, xpJ =5.7 rom, Xp2
= -13.5 rom)

The new geometry has an input impedance of 44 ,Q at resonance (5.1


GHz), as well as a greater surface wave efficiency (89 %) and gain (4.3 dBi).
However, the bandwidth of the patch narrows (0.3 %) as compared to the
conventional patch. Figure 4.3.3 shows the theoretical and measured input
impedance of the antenna. The discrepancy between the two can be attributed
to inaccuracies in the positioning of the shorting pin. Figure 4.3.4 shows the
measured E- and H-plane radiation patterns of the antenna. These patterns are
relatively smooth as is normally the case for these types of patch antennas,
however there is a slight scalping evident in the E-plane which appears to be
due to the presence of the surface wave, diffracting off the finite ground-
171
plane. The cross-polarization level of this more efficient patch is
approximately 15 dB higher than the conventional patch, which is due in part
to the presence of the shorting pin and also the large size of the modified
patch that allows higher order modes to radiate.

Figure 4.3.3 Input Impedance Response of Circular Patch Loaded with Short Circuit
(fstan =4.9 GHz; f slop =5.3 GHz)

II.

Figure 4.3.4 E- and H-plane Co-polar Radiation Patterns of Circular Patch Loaded with
Short Circuit at 5.I GHz

4.3-2.2 Patch loaded with annularring

Another possible antenna that satisfies the above conditions is an


annular ring loaded circular patch. Figure 4.3.5 illustrates a probe-fed circular
microstrip patch antenna that incorporates an annular ring. The patch of
radius R is mounted on a substrate of thickness d and with a dielectric

172
constant Cro The coaxial probe feed of radius ro is located at (pp, (j)p) from the
center of the patch. The annular ring is of inner radius a and outer radius b
and is placed concentrically about the patch antenna.

d E,

Figure 4.3.5 Probe-fed Circular Patch Loaded withAnnular Ring

To accurately predict the electrical characteristics of the modified


circular patch, the spectral domain electric field integral equation technique
was implemented (the same analysis as used for the previously reported high
efficiency antennas). The design considered here is based on the concepts
given in Section 4.3-1. By properly adjusting the radius of a (j)-symmetric
printed antenna, a critical dimension can be found that greatly reduces the
excitation of the dominant TMo surface wave mode. As mentioned before,
unfortunately this radius is much larger than the resonant dimension of a
conventional circular patch: therefore, the resonant frequency of the element
must be increased to take advantage of the surface wave efficiency.

The design procedure for this printed antenna is relatively


straightforward. The driven antenna, namely the circular patch, was designed
for 50 Q impedance match at the desired center frequency without the
presence of the parasitic element, namely, the annular ring. The outer radius
was chosen as in Section 4.3-1. Fine adjustments were then made to the probe
position and inner radius of the annular ring to achieve the desired input
impedance behavior. Also, the surface wave efficiency was examined during
the design process to ensure both high surface wave efficiency and good
impedance match were maintained over the band of interest.

The predicted and measured input impedance behavior of a circular


patch with an annular ring is shown in Figure 4.3.6 (see the figure caption for
dimensions). As can be seen from the results, very good agreement between
experiment and theory was achieved. The predicted and measured 10 dB
173
return loss bandwidths were 3.0 % and 3.7 %, respectively. The enhancement
in bandwidth (compared to the benchmark case) is an artifact of the presence
of the second radiator, namely the annular ring. The coupling between the
two elements improves the impedance bandwidth in a similar way as outlined
in Chapter 3 for any patch antenna that consists of two or more radiators. The
co-polarized patterns in the cardinal planes of the patch with a concentric ring
were measured at 5.07 GHz. Both the measured and predicted gain at this
frequency was 6.7 dBi. The resulting E- and H-plane patterns are quite
similar and have very little ripple in the pattern, as shown in Figure 4.3.7.
Although not shown, the measured peak H-plane cross-polarization level was
always more than 20 dB below the peak co-polar level. Also the calculated
surface wave efficiency of the annular ring loaded design was greater than 82
% over the 10 dB return loss bandwidth.
4.9GHz
~M~~

Figure 4.3.6 Input ImpedanceResponseof Patch Loaded with Annular Ring (a =8.05
rom, b = 16.25 mm, R = 4.74 mm, xp =1 rom)

".
Figure 4.3.7 E- and H-plane RadiationPatterns Patch with Annular Ring at 5.1 GHz

174
Figure 4.3.8 shows the calculated surface wave efficiency of the
benchmark case, the circular patch loaded with a shorting pin and the circular
patch loaded with an annular ring as a function of frequency. It is interesting
to note that the maximum surface wave efficiency does not correspond to the
resonant frequency for any of the designs.

95.-----,-----r---..------.---.,....----,

90

Conventional Patch •••.


Shorted Patch _.
RingLoaded Patch -
70
.......................................................................
65~:----:-':-::-:----:-+:-::-~~-~=----,~:----:--:-l
5.000 5.025 5.050 5.075 5.100 5.125 5.150
Frequency (GHz)

Figure 4.3.8 Comparison of Surface Wave Efficiencies

4.3-3 Multilayered version

4.3-3.1 Introduction

The annular ring coupled to the circular patch presented in the


previous subsection, is an easily manufactured version of the patches that do
not excite surface wave modes concept. However, of all these proposed
efficient antennas, a solution that is easily integrated with active devices has
not been presented. In this sub-section the surface wave suppressant is
located on a dielectric layer above the conventional printed antenna and active
circuitry and therefore allows a truly monolithic module. The multilayered
solution is an extension of that presented in the previous sub-section and
displays similar improved characteristics, namely improved bandwidth,
efficiency and gain. It should be noted that a probe-fed solution is presented
here due to its numerical ease, however the concept can be easily be applied
to an edge-fed antenna as probe and edge-fed patches behave in a similar
manner.

175
4.3-3.2 Design and Results

Fig. 4.3.9 shows a schematic of the multilayered antenna. As


mentioned previously, the annular ring used to suppress the TMo surface
mode is etched on a dielectric layer above the driven patch. For the case
considered here, a probe of radius ro, located at (pp, <!>p) is used to excite the
circular patch of radius R. The patch is etched on a substrate of thickness d t
and with a dielectric constant Erl and the ring is etched on a dielectric layer of
thickness d2 and with a dielectric constant fa located above the patch. To
accurately predict the performance of the printed antenna the spectral domain
integral equation technique incorporated before was used.

y z

Figure 4.3.9 Probe-fed Circular Patch with Parasitically Coupled Annular Ring

The design of the printed antenna is similar to the single layer


version, where the radius of the driven patch is chosen for the appropriate
frequency of operation and the outer dimension of the annular ring is selected
to maximize the surface wave efficiency. The location of the feed and the
inner radius of the annular ring are varied to provide fine control of the
impedance locus and have little effect on the overall efficiency of the antenna.
For the configuration considered here there is an extra degree of freedom,
namely the dielectric layer on which the annular ring is etched. After many
simulations, it was found that for optimum surface wave efficiency, this
dielectric layer should be as thin as possible and have a similar dielectric
constant to that of the substrate. It can be postulated that both these traits are
necessary to ensure maximum coupling of the excited TMo surface wave
mode to the annular ring.

176
To investigate the multilayered configuration, a prototype was
fabricated on similar material used in Section 4.3-2 for operation at
approximately 5.1 GHz. A thin layer of RT-Duriod 6010.5 (e, = lOA and d =
0.254 mm) was used to etch the annular ring that was then adhered to the
substrate. Figure 4.3.10 shows the predicted and measured return loss
behavior of the printed antenna (refer to the figure captions for the
dimensions). As can be seen from these results very good agreement between
theory and experiment was achieved. The predicted 10 dB return loss
bandwidth is 3.0 % compared to a measured value of 3.3 %. Figure 4.3.11
shows the predicted and measured E- and H-plane co-polar radiation patterns
of the multilayered antenna. Once again good agreement between theory and
experiment was achieved. As can be seen from these plots there is very little
ripple in the patterns, similar to the antennas presented in the previous
subsection. The predicted and measured gain of the antenna was 6.5 dBi. A
conventional patch mounted on the same material has a bandwidth of 1.9 %
and a gain of 3.8 dBi. The cross-polarized levels of the multilayered antenna
in each plane were greater than 30 dB below the co-polar patterns for all
angles that is very useful for applications requiring dual or circular
polarization.

-.---\
o ......
....... ....
.......... L--
·5
, ~. -~# ••

n, Theory
• • •• Experiment
"'\ ,,
I'

/:
/"
.
,

-30 1"'1"
-35
~
4.8 4.85 4.9 4.95 5 5.05 5.1 5.15
Frequency (GHz)

Figure 4.3.10 Predicted and Measured Return Loss of Multilayered Annular Ring Loaded
Patch(parameters: £'-. =10.4, tan 0) =0.002, d. = 1.905 mm, Ea =10.4, tan ~ =0.002, d2 =
0.254 mm, R =4.57 mm, Pp = 1.13 mm, <l>p =0, ro =0.45 mm, a =8.5 mm, b = 16.65 mm)

The surface wave efficiency of the multilayered version is greater


than 78 % over the entire 10 dB return loss band. This value is marginally
less than the single layer version (82%) and can be attributed to a slight
reduction in coupling between the TMo surface wave mode generated and the
177
annular ring. The surface wave efficiency of a conventional patch mounted
on this substrate is approximately 66 % over the same frequency band. It
should be noted that for both annular ring configurations, the peak in surface
wave efficiency always occurs before the mutual resonance between the patch
and the annular ring and cannot be easily moved.

Figure 4.3.11 Predicted and Measured E- and H-plane Radiation Patterns of Multilayered
Annular Ring Loaded Patch

4.4 Hi-lo stacked patches

4.4-1 Introduction

In Section 4.3, several highly efficient antennas were presented that


overcome the detrimental surface wave effects. The antennas suppress the
excitation of the TMo surface wave mode by making the size of the radiator
electrically large. Although all cases successfully enhance the surface wave
efficiency, there are two fundamental problems with these configurations.
Firstly, the antennas have narrow impedance bandwidths, making these
solutions unsuitable for most communication systems. Also the overall size
of the antennas is large (greater than 0.5 1..0) and therefore these elements are
unsuited for scanning array applications.

In this section we present a highly efficient patch antenna that is


broadband and can easily be integrated into arrays. The printed antenna
consists of a simple stacked configuration with the lower element mounted on
the high dielectric constant material and the upper patch mounted on foam.
Bandwidths approaching 30% have been achieved using this configuration

178
and importantly the surface wave efficiency over this band is high, greater
than 85 %. Comparisons with conventional patch antennas are also given.

4.4-2 Geometry and Design

Figure 4.4.1 shows a schematic of the proposed stacked


configuration. Here a probe-fed case is considered due to its numerical ease,
however the concept can be easily applied to an edge-fed antenna as probe
and edge-fed patches behave in a similar manner. The lower patch of radius
R1 is etched on the grounded high dielectric constant (ftl) substrate of
thickness d., The second patch is etched on a dielectric layer of thickness dz
and with a dielectric constant Er2 located above the lower patch. In the case
considered here, a probe located at (pp, ~p) and of radius ro is used to excite
the lower patch. To accurately predict the performance of the stacked printed
antenna the spectral domain integral equation technique presented in [4] was
used.

Figure 4.4.1 Probe-fed Hi-lo Stacked Patch

As previously mentioned, we believe this to be the first case in which


a stacked geometry has been proposed consisting of high and low dielectric
constant materials. Previously it had been postulated that to maximize the
impedance bandwidth, low dielectric constant materials should be used, in
particular a combination of foam (Er = 1.07) and teflon fibreglass (e, ::= 2.2)
(refer to chapter 3). It was thought that using such a conglomeration would
ensure the patches are approximately the same size and therefore maximize
the coupling between the two radiators. By doing so a resonant loop with
good impedance characteristics can be achieved. However, a similar
impedance behavior can be obtained when the patches are not of comparable
dimensions, as was shown in the optimization section of Chapter 3 and will
also be shown here.
179
To develop the proposed stacked configuration a simple design
procedure can be adopted. As with most stacked patch geometries the first
step involves designing the lower patch to resonate at the appropriate
frequency without the presence of the parasitically coupled element. Most
designs would then proceed with using an element of similar size to the
bottom radiator for the top patch, however this is not the case here. For the
hi-lo dielectric constant structure, the next step involves designing the top
patch for resonance at the required frequency ignoring the lower element and
the high dielectric constant substrate. Thus the top element is designed as if
the patch was mounted on a thick layer of grounded foam. The simplified
approach tends to give a good starting point for the size of this element as it
can be postulated that the significant dielectric constant difference between
the two layers behaves as a good reflector, or like a ground plane. Slight
adjustments of the relative dimensions of the patches and the feed position
provide the fine tuning mechanisms required to give the desired resonant loop
in the impedance behavior centered at 50 n.

4.4-3 Results

A hi-la stacked configuration was designed for operation at


approximately 5GHz using a grounded dielectric layer of Erl = 10.4 and d, =
1.905 mm and an upper dielectric layer of foam (fr2 = 1.07, d2 =4 mm). The
lower substrate has a similar dielectric constant to material typically used for
MMICs and OBICs. The predicted and measured input impedance behaviors
of the stacked patch are shown in Figure 4.4.2. As can be seen from these
results fairly good agreement was achieved. The slight discrepancy can be
attributed to the tolerances of the materials utilized as well as the thin layer of
dielectric material required to etch the top patch conductor. Here a thin layer
of RT-Duroid 5880 with Er =2.2 and d =0.125 mm was used. The predicted
and measured 10 dB return loss bandwidths are 30 % and 28 %, respectively.
To put these results into prospective, a conventional probe-fed patch mounted
on the high dielectric constant material has a bandwidth of 1.9 %.

Figure 4.4.3 shows the measured E- and H-plane co-polar radiation


patterns of the hi-lo stacked patch antenna. As can be seen from these plots
there is very little ripple in the patterns, similar to the antennas presented in
Section 4.3. The predicted and measured gain of the antenna was 7.5 dBi. A
conventional patch mounted on the high dielectric constant material has a gain
of 3.8 dBi. The cross-polarized levels of the stacked patch in each plane were

180
greater than 20 dB below the co-polar patterns for all angles that is very
useful for applications requiring dual or circular polarization. This will be
addressed in chapters 6 and 7.

Figure 4.4.2 Input Impedance Behavior of Hi-lo Patch (parameters: €r. = 10.4, tan Sl =
0.002, d 1 =1.905 rom, Ea = 1.07, tan ~ =0.002, d2 =4.0 rom, R 1 =4.74 rom, Pp =3.49 rom, %
=0, ro =0.325 rom, R2 =10.8 rom)

II.

Figure 4.4.3 Measured E- and H-plane Radiation Patterns of Hi-lo Stacked Patch

A plot of the theoretical surface efficiency is shown in Figure 4.4.4 .


As can be seen from this graph, the surface wave efficiency of the stacked
patch is greater than 85 % over the entire 10 dB return loss bandwidth. This
compares to approximately 66 % for the conventional patch and greater than
82 % for the efficient configuration presented in Section 4.3. For the hi-lo
181
stacked patch, it appears that the top patch draws the surface wave power
excited from the lower element. Importantly, the overall size of the stacked
patch is significantly less than the cases in Section 4.3 and therefore can
easily be integrated into scanning arrays. It is interesting to note that the
efficiency presented here across the 10 dB bandwidth, is comparable to that
achieved for a stacked patch configuration consisting of a combination of
materials similar to those used in Chapter 3.

---
95

90
.>~
~85
1'\
c
~80
V \
IE
W 75
\
~ \
~70
365
\
\
-.
~
J:60
55

50
4 4.5 5 5.5 8 8.5 7
Frequency (GHz)

Figure 4.4.4 Predicted Surface Wave Efficiency of Hi-lo Stacked Patch

4.5 Photonic Bandgap Structures


4.5-1 Overview of Photonic Bandgaps

4.5-1.1 Introduction

Photonic Bandgaps (PBGs) are periodic dielectric, metallic or


metallodielectric materials also known as photonic crystals (PCs). These
artificially engineered periodic materials forbid the propagation of
electromagnetic waves within a designed frequency range referred to as a
bandgap. The frequency bandgap depends primarily on the lattice constant
(spatial period) and the dielectric permittivity of the constituent materials.

Initially, photonic bandgaps were investigated for the control of the


optical properties of materials, i.e. controlling the emission and propagation of
light. However, with the fabrication of the Yablonovite structure [5], it was
demonstrated that photonic bandgaps are scaleable with frequency. This was
the first structure where the photonic bandgap was shifted from the optical
frequency down into the microwave frequency.

182
These artificial composite structures (photonic bandgaps) have
attracted a lot of attention due to their unique and favorable properties and
their suitability to a variety of applications, namely: antenna gain
enhancement and radiation pattern shaping applications [6]-[8], filtering
applications including high Q filters, frequency selective surfaces (FSS) and
space filters, resonant cavities [9] and waveguide structures [10, 11].

4.5-1.2 High Impedance Ground Plane

The high impedance ground plane (mGP) is easily identified from


other metallodielectric photonic bandgaps by the 'mushroom' like design,
which consists of metallic elements located at the surface of a grounded
dielectric slab with each element connected to the ground plane with a via.
The elements are mainly in the form of squares or hexagons and the structure
is either a two layer or three layer design as shown in Figure 4.5.1a and Figure
4.5.1b respectively, but are not limited to these configurations. The mGP is
classified as a 2D structure and the fundamentals behind both two-layer and
three-layer designs are the same. They rely on the increased capacitance
created between metallic elements to further push down the electromagnetic
bandgap frequency spectrum, either from the fringing capacitance between
adjacent elements in the two-layer design or the capacitive loading between
the metallic elements for the three-layer design. With the inclusion of these
metallic elements the constraint of linking spatial period and electromagnetic
frequency has been relaxed, creating versatile photonic bandgap designs.

:r-r-r:r-r-r-r-r side view

lop view
(a)

183
I I
side view

top view

(b)

Figure 4.5.1 Sievenpiper's High Impedance GroundPlane [12]: (a) Two-layer design,
(b) Three-layerdesign

The HIGP exhibits a high surface impedance to electromagnetic


waves within its frequency stop-band compared to the perfect electrical
conductor (PEe) ground plane which exhibits low impedance for all
frequencies. T4e HIGP is thus defined by two primary characteristics. First,
the high impedance surface creates image currents and reflections that are in-
phase with a source rather than out-of-phase as for a PEC surface. Second,
the high impedance surface suppresses surface waves that would propagate
freely on a PEC surface [12].

These characteristics combined with the fact that it is currently the


most size efficient/compact photonic bandgap material, has made the HIGP
the most feasible replacement for the conventional (PEe) ground plane in
antennas which operate in the low microwave region and need to be
electrically small.

184
4.5-2 High Impedance Ground Plane Design

4.5-2.1 Introduction

As already described in Section 4.2, surface wave excitation occurs


on most substrate-based printed antennas because the lowest order surface
wave mode (TMo) has a zero cut-off frequency. These waves become
stronger as the substrate thickness and relative dielectric constant are
increased. They propagate along/near the surface of the substrate and are
radiated into free space at the truncation of the dielectric substrate. The
radiated surface waves can significantly distort the radiation pattern hence
degrade the performance of the antenna.

Several schemes can be found in the literature to suppress these


surface waves including the use of photonic band-gap (PBG) structures (for
example, [13], [14], [10], [11]). Probably the most effective and compact
approach thus far in the microwave domain makes use of periodic structures
composed of metallic pads connected to the ground plane with vias. This
novel PBG structure, known as a high impedance ground plane (mGP), was
first proposed by Sievenpiper [12].

In this section an outline of the methodology used to design a two


layer mGP is given. The structure is simulated using Agilent's High
Frequency Structure Simulation (HFSS) to confirm the band-gap frequency.
Surface wave measurement techniques developed by Sievenpiper [12] are
employed to verify the band-gap of the designed structure.

4.5-2.2 High Impedance Ground Plane Design

The high impedance electromagnetic ground plane can be realized as


a two-layer or three-layer structure of various element shapes and lattice
configurations. The fabrication technique of the three-layer structure as
described by Sievenpiper [12] makes use of electroplated through vias
together with chemical etching techniques. This method requires special
facilities and can be quite expensive. It was therefore decided that a two layer
structure be used as shown in Figure 4.5.2. The complex fabrication can be
replaced by simply soldering metal vias to the bottom metal layer and metal
elements on the surface of the substrate. The square lattice design was chosen

185
because of the ease of integrating this high impedance ground plane with a
square or rectangular microstrip patch antenna.

~ COnduclin g Vias
"'rr~T----T""'--T---U_ Metal Layer - - - __
Side Vitw

Figure 4.5.2 Two Layer, SquarePattern,HighImpedance GroundPlane

The effective medium model is a common tool used in designing and


characterizing PBG structures. However, rather than characterizing the
photonic band-gap structure, the surface impedance model can be used to
predict the band-gap position in the frequency spectrum and its width. The
surface impedance model is only useful when the physical size of a PBG unit
cell is much smaller than the wavelength of the desired bandgap frequency
[12].
c

Figure 4.5.3 Equivalent LC CircuitModel

Since the unit cells are small compared to the wavelength, the
electromagnetic properties can be described using lumped circuits.
Sievenpiper models his design as a parallel resonant LC circuit shown in
Figure 4.5.2. The capacitance in this two-layer model is dominated by the
fringing electric field between the adjacent elements, as shown in Figure
4.5.3. The fringing capacitance C is given by the expression:

186
C -- w(£! + £2)
Il
Cosh-
g
-l(a) (4.4)

where w is the edge of an element, £] is the permittivity of a vacuum,


£2 is the permittivity of the substrate, g is the gap between adjacent elements ,
and a is the center-to-center spacing between conducting vias. The
inductance L originates from the current loops within the structure, as shown
in Figure 4.5.3 and is dependent primarily on the substrate material:
L =Jlrflot (4.5)
where Pr is the permeability of the dielectric material (typically
assumed equal to 1.0), Po is the permeability of a vacuum, and t is the
substrate thickness.

The model is assigned a surface impedance Z, equal to the impedance


of the parallel resonant LC circuit. Here C is the capacitance in Flm, L is the
inductance in Him and (J) is the angular frequency in rad/sec.
Z = jwL (4.6)
l-w 2LC

At resonance the impedance is very high and corresponds to the


center frequency of the band-gap. In this region the currents on the surface
radiate efficiently, and the structure can suppress both TM and TE surface
waves. It also reflects external electromagnetic waves without phase reversal
[13]. The resonant frequency to; is found by:
1
(4.7)
0J0 = .JLC
while the bandgap bandwidth ~w is obtained by the relationship:

#
--a;-fI
!J.OJ _
(4.8)

From Equation 4.8 the upper and lower frequency limits of the bandgap can
be predicted.

flow = _1 (Wo- I1w] (4.9)


21l 2
187
thigh = _l_(mo
21l
+ /1m)
2
(4.10)

Based on this surface impedance model, an approximation of the


physical dimension for a desired center frequency can be found. Given that
the same substrate for the high impedance ground plane is to be used for the
shorted patch antenna, a substrate thickness t = 9.51 mm and dielectric
constant e; = 2.2 have been chosen. Figure 4.5.4 plots the equivalent circuit
resonant frequency (fo) as a function of element width (w) and gap between
elements (g).
5
. \\
,.~
.
\
\
,~
~
, \\
•...... g s 0.25rnrn
-
\\
~
· · ··g.O.5mm
-g.1.Onvn
,.
~
,
\\
_.- g-2.Omm

. -, -. •

.<,..."
\
\

'.
'.-, ' . .. ~ .
.. .. ",
"

'. '. ", ~ ". ". ....


::-
........ ~ r;:.:
........ .......
-,,,,"

... ... ....... s::.::


~
..-...... ..........
...•...
1
o 2 4 6 8 10 12 14 16 18 20
Element Width (mm)

Figure 4.5.4 HIOP Resonant Frequency (Surface Impedance Model)

The surface impedance model is a simple tool in determining the


physical dimensions for a desired bandgap frequency. For a desired bandgap
frequency of approximately 2.5 GHz, many solutions exist. A compromise
between ease of manufacturing, size and bandgap bandwidth provides the
most viable solution. Therefore an element width w = 9 mm and gap
separation g = 1 mm was chosen. With a calculated C = 0.33 pF and L =
11.95 nH the resultant resonant frequency is 2.56 GHz and using Equation 4.9
and Equation 4.10 the calculated lower and upper frequency bandwidth are
1.89 GHz and 3.15 GHz, respectively.

One disadvantage of using the surface impedance model is that it does


not take into consideration the radius of the conducting vias connecting the
metal elements to the continuous bottom metallic layer. For manufacturing
purposes a practical via radius r ov of 0.5 nun was selected and to further
confirm the bandgap using the above-mentioned dimensions, the structure
188
was modeled using HFSS. To reduce the computational intensity, the
structure was divided into a 5xl unit cell as shown in Figure 4.5.5 and was
imaged infinitely in width by using a PMC (Perfect Magnetic Conductor)
boundary.
Output (port 2)

3D vrew Input (port 1)

Figure 4.5.5 HFSS simulation 5xl unit cell for HIGP <Parameters: s, =2.2, d =9.51
rnm, W = 9 mm, g = I mm, rOY = O.5mm)

4.5-2.2 High Impedance Ground Plane Results and Discussion

5xl unit cell in Figure 4.5.5 for the HIOP structure was simulated
and the S21 transmission results plotted in Figure 4.5.6. A reference
simulation is also included in this figure, that is, a grounded dielectric slab
without the HIOP. As can be seen from the results the centre frequency of the
HIOP bandgap closely corresponds to the resonant frequency calculated using
the surface impedance model (refer to Figure 4.5.4). Furthermore the
predicted bandgap from 1.89 OHz to 3.15 OHz using the surface impedance
model compares well to the simulation results shown in Figure 4.5.6.

The suppression of surface wave modes can be detected in the


transmission between two identical antennas positioned near the surface. TM
and TE modes can be distinguished by varying the polarisation of the
antennas. The practical method of measuring surface waves in the microwave
frequency band, developed by Sievenpiper [12], was performed.

As mentioned previously, the purpose of this high impedance ground


plane is to replace the PEe ground plane in small microstrip antennas.
189
Therefore the mGP needs to be small since the objective is to improve the
performance of these small antennas, without increasing the overall size.
Rather than testing a large mGP, it is more beneficial to determine the
performance of a structure approximately the size of the PEe ground plane it
is to replace. For this purpose a 5x5 design as shown in Figure 4.5.1 of
overall dimensions 49mm x 49mm was fabricated. The effectiveness of
suppressing TM and TE waves were measured and the bandgap verified.

o .......... ............
~--
-5 ..".- I-"""

iii
:!:!. ·10
o
e
V / I...
~ Reference I
u; -HIGP
~ ·15 I
In
e
l!
I- .20
N
~
/V
I
<II
-25

-30
1 2 3 4 5 6
Frequency (GHz)

Figure 4.5.6 HFSS simulated band-gap result

In TM surface waves, the electric field forms loops that extend


vertically out of the surface. TM waves can be measured using a pair of small
monopole antennas oriented normally with respect to the surface [12], as
shown in Figure 4.5.7. The vertical electric field of the probe couples to the
vertical electric field of the TM surface waves.

Figure 4.5.7 TM surface wave configuration [12]

190
The results are shown below in Figure 4.5.8. A reference
measurement on a PEC was performed, to characterise the behaviour of the
TM probes and the environment the measurements were performed in. The
surface of the PEC was the same size as the mGP under test. A steep
decrease in transmission at 2.2 GHz signifies the TM band edge.
-15
...... #
.,- .#.,
-20 "
\L 1
f'\-...I r-\'
r
1.

.' . ~
-25

i ·30 ,"
•• ...1\ u \ rV\V
g -35
iii
A }/ \ r/v
~ -40

sc: -45 ~
'"
_ -50
~.,I" '- 6' I

I;j
-55
~ 1- •_. Reference ~
I TM Measurement
-60

-65
1 1.5 2 2.5 3 3.5 4
Frequency (GHz)

Figure 4.5.8 TM surface wavemeasurement results

In TE surface waves, the electric field is parallel to the surface, and


the magnetic field forms vertical loops that arc out of the surface. They can
be measured with the same small monopole antennas as the TM surface wave
measurement, only positioned parallel to the surface. The horizontal electric
field of the antenna couples to the horizontal electric field of the TE waves.
However, using monopole antennas for TE measurements complicates
measurements because of the strong cross-coupling to TM waves. An
alternative antenna for improved TE wave measurement is the small loop
antenna [12] shown in Figure 4.5.9. The plane of the loop is positioned
parallel to the surface, creating a vertical magnetic field. The vertical
magnetic field of the loop couples to the vertical magnetic field of the TE
surface waves.

Again a reference measurement on a PEC was performed, to


characterise the behaviour of the TE probes and the environment the
measurements were performed in. The surface of the PEC was the same size
as the mGP under test. From Figure 4.5.10 it is evident that there is a
suppression of surface waves from approximately 2.2 GHz to 3 GHz with a
191
centre frequency of 2.6 GHz. The steep rise at 3 GHz signifies the TE band
edge. However the TM band edge is also visible due to the probe cross-
coupling to TM waves [15] at 2.2 GHz. The suppression of surface waves is
greater than 10 dB within the bandgap. Outside the bandgap the transmission
of surface waves is significantly greater. The TE measurements curve
characteristics resemble the simulated results in Figure 4.5.6. However, the
measurement results show a slight reduction in bandwidth in the lower
frequency range than the simulated results.

Figure 4.5.9 TE Surface WaveConfiguration [12]

-30 r - - - , -- - ....----,-----,------,r-------,

-40 I----+----f----+----+---jl---~

iii' -50 1----+---jf'lHW~~,..,....,....,?__fjAf'o.r"7'CJhr-...."._/I


~
c
~ -60 t - - --f----:::#..W- t --t-- - - t t- - --ir-H---i
1/1
E
~ -70 Jr-F\uU'~~/--+--\\-_Hfl--_lk_t+---jI---L-___i

...I!
N -80 OOU-J' -+-- - + -- --I-+--4fH-- - - jI--- ___i
(J)

-90 1--- - +-- - +-- - -+-1

1.5 2 2.5 3 3.5 4


Frequency (GHz)

Figure 4.5.10 TE Surface WaveMeasurement Results

The measurement of TE and TM surface waves on such a small


structure proved to be a difficult and sensitive task. A slight shift in the probe
position caused transmission variations of approximately 5 dB. The small
192
separation, hence strong direct and cross polarization coupling between
probes attributed mainly to the measurement variation. Multi-path effects
were also present which is the cause of the narrow band peaks and troughs. In
chapter 5 we will apply this developed map to a microstrip patch
configuration.

4.6 Summary

In this chapter we presented techniques used to improve the surface


wave efficiency of microstrip patch antennas. The concept of surface wave
modes in a microstrip patch antenna were discussed and the concept of patch
antennas that do not excite the TMO surface wave mode were presented. An
annular ring loaded probe-fed circular patch antenna based on this concept
was designed, fabricated and measured in Section 4.3. The parasitically
coupled microstrip patch has a greater bandwidth, surface wave efficiency and
gain that a conventional patch constructed on the same material. Additionally
the antenna is no more difficult to build than a standard microstrip patch
antenna.

A multilayered, highly efficient printed antenna was presented in


Section 4.3. This antenna incorporates an annular ring etched on a
parasitically coupled layer. The presence of the ring significantly suppresses
the excitation of surface waves. The multilayered antenna has similar
radiation and impedance characteristics as the single layered version with the
additional feature that it can be easily integrated with active devices.

In Section 4.4 a highly efficient, broadband printed antenna suitable


for MMICs and amcs was presented. The antenna utilizes a simple stacked
configuration with the lower element etched on the high dielectric constant
material used to grow the microwave or photonic devices. Bandwidths
approaching 30 % and surface wave efficiencies greater than 85 % across this
band were achieved. Due to its relatively small size, this simple printed
antenna can be easily integrated into scanning arrays, which will be shown in
subsequent chapters.

In Section 4.5 a high impedance ground plane was designed. The


design methodology was presented, and the structure parameters determined
from the surface impedance model for the desired frequency bandgap. The
structure was then modeled using HFSS. The simulated results confmned the
bandgap frequency and the bandwidth closely matched the surface impedance
model predictions. Finally, the structure was fabricated and tested. TE and
193
TM measurements were performed. The results taken the bandgap of the
structure was clearly evident and good correlation between simulation and
measured results were achieved.

Common to all the procedures presented in this chapter is a smooth


radiation pattern and minimal cross-polarization levels. These are key
attributes when considering the design of circularly polarized antennas and
some cases will be presented in subsequent chapters that take advantage of
this. Also the concepts presented in this chapter will be applied to two later
chapters when considering integration of patch antennas with active
microwave components and photonic devices and also when trying to
improve the radiation properties of a small printed antenna for use on a
mobile communication handsetterminal.

4.7 Bibliography

[1] J.F. ZUrcher and F.E. Gardiol, BroadbandPatchAntennas,Artech House Inc, 1995.
[2] D. R. Jackson, J. T. Williams, A. K. Bhattacharyya, R. L. Smith, S. J. Buchheitt and
S. A. Long, "Microstrip Patch Designs That Do Not Excite Surface Waves," IEEE
Trans. Antennas Propagat., vol. AP-41, pp. 1026-1037, August 1993.
[3] R. B. Waterhouse, "Small Microstrip Patch Antenna," Electronics Letters, vol. 31,
pp. 604 - 605, Apr. 1995.
[4) 1. T. Aberle, D. M. Pozar and J. Manges, "Phased Arrays of Probe-fed Stacked
Microstrip Patches", IEEE Trans. Ant. & Prop., vol. AP-42, pp. 920-927, July 1994.
[5) T.K. Lo, Y. Hwang, E.K.W. Lam, B. Lee, "Miniature Aperture-Coupled Microstrip
Antenna of Very High Permittivity," Electron. Lett., vol. 33, pp. 9-10, Jan. 1997.
[6] J.D. Shumpert, W.J. Chappell and L.P.B. Katehi, "Parallel-Plate Mode Reduction in
Conductor-Backed Slots Using Electromagnetic Bandgap Substrates," IEEE Trans.
Microwave Theory and Tech., vol. 47, pp. 2099-2104, Nov. 1999.
[7] M. Thevenot, C. Cheype, A. Reineix and B. Jecko, "Directive Photonic-Bandgap
Antennas," IEEE Trans. Microwave Theory and Tech., vol. 47, pp. 2115-2122, Nov.
1999.
[8] T.H. Liu, W.X. Zhang, M. Zhang and K.F. Tsang, "Low Spiral Antenna with PBG
Substrate," Electron. Lett., vol. 36, pp. 779-780, Apr. 2000.
[9) M.M. Beaky, lB. Burk, H.O. Everitt, M.A. Haider and S. Venakides, "Two-
Dimensional Photonic Crystal Fabry-Perot Resonators with Lossy Dielectrics," IEEE
Trans. Microwave Theory and Tech., vol. 47, pp. 2085-2091, Nov. 1999.
[10) F. Gadot, A. Ammouche, A. de Lustrac, A. Chelnokov,F. Bouillault, P. Crozat, 1.M.
Lourtioz, "Photonic Band Gap Materials for Devices in the Microwave Domain,"
IEEE Trans. Magnetics; vol. 34, pp. 3028-3031, Sep. 1998.
[11) F. Yang, K. Ma, Y. Qian and T. Itoh, "A Novel TEM Waveguide Using Uniplanar
Compact Photonic-Bandgap (UC-PBG) Structure," IEEE Trans. Microwave Theory
and Tech., vol. 47, pp. 2092-2098, Nov. 1999.
[12] D. F. Sievenpiper, "High-Impedance Electromagnetic Surfaces," Ph.D. dissertation,
Dept. Elect. Eng., University of Califomia at Los Angeles, Los Angeles, CA, 1999.

194
[13] D. F. Sievenpiper, L. Zhang, R. F. Jimenez Broas, N. G. Alexopolous, E.
Yablonovitch, "High-Impedance Electromagnetic Surfaces with a Forbidden
Frequency Band," IEEE Trans. Microwave Theory and Tech., vol. 47, pp. 2059-
2074, Nov. 1999.
[14] R. Gonzalo, P. de Maagt, M. Sorolla, "Enhanced Patch-Antenna Performance by
Suppressing Surface Waves Using Photonic-Bandgap Substrates," IEEE Trans.
Microwave Theory Tech., vol. 47, pp. 2131-2138, Nov. 1999.
[I5] R. F. Jimenez Broas, D. F. Sievenpiper, E. Yablonovitch, "A High-Impedance
Ground Plane Applied to a Cellphone Handset Geometry," IEEE Trans. Microwave
Theory and Tech., vol. 49, pp. 1262-1265, Jul. 2001.

195
Chapter 5 Small Microstrip Patch Antennas

5.1 Introduction

As stated in Chapter 1, one of the many advantages of microstrip


patch technology over its competitors is its low profile and hence small
volume. Another key advantage of this printed antenna is the relative ease in
which it can be connected to the feed network, as was highlighted in Chapter
2. For these reasons antenna design engineers deduced that microstrip patch
antennas could be utilized for applications requiring where there was very
limited space to mount the antenna. One such global application is for
wireless communication handset terminals.

In recent years, the market for wireless communication handheld


devices has grown exponentially with the rapid development of new products
and services. Current devices on the market, cordless/cellular telephones,
wireless Personal Digital Assistant (PDA), and Global Positioning Systems
(GPS)/Satellite receivers utilize frequency spectra from 400 MHz to 3 GHz
for communication [1 - 6]. Importantly it appears that the technology is being
lead be today's savvy consumers demands for handsets to be low cost,
compact in size, and with a look that is aesthetically pleasing. Therefore,
much recent research effort has been devoted to physically reducing the size
of the wireless handset.

Technical advances in integrated circuit technology have lead to a


continual decrease in the overall size of radio transceivers for wireless mobile
handsets. As the overall handset size decreases, the antenna size must
decrease too. Of course however, there are fundamental limits in terms of the
trade-off in radiation and other characteristics and significantly reducing the
physical area occupied by the antenna. Progress to date has been somewhat
limited because in many applications bandwidths in the order of 10% are
required to cover allocated frequency bands and size reduction is often
obtained at the expense of bandwidth [7].

It is increasing to note the fundamental limits of the impedance


bandwidth (or Q) and gain of an antenna versus its size and the consequence
on the overall system performance. As the volume of an antenna decreases,
so does its bandwidth and gain . In a communication link such as a mobile
network, continual coverage is very important and so with a cellular base

R. B. Waterhouse, Microstrip Patch Antennas: A Designer’s Guide


© Springer Science+Business Media New York 2003
station configuration, it is essential that the mobile terminal have a very broad
radiation pattern to minimize the dynamic range requirements on both
receivers and possible outages. Therefore, in cellular networks where there is
minimal overlap of the base station coverage, a low gain handset antenna is
required. So having a small antenna actually facilities this requirement. Also,
as is the case for any communication link, any shortcomings of one end of the
link can be compensated for at the other end: large powers can be transmitted
from the base station and also sensitive receivers can be used there. Once
again this allows for a less than optimal performing antenna at the handset
terminal. Having said that, that a broad radiation response is necessary, some
gain from the handset antenna is necessary. Also, reasonable matching at the
operating frequencies is required as both these parameters will affect the
overall link performance, with the gain characteristic having a more dramatic
impact on the budget. One reported argument for requiring a good impedance
match at the antenna port is the case for minimizing reflections back into the
transmit amplifier, however, more commercial amplifiers for these systems
are buffered and so this really isn't too much of a concern.

5.1-1 History of antennas for wireless systems

The antennas for mobile units are fundamental parts of the overall
mobile system. The evolution of mobile systems requires portable antenna
elements that are small, low profile, lightweight and relatively inexpensive to
manufacture. As mentioned before, some of the desired features for mobile
antennas include omnidirectional radiation pattern in the horizontal plane,
broad bandwidth to accommodate spread-spectrum services, and ease of
construction. However, the fundamental problem is the size of the antenna
tends to be large for low microwave frequencies, typically less than 3 GHz.
As stated earlier, as the volume of the antenna reduced, the performance of
the antenna in terms of impedance and radiation patterns is sacrificed to
achieve reduced volume [7].

Antennas for mobile handsets have typically been versions of wire


antennas [8 - 10] namely, monopoles, sleeves and helices. These antennas
have been used for the past few decades since the first analog mobile systems
in the 1980s. At present, wire antennas are still being used in some mobile
handsets because of their proven performance. However, the major
disadvantages with utilizing wire antennas in handsets are that, the antennas
can: easily be broken; difficult to manufacture; and when used for diversity to
combat multipath effects, the overall size of the handsets increases
substantially.

198
The dipole antenna is the simplest form of wire antenna. The basic
dipole antenna structure is a Y2A center-fed linear cylindrical antenna and it is
this unwieldly dimension that has seen its restricted use. Thick dipoles are
considered broadband while thin dipoles are more narrowband. For very thin
dipoles, small perturbations in the operating frequency will result in large
changes in its operational behavior. The bandwidth can be increased for thin
dipoles if the length to diameter ratio of the dipole is decreased.

To date, the !,4A monopole antenna is probably the most common


mobile antenna. This antenna has the simplest structure and normally employs
a flexible antenna element. The radiation pattern and bandwidth
characteristics of the antenna are the same as a Y2A dipole antenna when the
ground plane is an infinitely large perfect conductor, due to the effect of an
image of the element formed by the ground plane. The input impedance is
typically half that of a dipole antenna. Theoretically, the directivity is 3 dB
larger than that of a dipole antenna because the radiation power is radiated
only to the upper half space of the ground. When mounted on a handset, the
size of the ground plane is small and the direction of maximum radiation tilts
somewhat upwards from the horizontal plane. Hence, the effective gain of the
antenna is usually lower than that of a half-wavelength dipole antenna.

The sleeve dipole antenna is another version of a wire antenna. The


center conductor of a coaxial cable is connected to an element whose length is
a !,4A and the outer conductor of the coaxial cable is connected to a cylindrical
skirt whose length is lAA. The coaxial cylindrical skirt behaves as a lAA choke,
and thus most of the current does not leak into the outer surface of the coaxial
cable. Adding chokes on the lower part of the coaxial cable is usually done to
improve the radiation pattern by further suppressing the current leakage from
the skirt. As a result, this antenna has almost the same characteristics as a Y2A
dipole antenna.

The helical antenna is another basic electromagnetic radiator. The


antenna consists of a conducting wire wound in the form of a screw thread
forming a helix. The helix is normally used with a ground plane with the
diameter of the ground plane typically %A. The antenna operates in many
modes; however, the two principal ones are the normal (broadside) and the
axial (endfire) modes. Operating in the normal mode of the helix gives an
antenna solution that is typically small in dimension. The axial mode is
usually used for applications requiring circular polarization over a wider
bandwidth and in this mode of operation the helix is more efficient, however
it is also larger in size. Helices are usually used for applications of satellites,
namely GPS and Mobile Satellite Communications.
199
To alleviate the problems and drawbacks encountered when wire
antennas are used for mobile handsets, low profile antennas such as the Planar
Inverted-F antenna (PIFA [11-14]) and microstrip antennas [15-17] are
possible candidates due to their compactness and robustness. Planar Inverted-
F antennas are derived from folded monopoles and consist of a planar radiator
element, a feeding point and a shorting pin located at the edge of the radiator
element. Impedance matching can be obtained by varying the feedpoint and
the bandwidth increases in proportion to the height of the antenna. A
performance analysis of a PIFA antenna mounted on a mobile handset has
been presented in [12]. Slots [13] and capacitive loading [14] have also been
implemented in the PIFA to reduce the size of the antenna, however, the
resulting solutions are still considered large in the low microwave frequency
spectrum.

As mentioned before, in mobile and wireless communication where


size, weight, cost and aesthetically pleasing structures are constraints, low
profile antennas are required. To meet the requirements of mobile and
wireless communication systems, microstrip antennas are possible candidates
due to their key advantages. However, their large physical size at low
microwave frequencies (1 - 3 GHz) has deterred the use of these antennas.
Techniques in reducing the size of the printed conductor have been
undertaken in an attempt to allow the antenna to be incorporated into a mobile
handset.

Several methods have been reported in the literature to reduce the size
of the printed conductor of a microstrip patch. One such method of reduction
is to use high dielectric constant material [18]. This is a logical extension of
what was presented in chapter 2 on how to reduce the physical size of the
patch. The antenna in [18] was mounted on a high dielectric material and
loaded with superstrate material (ft. ~ 38). Only poor efficiency due to surface
wave excitation and narrow bandwidths have been achieved so far with this
form of patch. The availability of low cost and low loss high dielectric
constant material is another issue.

A more common approach adapted to decrease the size of the patch


conductor is to short-circuit the edge of the microstrip patch with a shorting
wall/plane [19 - 20], known in microstrip .patch antenna technology as the
quarter-wave patch. Using this procedure can yield reasonable bandwidths,
however, at low microwave frequencies, the conductor size is still too large.
Short-circuiting the edge of the microstrip antenna with a series of shorting
posts can also be used to reduce to size of printed conductor. The number of
shorting posts used determines the degree of size reduction. The most
200
significant reduction is achieved when only a single shorting post is used and
it is located in close proximity to the feed point [21]. Antenna reduction by a
factor of three or so was reported albeit at the expense of reduced bandwidth.
This last method appears to be very promising.

Replacing the shorting post with a low resistance chip resistor can
also demonstrate reduced antenna size [22]. A consequence of this action is
that the antenna bandwidth is also enhanced. However, the problem with
utilizing a chip resistor is that the efficiency and gain of the antenna are
dramatically reduced due to the power dissipated in the resistor that
significantly reduces the power budget.

Incorporating slots in a microstrip antenna reduces the resonant


frequency of the antenna. Hence reduces the overall size of the antenna. It has
been shown in [23] that utilizing this technique provides a narrow bandwidth,
in this case less than 3%. Although size reduction has been achieved using
this technique, the reduction achieved has not been significant compared to
utilizing a single shorting post. Also, the bandwidth reported was narrow.

Having addressed the merits and drawbacks on the above methods to


reduce the size of the printed conductor, it can be concluded that utilizing a
single shorting post in close proximity to the feed location resulted in a
significant reduction in antenna real estate. Therefore, in the next section a
more detailed study will be presented. Included here is a parameter study of
the variables associated with the shorted patch and the effects they have on
the resonant frequency of the antenna. From this study, an understanding of
the behavior of shorted patch antennas can be achieved such that a design
methodology can be proposed.

It should be noted that throughout this chapter the term: shorted patch
is used to describe a patch incorporating several shorting posts. It does not
refer to the classical quarter-wave patch antenna where an edge of the radiator
is terminated with a short circuit plane.

This chapter is organized as follows: Section 5.2 presents an overview


of shorted patches, including design trends and methods for easing the
manufacturing tolerances; Section 5.3 discusses techniques that can further
reduce the size of the patch conductor on a shorted patch; Section 5.4 presents
a winged shorted patch and its performance; Section 5.5 discusses the
concepts of shorted printed spirals to further reduce the size of the printed
conductor; Section 5.6 then summarizes some methods that can alleviate some
of the issues of utilizing small printed antennas. In this section techniques are
presented that enhance the bandwidth of the shorted patch and reduce the
201
cross-polarization levels. Circularly polarized printed antennas incorporating
shorted patches are also summarized. Finally the impact of incorporating a
shorted patch in a handset on Specific Absorption Rate is presented in Section
5.7.

5.2 Shorted Microstrip Patches

5.2-1 Introduction

A typical shorted patch configuration utilizing a circular patch and a


probe feed is shown in Figure 5.2.1a. Here a shorting pin is located in close
proximity to the feed pin resulting in a significant reduction in overall patch
size [21]. The impedance behavior of the shorted patch radiator can be
qualitatively explained in terms of simple circuit theory. The typical tank
RLC network in series with an inductor used to describe a probe fed patch is
here loaded with a series LC circuit, as shown in Figure 5.2.1b. The LC
combination represents the effect of the shorting pin on the input impedance
of the antenna. When the shorting pin is located near the feeding pin, a strong
capacitive coupling between the feed and short occurs. The capacitive nature
is similar to that of two parallel wires [24] and is strongly dependent on the
dielectric constant of the material, the closeness and the radii of the pins. This
capacitive effect counters the usual inductive nature of a patch antenna below
resonance. Thus, depending on the amount of capacitive coupling, the size of
the shorted radiator can be dramatically reduced. As the shorting pin is
moved away from the feed, the capacitive coupling decreases and the feed
sees the shorting pin as an inductive reactance, similar in form to that
typically used to represent the effect of the feed itself. The inductance
increases as the distance between the pins increases [24]. This inductive
effect raises the resonant frequency of the shorted configuration. Hence at a
fixed frequency the modified patch size can be increased or decreased,
depending on the distance of the shorting pin from the feed. Of course this
over simplified approach ignores coupling between the fields from the edges
of the patch and the pins. As shown later in this section, all of these relative
dimensions of the printed antenna play a role in determining the impedance
behavior of the radiator.

The organization of this section is as follows: Section 5.2.2 presents


the impedance and radiation characteristics of a shorted circular patch
incorporating a single shorting post. The antenna is mounted on a relatively
thick substrate with a low dielectric constant to enhance the bandwidth. A
comparison between theory and experiment is provided; Section 5.2.3 gives a
brief comparison between circular and rectangular shorted patches . Also in

202
this section a parameter study showing the effects of the height of the
substrate, the dielectric constant of the material, the size of the feed and
shorting pins as well as the effect of a cover-layer on the electrical
performance and manufacturing ease of the shorted patch is provided. From
this study, valuable insight into the optimum design of shorted patches is
achieved; Section5.2.4 addresses the problemof the strong dependence of the
impedance behaviorof the shorted patch on the close proximity of the feed to
the pin. Here several simple techniques are summarized which overcome this
short fall. The advantages and disadvantages of each method are discussed in
this section.
y
y

x f ..
w
(Xp.Yp)
x

(xl'" »s)
~L~

(a) (b)
z

t,

Figure 5.2.1 Schematic Diagrams of ShortedMicrostrip PatchAntennas: (a) Circular


Patch; (b) Rectangular Patch

5.2-2 Characteristics of Shorted Patches

5.2-2.1 Introduction

Although conceptually quite easy to understand, a shorted patch is a


complicated radiating structure. As mentioned previously the key to
analyzing the performance and understanding the limitations of any variation
of a microstrip patch antenna is an accurate and fast analysis tool. Thus to
obtain an accuraterepresentation of the performance of a shorted patch, a full-
wave analysis is required. Throughout this characterization, an analysis based
on a full-wave spectral domain Green's function/Galerkin' s technique was
utilized. Attachment modes similar to those introduced in [25] were used to
model the discontinuities associated with the feed and shorting pins. In this
section validation of the analysis will be given as well as a method to increase
203
the impedance bandwidth of a shorted patch to that required for commercial
mobile applications.

Consider a circular probe-fed patch loaded with a single shorting post


as shown in Figure 5.2.1a. Using the philosophy adopted for conventional
microstrip patch design. to achieve reasonable bandwidth a thick. low
dielectric substrate should be utilized. Figure 5.2.2 shows the predicted and
measured input impedance locus of a shorted patch designed for an input
impedance of 50 .Q at a frequency of 1.9 GHz mounted on 10 mm foam (refer
to Figure 5.2.2 captions for the relevant dimensions). As can be seen from
these plots. very good agreement between theory and experiment was
achieved. The predicted and measured 10 dB return loss bandwidth was 5.3
% and 6.8 %. respectively. The impedance bandwidth is significantly more
than that achieved using the technique outlined in [18] and also previously
recorded bandwidths of shorted patches [21.26,27]. The radius of the patch
is 10.65 mm, allowing the antenna to be easily accommodated on a typical
handset. Slight discrepancies between the theoretical and experimental loci
can be attributed to the thin layer of dielectric required to mount the conductor
on the foam. This layer of dielectric (height of 0.254 mm, e, = 2.2) was not
taken into account in the analysis. The separation distance between the feed
pin and the shorting post. ~, was 1.2 mm. An investigation of this parameter
will be presented in Sections 5.2.3 and 5.2.4.

Figure 5.2.2 Predicted and Measured InputImpedance of Circular Probe-fed Shorted


Patch(parameters: £,.=1.07, tanS = 0.015,d = 10 mm,R =10.65 mm, xp = 6.2 mm, Yp = 0, xps
= 8.3 mm, Yps = 0, ro= 0.325 mm, fo. = 0.6 rom)

The predicted and measured Be far-field radiation patterns in the E-


and H-planes of the shorted circular patch are shown in Figures 5.2.3 and
204
5.2.4, respectively. Once again good agreement between experiment and
theory was achieved. For the experiment, the shorted patch was mounted on a
large ground plane (4Ao x 4Ao) to reduce diffraction off the edges. From the
results presented in Figures 5.2.3 and 5.2.4 the ripple commonly associated
with a finite ground plane and more predominate in the E-plane is still
evident. The radiation patterns in these figures are similar to a top loaded
monopole. It is interesting to note that in the E-plane, there is a slight dip in
the pattern at broadside (8 =0°), not a deep null as usually expected for a top
loaded monopole. It can be postulated that this is due to the excitation of
modes on the patch conductor. The extent of the decrease in magnitude is
dependent on the relative position of the shorting pin and probe feed with
respect to the center of the patch. 0

110

Figure 5.2.3 Predicted and Measured E-plane Ee Far-field Radiation Patterns of Circular
Probe-fed Shorted Patch at 1.9 GHz.

110

Figure 5.2.4 Predicted and Measured H-plane Be Far-field Radiation Patterns of Circular
Probe-fed Shorted Patch at 1.9 GHz

205
It should be noted that although the H-plane cross-polarization level
of this antenna (Figure 5.2.4) is relatively high, for the application of mobile
communications handsets this is not a major concern, as much of these fields
will diffract off the edges of the small, finite ground plane within the handset
[29]. The H-plane co-polarized field, E~, is very similar to that of a
conventional patch and therefore is not shown here. The predicted gain of the
antenna is 3.6 dBi. The location of the gain maximum is similar to a top
loaded monopole.

5.2-2.2 Parameter Study of Shorted Patches Incorporating a Single Shorting


Post

In this section, a parameter study of the shorted patch is given and


yields valuable insight into the design of this radiator. From the study, trends
that give the minimum size as well as a design methodology are established.
A comparison of the properties of a shorted patch and a conventional patch
was given in [28] and is not included here.

Schematic diagrams of shorted circular and rectangular patches are


shown in Figure 5.2.1. As can be seen from this figure, there are many
degrees of freedom in the design of one of these radiators, with each affecting
the resonant frequency and input impedance of the antenna. Thus there may
be more than one solution for a given specification and given substrate
parameters. For example, consider the previous stated requirements, namely
50 Q input impedance at 1.9 GHz. Another solution other than that given in
Figure 5.6.2 is the following: R = 11.4 mm, xp =3.5 mm, ro =0.325 mm, xps =
5.7 mm, ros = 0.6 mm. For both cases the radii of the feed and shorting pins
are kept constant. The new solution has the same impedance bandwidth and
gain as the previous solution. The differences in these cases are: (1) the
radius of the patch conductor has increased by 0.75 mm, or 7 %; (2) 11 has
increased by 0.1 mm; and (3) the dip at broadside in Be in the E-plane has
decreased by 2.5 dB. The most important apparent trend that can be
established from these results is that to minimize the size of the printed
conductor, the shorting pin should be located close to the edge of the patch.
The reduction in size is achieved at the cost of reducing 11. This trend has
been tested for many cases (using different substrates) and seems to hold. For
most cases considered ratios of the short position to the radius of the patch,
namely, xpJR" are usually between 80% to 90%. This allows for relatively
easy attachment of the pin to the printed conductor and a minimal patch
radius.

206
Another solution for the given specification can be achieved using the
rectangular shorted patch configuration presented in Figure 5.2.lb. The
= =
dimensions of this case are: L 17.3 mm, W 21.3 mm, xp 5.0 mm, ro = =
= =
0.325 mm, xps 7.4 mm, ros 0.6 mm. Once again, the impedance bandwidth
and radiation patterns are very similar to that presented in Figure 5.2.2-4 .

The performance of a shorted patch is very dependent on the substrate


parameters. As a function of substrate height, the impedance trends are very
similar to those of a conventional patch, that is, as the height increases the
impedance bandwidth increases. For a circular shorted patch mounted on
foam of height 5 mm, the impedance bandwidth is 2.9 % when the dimensions
are optimized for an input impedance of 50 Q at 1.9 GHz. Another trend
similar to a conventional patch is that as the substrate height is increased, the
patch conductor becomes smaller. This is useful for minimizing the size of
the patch conductor. There is yet another very important trend which occurs
when increasing the substrate height of a shorted patch. As the height
increases, I!J. increases. For this case, an increase in substrate thickness by a
factor of two, I!J. increases by almost a factor of three . This reduces the
stringent accuracies required in fabrication of the radiator. Thus shorted
patches are best designed using thick substrates. A summary of the
dependence of bandwidth and I!J. on the electrical thickness of foam substrate
is presented in Figure 5.2.5 .
7.---..,-----.----r----,-----.--,.....,...---,

0.8 f>

0.8
~
0.4

1 '--"-tJr='-----'------'--'---'-----..---I.---'---''--o--J 0.2
o 0.03 0.04 0.05 0.08 0.07

Thickness , Ao

Figure 5.2.5 Performance Trends for Shorted Patches Mounted on Foam (e, =1.07)

The dielectric constant of the substrate material plays a significant


role in the performance of a shorted patch. Predictably, as the dielectric
constant is increased the size of the patch conductor becomes smaller as does
the separation between the probe feed and shorting pin. Once again using the
specifications given previously, a shorted patch mounted on a substrate with d

207
= 10 mm and e, =2.2 has the following dimensions: R =7.8 mm and 11 =0.28
mm. Hence the fabrication tolerances become very stringent. Another very
important consequence of using a dielectric material is the significant
reduction in bandwidth. Although a decrease is expected when comparing to
conventional microstrip patch design, the degree of reduction is larger. For
the Er =2.2 case, the 10 dB return loss bandwidth is only 3.4 %. This can be
explained by considering the simple circuit theory outlined earlier. Recall
that the capacitive coupling between the probe feed and the shorting pin used
to counter the inductive nature of the patch below resonance is proportional to
the dielectric constant of the material. Thus, the rate of change of capacitance
with respect to frequency is also proportional to s, This being so, then the Q-
factor of the resonator will be very high and hence the bandwidth narrow.

Another problem with using a dielectric material is the decrease in


efficiency if the material itself is lossy. As the feeding and shorting pins are
spaced very close together for this case, the conduction loss between the pins
increases [24]. This loss is even further enhanced when lossy material is
used. From the above discussion it is apparent that shorted patches are very
suited to foam substrates.

For typical mobile communications handsets, it is required that the


conductor of the radiating element be protected from the environment. This is
usually achieved by using a radome or coverlayer. For conventional
microstrip patches, incorporating a coverlayer typically reduces the size of the
element. Indeed the same effect occurs for shorted patch configurations,
however, an important side effect also occurs. Consider the same
specifications, the same substrate and the same pin sizes given in Figure 5.2.2,
however this time a coverlayer is used. The coverlayer is a thin dielectric
layer of height 0.508 mm and a dielectric constant of 2.2. The dimension of
the covered shorted patch are: R = 10.3 mm.x, =5.7 mm, xps =8.0 mrn, Thus
as expected the overall dimensions are reduced, however, 11 has increased to
1.375 mm, Thus by incorporating a coverlayer, the stringent tolerances on the
relative positioning of the probe and shorting pins can be improved slightly.
The radiation patterns and impedance locus for the covered shorted patch are
similar to that presented in Figures 5.2.2, 5.2.3 and 5.2.4. The surface wave
efficiency was calculated as 90 %.

For all cases considered so far, the radii of the feed (ro) and shorting
(rOs) pins have been kept constant. To obtain insight into the effect of each
radius on the overall size of the patch conductor and impedance
characteristics, several new cases were investigated. For the first series of
cases, the feed pin was kept constant (ro = 0.325 mm) and the radius of the

208
shorting pin varied. The general trend observed was the impedance locus of
the printed antenna remained similar, once the other dimensions were
modified to ensure 50 Q at 1.9 GHz. It was observed that as the radius of the
shorting post was reduced, the bandwidth did slightly decline, however only
by a fraction of a percent. The main effect of the size of the shorting pin was
on the overall size of the patch conductor as well as t:... The trend observed
was as the radius of the shorting pin is increased, the size of the patch
conductor increases, as does t:... For the previously incorporated substrate, an
increase in the radius of the shorting pin of 0.1 mm, is accompanied by an
increase in the radius of the patch by 0.4 mm and an increase in t:.. of 0.4 mm.
Hence a smaller patch size can be achieved with a thinner shorting pin
conductor, although the problem with accurate positioning of the pin with
respect to the feed becomes worse. Interestingly, the opposite trends occur
for the feed pin radius. As the radius of feed pin is decreased, while keeping
the shorting pin size constant, the radius of the patch conductor increases, as
does the separation distance between the probe and shorting pin required to
achieve 50 Q at 1.9 GHz. Similar incremental values for the radius of the
patch and t:.. were obtained as in the previous case. This phenomenon will be
examined in more detail in the next section.

From the parameter study given, a design methodology can now be


established. For reasonable bandwidths, thick foam substrates should be used.
The use of an electrically thick substrate also reduces the radius of the patch
conductor and increases t:... The shorted patch is very suited to the inclusion
of a coverlayer, which once again, reduces the size of the patch as well as
improves t:... The radii of the pins also play an important role in the overall
size of the patch conductor as well as t:... Since a compromise must be made
between Rand t:.., we found good results can be achieved when ros is
approximately 0.04 Ao, approximately twice that of roo Once again these
values really depend on which is more important, the overall size of the
antenna or the ease in fabrication.

The design procedure of a patch incorporating a single shorting post


is relatively straightforward. As a starting point, the radius of the patch
should be set to approximately A/16 (where Ag is the guided wavelength).
The shorting post should be located at approximately 80 % - 90 % of this
value. As pointed out previously, the closer the shorting pin to the edge, the
smaller the overall size of the patch. The probe feed should be located in
close proximity to the shorting pin and then adjusted to achieve 50 Q . Of
course the design strategy will depend on the thickness of the substrate. As
with the case for a conventional probe fed patch etched on a thick substrate,
the design is somewhat more complicated by the strong inductive nature of
209
the electrically longer probe. However, this design procedure is still relatively
accurate, requiring only slight adjustment of the dimensions.

5.2-2.3 Techniques to Improve Mechanical Tolerances

As was seen in the previous section, the close proximity of the feed
and shorting pins is a major design flaw in the shorted patch antenna.
Although some of the procedures outlined in Section 5.2.3 can reduce the
stringent design tolerances on 11, more robust solutions have yet to be
provided. In this section, the issue of improving the manufacturing ease of a
shorted patch is addressed and two techniques are proposed.

Increase the Number ofShorting Posts

The ftrst technique involves incorporating more shorting posts on the


antenna. The philosophy behind this method can be extrapolated from some
of the results presented in Section 5.2.3. Recall the larger the radius of the
shorting pin, the greater the 11. Another way of interpreting this is the
stronger the shorting plane developed, the greater the 11. Figure 5.2.6 shows
examples of this technique, where probe fed circular patches consisting of two
and three shorting posts are presented. As can be seen from these schematics
there are even more degrees of freedom, in particular the distance between the
shorting pins, or the angular displacement of the pins from the x-axis, q> (refer
to Figure 5.2.6).
y
y

__ -+_~x
I'--_ _~~x

(a) (b)

Figure 5.2.6 Top Viewof Probe-fed Patches Incorporating Multiple Shorting Posts: (a)
Two Posts; (b) ThreePosts

Consider the patch with .two shorting pins shown in Figure 5.2.6a.
Using the same substrate parameters and pin radii given earlier, a patch
configuration with the following dimensions was designed for 50 .Q input
210
impedance at 1.9 GHz: R = 13.2 mm; xp =4.95 mm, Xpst = xpsz = 11.08 mm
and CPt =-cpz = 10° (where cpj is the angle between the ilb shorting pin and the x-
axis). The input impedance locus of this shorted patch is shown in Figure
5.2.7. As can be seen from this figure, the 10 dB return loss bandwidth is
approximately 7.9 %. Thus the improvement in bandwidth and ~, here
approximately 6 mm, are achieved at the cost of increased patch real estate.
The predicted radiated Eo in the E- and H-plane are presented in Figure 5.2.8.
These results are similar to the patterns presented in Figures 5.2.3 and 5.2.4,
with the exception that there is less of a dip at broadside in the E-plane, by a
factor of 2 dB. Thus more power is being radiated with this configuration
than a patch incorporating a single shorting post. There is also a more
pronounced asymmetry in the E-plane pattern as a result of the presence of
more shorting pins.

Figure 5.2.7 Predicted Input Impedance of Circular Probe-fed Shorted Patch


Incorporating Two Posts (parameters: E,- = 1.07, tan S = 0.015, d =10 rom, R = 13.2 rom, xp =
= = =
4.95 rom, Yp =0, Xpsl =Xps2 =11.08 rom, ~I =-~ 10°, ro 0.325 rom, ros 0.6 rom)

,eo

Figure 5.2.8 Predicted Eo Far-field Radiation Patterns of Circular Probe-fed Patch


Incorporating Two Posts at 1.9 GHz

211
As mentioned before, another degree of freedom with the multiple
shorting post antenna is the angular displacement of the posts. Consider the
patch with two shorting posts with the angular displacement of the pins
increased to ± 20°. For an input impedance of 50 Q at 1.9 GHz the
dimensions of this shorted patch are: R = 14.65 rom, xp = 3.65 rom and Xpsl =
X ps2 = 12.7 rom. The corresponding bandwidth of the shorted patch
configuration is 9.4 %. Thus to achieve greater bandwidth, the shorting posts
should be positioned further apart, once again at the expense of increased
patch area. An explanation of this phenomenon is that as 4> increases, the
coupling between the shorting pins decreases, as a function of cos 4>. To
counter this effect, stronger patch modes are required to maintain the
coupling. Thus the patch conductor size must be made larger.

Incorporating more shorting pins increases further the effective size of


the shorting plane. A triple post circular patch was designed for the
previously given specifications and substrate parameters. The relevant
dimensions are: R = 15.4 rom, x, =2.35 rom, Xpl = 13.3 rom, Xp2 = Xp3 = 14.1
rom, 4>1 = 0~.4>z = -4>3 = 20°. The 10 dB return loss bandwidth of this patch
configuration is approximately 10 %, which is similar to the bandwidth of a
conventional circular patch mounted on this substrate. As the number of
shorting pins along the same plane increases, the performance of the printed
antenna approaches that of a conventional quarter-wave patch. Once again
the increase in bandwidth as well as 11 is at the cost of increased patch size.
The radiation patterns for this configuration are similar to that presented in
Figure 5.2.8, with the exception that the dip is further reduced by 1 dB. It is
interesting to note that the magnitudes of the H-plane cross-polarized fields
are still relatively high. This finding is consistent with the postulation made
in [30] about quarter-wavelength patches.

Another approach to improve the manufacturing ease using multiple


shorting pins can be formulated from the simple theory outlined in Section
5.2.1. Here the second post is positioned on the opposite side of the probe
with respect to the first shorting pin. In essence, the two pin combination
effectively halves the required total capacitive coupling required and therefore
the distance between the probe pin and the shorting pins should be increased.
Unfortunately this technique has only limited success, especially when
electrically thick material is used, as the self inductance of the pins ensure the
overall size of the patch becomes significantly large, comparable to a quarter-
wavelength patch.

212
Swap the Relative Location of the Feed with respect to the Post

A simple approach to alleviate the problem of the stringent tolerances


on t!. of a shorted patch can be derived from another set of results presented in
Section 5.2.2. As was shown earlier, the radiation performance and
impedance characteristics of a single shorted patch are similar for different
positions of the feed and shorting pins if the radius of the patch is
appropriately modified. This implies the configurations are very similar
electrically. The key parameter in these designs-is the relative position of the
pins with respect to each other. Consider now swapping the positions of the
feed and shorting pins with respect to the edge of the patch. Thus in Figure
5.2.1, (xps, YPS) now becomes the feed location and (xp, yp) is the shorting post
location. This is shown in Figure 5.2.9.

___ t - - ~ X

shorting post

Figure 5.2.9 Schematic Diagram of Shorted Patch Configuration with Swapped Post and
Feed Pins

After redesigning this configuration, t!. was increased from 1.3 mm to


approximately 4.9 mm. The increase in spacing is achieved at an increase in
patch size, by 1.45 mm (refer to Figure 5.2.10 caption for the relevant
parameters). The predicted and measured input impedance behavior of this
shorted patch is shown in Figure 5.2.10. As can be seen very good agreement
between experiment and theory was achieved. The increase in bandwidth to
7.1 % is another feature of this design. An important trait of this novel
technique is also apparent from Figure 5.2.10. Note that the resonant
frequency shift is 0.2 % in this case, compared to a shift of 0.7 % for the
conventional shorted patch shown in Figure 5.2.2. Thus the novel design is
less sensitive to fabrication errors or substrate variations. The far-field
radiation patterns were measured and are very similar to those presented in
Figures 5.3 and 5.4 and therefore will not be repeated here.

213
Figure 5.2.10 Predicted and Measured Input Impedance of Shorted Microstrip Patch
= = =
(parameters: Er 1.07, tan S 0.015, d 10.0 mm, R 12.1 rom, xp 10.1 mm, Yp O.xps = = = =
= =
5.2 rom, Yps 0, rO 0.325 rom, rOs 0.6 rom) =
As mentioned before, there are several degrees of freedom in the
design of a shorted patch of which the radius of the patch and the locations
shorting post and probe feed are the most critical in determining the
impedance characteristic. Many simulations were run which considered the
effect of fabrication errors on the impedance behavior of shorted patches. To
this end, we have assumed errors of ± 0.25 mm in each of the previously
mentioned parameters. Variations in each parameter were considered
accumulatively and Figure 5.2.11 displays the worse case scenarios in terms
of change in resonant frequency and return loss for both the traditional and the
modified shorted patches.
O.------,- - ---,.- - --r-- --r-- - --r-- --,

·5

iii' ·10
S
] -15
I
I

E
:J
-20 - _ 1-
I
_

I
~
a: -25
I
- - - - 1- - - - -
I
I I
-30 - - - - r - - - - - ,- - - - -
I I
I I
-35 ..................'"""--''-'-~ .........~~.....x. ~~.................................L..~~ ..J
1.75 1.8 1.85 1.9 1.95 2 2.05
Frequency (GHz)
Figure 5.2.11 Effect of Fabrication Errors on Return Loss Behavior of Traditional
Shorted Patch ( • ) and Modified Configuration ( • ). The solid lines are the original
cases and the dashed lines show the worse case scenarios

214
As can be seen from Figure 5.2.11, the changes are more dramatic for
the traditional case, once again highlighting the robustness of the modified
patch configuration. It should be noted that the cases considered here are
relatively broadband in nature, hence the effects of fabrication error are
somewhat reduced. For cases where thin substrates are used, these errors will
be more pronounced.

5.2-3.4 Other conductor shapes for shorted patches

The previously discussed single shorting post design strategy can also
be applied to a printed annular ring to further reduce its surface area. Figure
5.2.12 shows a schematic of a shorted annular ring and the input impedance
variation of such an antenna using 10 mm 51 HF Rohacell foam as the
substrate. The dimensions of the structure are given in the figure caption. As
can be seen from the measured and predicted impedance loci, very good
agreement between the two is evident. The predicted and measured 10 dB
return loss bandwidths of the antenna are 5.8 % and 5.9 %, respectively. The
slight shift in resonant frequency apparent in Figure 5.2.12 can be attributed
to the thin layer of dielectric material required to etch the patch conductor.
This layer (height of 0.254 rom, e, = 2.2) is not included in the analysis. For
the theoretical results displayed in Figure 5.2.12, two attachment modes and
twenty-five entire domain basis functions were used. The radiation patterns
of the shorted ring were measured and are very similar to those presented in
Figure 5.2.3 and 4 of a shorted circular patch. For the sake of brevity these
patterns will not be given here. The gain of the antenna was measured as 3.5
dBi, compared to a theoretical value of 3.6 dBi.
y

t:

Figure 5.2.12 (a) Schematic Diagram of Shorted Annular Ring; (b) Input Impedance
Behavior of Shorted Annular Ring (parameters : d = 10.0 rom, E,. = 1.07, tan 0 = 0.01, a = 6.0
rom, b = 10.3 mm, Pp = 7.0 rom. $p = 0, ro= 0.325 rom, PI" = 9.3 rom, $ps = 0, ros = 0.6 rom)

215
It is interesting to compare the properties of the shorted annular ring
with the previously presented shorted patches. Using the same dielectric
materials, the outer radius of the shorted ring is slightly smaller than the
radius of a circular shorted patch, by approximately 5%. The bandwidth and
the gain of the shorted ring are once again slightly less than the circular patch
version, by 0.5 % and 0.2 dBi, respectively. These results can be attributed to
the smaller area of the ring-based antenna. Despite these relative
shortcomings in the electrical performance of the shorted annular ring there is
a distinct advantage. The gap between the shorting and feed pins is greater,
by 14 %. This allows for easier manufacturing of the antenna and makes it
less susceptible to errors in the positioning of these pins. As was shown in
Section 5.2.2, the coupling between the feed, the shorting post and the closest
radiating edge plays a significant role in the impedance behavior and hence
the size of the patch. Like for the conventional stacked patch case considered
earlier, the annular ring has another degree of freedom that contributes to the
impedance behavior, namely, the inner conductor edge. Thus the strong
capacitive coupling required for a shorted patch to operate can be distributed
amongst all these variables. It was shown in Section 5.2.2 that for minimum
area the shorting pin must be located closest to one of the radiating edges of
the patch. However, it was also shown that for maximum spacing between
the feed and shorting pins, the feed should be located nearest to this edge.
Hence for a circular or rectangular patch configuration, a compromise must be
made. However, by using an annular ring both criteria can be achieved. Here
the feed pin is located nearest the inner conductor edge and the shorting pin
located near the outer edge, providing a shorted patch of minimal area as well
as acceptable spacing between the pins.

5.3 Further Size Reduction Technique for Shorted Patches

5.3-1 Introduction

Although the conventional shorted patch can yield significant size


reduces, there are still applications where the reduction provided is not
enough. Two methods for further reducing the size of a shorted patch are
presented in this section. This technique utilizes strategically positioned
notches within the patch conductor near the shorting pin and yields a further
reduction in the conductor size.

5.3-2 Configuration and Design Strategy

216
Throughout this investigation, edge-fed rectangular shorted patches
are examined, although the findings can be applied to any other form of the
shorted patch. To accurately analyze this style of shorted patch, the full-wave
software package, Ensemble 5.1 [31] was utilized. Figure 5.3.1 shows a
schematic of the proposed shorted patch antenna. As can be seen from the
top-view of the antenna, there are two 'notches' each located on
symmetrically opposite sides of the feed, near the shorting pin.

Yt
z

i
L ~

L~
E,

i
• w

I
r
1
Shorting pin (rOo)

SMA feed

Figure 5.3.1 Schematic of Edge-fed Shorted Patch: (a) Top View; (b) Side View

The explanation of the effect of these deformations in the conductor


shape on the impedance performance of the shorted patch is as follows. As
was shown earlier this Chapter, a shorted annular ring patch is somewhat
smaller than a filled shape shorted patch conductor (such as a circle or
triangle). The postulation as to why this is so is that for the annular ring
configuration, an extra capacitance is created by the 'fringing fields'
associated with the close proximity of the shorting pin and coaxial feed to
both the inner and outer patch conductor edges. This additional capacitance
further negates the inductive nature of the shorted patch antenna below its
first resonance and hence lowers the operation frequency. hnportantly too, by
creating such an effect, the tolerances on the positioning of the feed and
shorting pin can be eased. The above-mentioned approach can be applied to a
filled shaped patch too.

As can be seen from Figure 5.3.1, there many degrees of freedom in


the design of the new shorted patch. Other than the variables associated with
the standard shorted patch (such as its length, etc) there are two notches.
These two notches can be interpreted as consisting of three 'conductor pads':
the first pad where the short circuit is located and the other two near the feed
line (refer to Figure 5.3.1). The dimensions of each pad will affect the
impedance behavior of the antenna.

217
Simulations have shown that the length of the first pad (L 1 in Figure
5.3.1) should be set to the diameter of the shorting pin to yield the smallest
size patch conductor. The width of the first pad (WI) plays a key role in the
overall size of the printed antenna and its input impedance nature. If WI is set
to the width of feed insert (Wfi in Figure 5.3.1) then the antenna behaves as a
standard short patch. As WI is reduced, the overall patch conductor size
decreases, due to the phenomenon described earlier associated with the fringe
capacitance. When the width is reduced to that of the shorting pin, the
maximum size reduction is achieved. However, as WI is decreased, the input
impedance is also reduced until this point where of course the antenna appears
almost as a short circuit at the feed port. Thus a compromise must be made
between maximum size reduction and input impedance behavior. This control
of the input impedance response alleviates to an extent the required very
accurate positioning of the shorting pin with respect to the feed of a standard
shorted patch [21].

The presence of the other two pads once again increases the fringing
capacitance around the post/feed thereby reducing the size of the patch
conductor. These additional pads have a secondary effect on the impedance
response.

5.3·3 Experimental Results

Figure 5.3.2 shows the theoretical return loss behavior of several


shorted patch antennas mounted on the same material (refer to the figure
captions for the parameters). Shown in this graph are a standard shorted
patch, a shorted patch incorporating just the first pad, and a shorted patch
utilizing the two notches. As can be seen-from Figure 5.3.2, the resonant
frequency steadily decreases with the new modifications to the shorted patch.
To confirm the theoretical results a shorted patch with the two notches was
fabricated and its return loss characteristic is also shown in Figure 5.3.2. As is
evident from the return loss plot, very good agreement between measurement
and experiment was achieved. The new patch is 75 % smaller than a
conventional microstrip patch antenna mounted on the same material and 20
% smaller than a conventional shorted patch [21].

The radiation patterns of the proposed shorted patch were measured


and are shown in Figure 5.3.3. For this measurement, the ground-plane was
very small, extending 1 em each side of the patch conductor and thus the near
omni-directional patterns were observed. The gain of the antenna was
measured as 1.0 dBi. As can be seen from Figure 5.3.3, the patterns are very
similar to a conventional shorted patch when a small ground-plane is used
[32].
218
o . _.-- - - '.•
", ·· ••
#-

-5
, ~'i ;:
::
'i

~10
I
~
III
,i
III I
,3.15
c::
:i t - ••• Standard S-P
Q).20 ~ ••••••.. 1 Pad Sop
a: 2 Notches Sop
~i - • _. Measured
-25
I

-30
0.5 0.55 0.6 0.65 0.7 0.75 0.8
Frequency (GHz)

Figure 5.3.2 Return Loss of Various Shorted Patches (Substrate Parameters: Er =2.2, d =
1.59 rom. Shorted Patch Dimensions: L = 44.5 rom, W = 48.75 mID, 4= 16 rom, Wf = 1.0 rom,
Ln = 11.5 mm, Wfi= 7 mm, L. = 1.0 mm, WI = 3.0 mm, ~ = 2.0 mm, Wz = 3.0 rom, xps = 0.5
mm,ros = 1.0 mm)
o

110

Figure 5.3.3 Measured Radiation Patterns of Shorted Patch

5.4 Winged Shorted Patch

5.4-1 Introduction

This section investigates a winged patch antenna incorporating a


single shorting post and a notch located near the shorting pin. The antenna is
219
significantly smaller than previously reported shorted patches, by more than a
factor of 25. The antenna is also relatively easy to manufacture and does not
require the shorting post to be located close to the feed position. An
integrated version of the proposed antenna mounted on a wireless
communication handset model is studied experimentally and is shown to have
an RF performance compliant with most cordless systems. It should be noted
that throughout this section, the terms shorted patch and wing antenna are
used to describe a patch antenna with a single shorting post and the proposed
configuration, respectively. The latter term is employed as a consequence of
shape of the printed conductor.

The order of this section is as follows: Section 5.4-2 presents the


characteristics of the shorted wing antenna. A brief study of the parameters
affecting the resonant frequency of the antenna is given. From this study,
valuable insight into the optimum design of the winged patches is achieved.
The impedance and radiation characteristics of a winged patch incorporating a
single shorting post is presented in Section 5.4-3. Comparison between
theoretical and experimental results is provided. Also a comparison in
conductor area with a 'standard' edge-fed shorted rectangular patch is
provided. Section 5.4-4 presents an integrated version of the wing antenna on
a wireless handset. Here, the structure of the handset is addressed and the
experimental impedance and far-field radiation characteristics are presented.

5.4-2 Characteristics of Shorted Winged Antenna

The shorted winged printed antenna is shown in Figure 5.4.1. Here, a


shorted patch of length L1 and width WI is edge-fed by a microstrip line of
length L, and width Wr. To couple power to the edge-fed antenna, a coaxial
probe of radius rp is located xp from the end of the microstrip line (refer to
Figure 5.4.1). The shorting post for this antenna is located at xps from the
contact between the microstrip line and the edge of the patch conductor.
Importantly, a notch of length L, and width Wn is located within the patch
conductor. The purpose of this notch is to utilize the principle outlined in
Section 5.3 to reduce the overall dimensions of the antenna. The printed
antenna is etched on a grounded substrate with a height, d and a dielectric
constant, e,

As mentioned in Section 5.3, the key to reduce the conductor size of a


shorted patch is to create extra capacitance to counter the inductive nature of a
patch below resonance. This is done here by locating the shorting pin near the
patch conductor edge and by contributing to the fringing fields via the
strategically positioned notch, as shown in Figure 5.4.1. As can be seen from
this schematic, there are many degrees of freedoms in the proposed antenna.
220
Apart from the standard variables associated with the standard shorted patch
(such as its length), there are the notch (size and location), the microstrip
feedline and the position of the coaxial probe. Many simulations were done on
a more general version of the antenna where the notch could be located
anywhere within the patch conductor and it was found that the geometry in
Figure 5.4.1 gave the optimum performance in terms of size reduction and
also ease of fabrication . As can be seen from the schematic in Figure 5.4.1,
the problem of the close proximity of the shorting post and the coaxial probe
position for the antenna in [21] has been resolved. This factor should make
the new antenna easier to manufacture. The key variables here are xps, Lz (or
Ln) and the distance between the short and probe feed. It was found that to
minimize the size of the patch conductor, the shorting post must be located as
close as possible to the conductor edge, in fact, as close to as many conductor
edges as possible. Once again this creates the maximum field disturbance and
thereby reducing the resonant frequency of the antenna. Thus minimizing xps,
Lz and increasing (Lr..:. xp) in Figure 5.4.1, reduces the patch conductor size of
the antenna for a given frequency of operation. Of course Lj, W h e, and the
thickness of the substrate play important roles in reducing the resonant
frequency and their effects are similar to that for a standard shorted patch
given in Section 5.2. x

t
4 W, ..
t, +- 1..--" w.
t -.. y
L, + i-
t .-i- x,.
+
L,
T
t
L,

~.-
....vr,- .-
x,

(a)

(b)

Figure 5.4.1 Schematic Diagramof PrintedWing AntennaIncorporating Single


ShortingPost: (a) Top View; (b) Side View

5.4-3 Experimental Results

Figure 5.4.2 shows the predicted and measured input impedance for a
wing antenna designed for operation at a frequency of 2.7 GHz mounted on
221
10 mm foam (refer to Figure 5.4.2 captions for dimensions). The predicted
and measured 10-dB return loss bandwidth was 7.4 % and 6.9 % respectively.
This impedance bandwidth is comparable with a standard shorted patch
presented earlier in this chapter. The measured far-field radiation
characteristics of the shorted winged patch are shown in Figure 5.4.3 (note:
broadside is at 1800 in this figure). For the measurement, the ground plane for
the antenna was very small, extending 15 mm each side of the printed
conductor. The radiation pattern is similar to that of the shorted patches
presented in this chapter, however, the cross-polarization level for this
antenna is somewhat higher, particularly in the H- plane, due to the small
sized ground plane. As was shown in [34], the size of the handset antenna
affects the radiation properties. The smaller the ground-plane size, the more
the fields diffract around the ground-plane edges. Of course, the size of the
ground plane cannot be too large as it will not be able to be accommodated for
on the handset. The 6ain of the antenna was measured as 1 dBi.

--.-- •••••.• • Experiment


- - Theory
iil -5
:E..
:l -10
.2 --_.. _._-
E
::I

e -15
-20
2.3 2.4 2.5 2.6 2.7 2.8 2.9
frequency (GHz)
Figure 5.4.2 Predicted and Measured Return Loss of Winged Shorted Patch
(Parameters: E, =1.07, tan &=0.015, d =10 nun, WI =19 nun, L I =6 nun, Wn =4 nun, L.. =5
mm, Lz =2 mm, Lr=7 nun, Wf = 1.2 mm, xp = I nun, Xps =1 mm, r p =0.6 mm, rps =0.325
m~

''0
Figure 5.4.3 Measured Far-field Radiation Patterns of Winged Shorted Patch at 2.7 GHz

222
To put these results into perspective, a standard edge-fed rectangular
shorted patch designed to operate at this frequency using the same dielectric
material has the dimensions of 66 mm x 46 mm. Thus the new antenna is
smaller by a factor of 26 comparing the conductor area. The bandwidth of the
wing antenna is slightly smaller, 7 % compared to 10 %.

Other variations of the wing antenna were investigated and


schematics of these are shown in Figure 5.4.4. Each one of these
configurations resulted in similar performance to that shown in Figure 5.4.2
and 5.4.3. The reason for this is evident in Figure 5.4.4 where a vector
representation of the current flow for each configuration is given. As can be
seen from these diagrams the current flow for each antenna is very similar. As
was stated before, the dimension L, is important in lowering the resonant
frequency; however, as shown in Fig. 5.4.4, this feedline can be bent around
the patch conductor, thereby not increasing the surface area of the antenna
much beyond WI x L,

r ,.

Figure5.4.4 Current Distribution for Different Versions of Printed Wing Antennas

5.4-4 Integrated Winged Antenna on a Mobile Handset Model

Printed antennas that can be integrated into a mobile handset offer


several advantages compared to external antennas such as monopoles and
helixes as mentioned in Section 5.1. The conformal antenna is less likely to be
damaged, if stategically located, radiate less power toward the head, reducing
the specific absorption rate [33] (which will be shown later in this chapter)
and are less sensitive to the geometry of the handset [34].

To investigate the performance of the wing antenna in a mobile


handset environment, a customized version of the handset specified in the
European framework for the Cooperation in Science and Technology Project
223
COST244 [35] was developed. To accommodate the antenna structure, a
section was removed thus enabling the antenna to be embedded in the
handset. It is important to note that the antenna in Figure 5.4.5 is not entirely
surrounded by metallic walls, unlike the cavity-backed version of a loaded
patch given in [36]. Since on a handset there is limited real estate, the cavity
walls would need to be located near the patch conductor edge, significantly
reducing the bandwidth of the antenna [38]. To get access to the input port of
the antenria, a small hole was drilled in the back of the handset.

Probe

Figure 5.4.5 Mobilecommunication handsetmodel (Dimensions: L = 46 mm,W = 27


mm, H =125 mm, H.44 mm, W. =10mm)

5.4-5 Experimental Results

Figure 5.4.6 shows the predicted and measured return loss of a wing
antenna on the developed handset designed for operation at 1.9 GHz. The
slight discrepancy between the experimental and calculated results can be
attributed to the handset not being taken into consideration during the
simulation. The measured bandwidth of the antenna on the handset was 5.7 %,
which is compliant with most cordless systems (for example the DECT [37]).
To improve the bandwidth performance to be consistent with PCS
requirements, stacking the printed antenna as in [38, 39] or using a resistive
termination as in [22] could be done, although this latter method should be
avoided if possible and this will be addressed later. The measured far-field
radiation pattern is shown in Figure 5.4.7 . The ripples observed are commonly
224
associated with the interaction of the handset and the antenna [33, 34]. The
measured gain of this configuration was 0.6 dBi.
o
" .... --~-~
--10
m
"C
';;-20
en
..2
E-30
::J
li
"'-40
-50
1.75 1.8 1.85 1.9 1.95 2
frequency (GHz)

Figure 5.4.6 Predicted andMeasured Return Loss of ShortedWinged Antenna on


Mobile Communications Handset (Parameters: e,. = 1.07,tan S = 0.015,d = 10mm, WI = 30
mm, LI = 10nun, Wn= 7 mm, L, = 14 mm, ~ =3 mm, Wr= 1.2 mm,Lr= 10 nun, xp= 1 nun,
xps = 1 nun, rp = 0.6 nun, rps = 0.325 nun)
o

Figure 5.4.7 Measured Far-field Radiation Patterns of Winged ShortedPatch at 1.9 GHz
on MobileHandset

It should be noted that with respect to size reduction, in the limit,


shorted patches are very similar to Planar Inverted F antennas (PIFAs), such
as those reported in [12, 14]. It is therefore pertinent to compare some of the
techniques used to reduce the size of PIFAs to the proposed shorted patch

225
method here. Although the volume of the PIFA in [14] is smaller, it is
substantially more difficult to manufacture than the proposed printed antenna
here. One of the key issues associated with any low cost, high volume
antenna is the ease of manufacturing and its susceptibility to manufacturing
errors. There are several very important properties of the proposed antenna
compared to that in [14] related to its ease of construction. Firstly, the
antenna here is a single layer structure (as opposed to the design presented in
[14]) and therefore there are no associated alignment problems. Also the
antenna here requires less solder joints and therefore is more robust. The
antenna in [14] also has a higher component account and therefore is more
expensive to manufacture. It should be pointed out that the size reduction
proposed here can be applied to each of [14], to give even further size
reduction, as these papers have not attempted to optimize the 'patch'
conductor shape, as done here.

The next higher order resonance for the two designed antennas
presented here are weakly coupled and occur at 8 GHz and 9.5 GHz,
respectively. These frequencies are somewhat higher than expected for a
small antenna and can be attributed to the non-conically shaped conductor
making it difficult for higher order resonances to be impedance matched.
This property of the antenna is very advantageous ; especially as tighter
emission requirements by government agencies are enforced for these
systems.

5.5 Shorted Spiral Patches

5.5-4 Introduction

In this section we present two shorted patch geometries that


significantly further decrease the resonant frequency of the printed antenna.
The antenna utilizes the meander line technique traditionally used to decrease
the size of the wire dipoles and monopoles operating in the low MHz
frequency range [40]. Here the wire (of set length) is folded back on itself to
reduce the physical length of the antenna in a given direction . This in tum
leads to a very important postulation: to decrease the resonant frequency of an
antenna for a given surface area, the current path must be maximized within
that area. This postulation can also be applied to the shorted patch concept.
Reductions in resonant frequency by factors of 5.5 and 10, respectively
compared to a standard microstrip patch antenna are achieved using two
forms of spiral shaped conductors. This frequency reduction corresponds to a
decrease in area by a factor of 111 and 261, respectively. This is the largest

226
reduction recorded in the literature. The two spiral shorted patches are
investigated theoretically and experimentally with their input impedance and
radiation characteristics given.

5.5-5 Configurations

Figure 5.5.1 shows a schematic diagram of the patch conductor of a


conventional shorted microstrip patch antenna and the two new spiral shorted
patches. Each antenna is fed by a coaxial probe of radius ro located at (xp, yp)
from the center of the antenna and has a shorting pin, radius rOs' located in
close proximity to the feed positioned at (xps, yps) from the center of the patch
to reduce the overall size of the antenna. The overall areas of each patch
geometries in Figure 5.5.1 are the same (L x W). Each shorted patch is
mounted on a grounded substrate of thickness d and dielectric constant e, (not
shown in Figure 5.5.1). As mentioned before, the objective here is to increase
the length of the current path for a set patch conductor area. For the first new
printed antenna (Figure 5.5.lb), the conductor track is folded back onto itself
in the two dimensions to maximize this current path length. In the second
configuration, a standard rectangular spiral shape is utilized. For both
geometries, as can be seen from Figure 5.5.1, there are several degrees of
freedom, other than the standard parameters associated with a conventional
shorted patch presented earlier. Each new configuration has three dependent
variables for a constant antenna area, namely, the number of bends in the
conductor, the conductor width and the gap between conductors. It should be
noted that the thickness of the conductor track has a lower limit set by the
diameters of the probe and shorting pins. To investigate the performance of
the proposed shorted patch configurations in terms of lowering the resonant
frequency Ensemble 5.1 was used.

Through many simulations, it was found that increasing the gap


between conductors for both configurations reduced the resonant frequency of
the antenna. Probably the most important variable is the number of bends
existing in the conductor track. It was found that the lower the number of
bends, the lower the resonant frequency that could be achieved. Thus
configuration 2 (Figure 5.5.lc) should, have a lower resonant frequency than
configuration 1 (Figure 5.5.lb). This can be easily interpolated using an
equivalent circuit for the shorted patch. By creating a bend in the track, a
'fringe' capacitance is formed. This capacitance counters the capacitance
associated with having the shorting pin located in close proximity to the feed
[21,32], thereby increasing the resonant frequency. However, this increase is
minor compared to the significant reductions that can be achieved by
maximizing the conductor length using either configuration.

227
Shorting pin <x,.. y.,l

I .
\
Probe <x,. Y~

--- - L - - --.
Probe (x,.. Y,) Shorting pin (x,.. Y,.)
(a) (b)

(e)

Figure 5.5.1 Schematic Diagrams of Conventional Shorted Patchand Two Printed


ShortedRectangular Spirals

To test the proposed shorted patches, the geometries in Figures 5.5.1b


and c were developed. Here the material used and the area was the same as
that for a conventional rectangular shorted patch designed for resonance at 1.9
GHz in the previous section. A thin layer of Taconic TLC30 was used to etch
the copper conductor of the patches. The outer dimensions are 17 mm x 25
mm. The track width for each configuration was 1.2 mm, with a gap width of
1 mm. The size of each ground-plane was 4 em x 4 em, Photographs of the
manufactured antenna configurations 1 and 2 are shown in Figure 5.5.2 and
5.5.3 respectively.

5.5-3 Experimental Results

Figure 5.5.4 shows the measured return loss of the two proposed
shorted patches . As can be seen from Figure 5.5.4, configuration 1 has a
resonant frequency at 715 MHz and configuration 2 has a resonant frequency
at 440 MHz. The simulated resonant frequencies were at 690 MHz and 460
MHz, respectively. These frequencies are significantly lower than a standard
microstrip patch antenna of the same area (f, = 4.5 GHz) as a well as a
conventional shorted patch (f, = 1.9 GHz). The reductions in resonant
228
frequency correspond to area size reductions of the order of 111 and 261,
respectively. Of course the 10 dB return loss bandwidths of the two
configurations shown in Figure 5.5.4 are very narrow, however this is only
due to the material thickness not being scaled for the new operating
frequencies 450 MHz & 735 MHz, respectively. There does exist a
compromise between size reduction and achievable bandwidth, which is
consistent with fundamental antenna theory [10].

Figure 5.5.2 Photograph of Configuration 1

Figure 5.5.3 Photograph of Configuration 2

229
0
~
. r#,
~~ I
·5 .• !
iii !:l
~
III ·10
:!
III
~u
.
0
..J u

··
C u
:2 -15
Gi
a:

···
- • - conllgur.Uon 1
.. .. . .. conllguflUon 2
·20 Theory(conllg.2)
••• •Theory(conlllJ.l)

-25
0.3 0.4 0.5 0.6 0.7 0.6 0.9
Frequency (GHz)

Figure 5.5.4 Return Loss of Configuration 1 and Configuration 2 (Parameters: e, 1.07, =


d =10 mm, L =17 mID, W =25 mm, WI =1.2 mID, w g =1 DUD, ro =0.6 mm, rOs =0.325 mm:
configuration 1: xp =-3.1 mm, Yp =-7 mm, xps =-3.1 mID, Yps =-5 mm; configuration 2: xp =-
9.5 mm, Yp =7.4 mm, xps =-11.5 DUD, Yps =7.4 mm)

It is interesting to note that although the conductor track of


configuration 1 is longer than that of configuration 2, by a factor of
approximately 1.7, its resonant frequency is higher. This can be attributed to
the comment made previously with respect to the number of bends within the
structure, for these cases 21 and 14, respectively.

Figure 5.5.5 shows the co-polar and cross-polar radiation patterns of


configuration 1 measured at 700 MHz. These patterns are similar to
conventional shorted patches when mounted on very small ground planes.
The gain of the antenna was measured as 0.3 dBi, although an accurate
measurement of this quantity is very difficult due to the small size of the
antenna [41]. Similar radiation performance for the second configuration
(Figure 5.5.Ic) was obtained at 440 MHz and so for the sake of brevity the
results will not be shown.

,.
Figure 5.5.5 Measured Radiation Patterns of Configuration 1
230
5.6 Improving the Performances of Small Microstrip Patches

5.6-1 Introduction

In the previous section, we investigated size reduction techniques for


shorted microstrip patches. The main objective of these low profile antennas
was to reduce the overall size of the patch conductor. However, little attention
was paid to the impedance bandwidth and polarization aspects of the radiated
fields and as a result all of these configuration have very narrow bandwidth
and high cross-polarization levels. The objective of this section is investigate
novel small printed antennas with low cross-polarized fields and with
enhanced bandwidth performance.

In Chapter 3, we saw that vertically stacking several patches can


significantly enhance the bandwidth of the antenna. This concept is applied in
Section 5.6-2 to a shorted patch case. Several configurations are investigated.
In Section 5.6-3, a technique is discussed that incorporates a balanced feed
arrangement for a shorted patch to reduce the cross-polarization levels. In
Section 5.6-4 another method utilizing a balanced feed is presented. This feed
configuration is applied to dual concentric ring printed antennas. The new
dual concentric shorted rings have a compact structure due to each of the
annular rings loaded by a localized dielectric layer. Also, improved bandwidth
performance is achieved by planar coupling the rings.

In Section 5.5, shorted spiral printed antennas were presented. It was


shown that significant reduction in antenna real estate was achieved by
maximizing the current paths of the antennas. The major drawback with the
antenna configurations was that it exhibited bandwidths of less than 1%
making it not suitable for most wireless applications. In Section 5.6-5 we
present a bandwidth enhancement technique to improve the bandwidths of the
spiral conductors. The technique utilizes two shorted interleaved spiral
conductors and has shown to have a significant improvement in bandwidth.
The dual spiral printed antennas is experimentally investigated with its input
impedance and radiation properties presented.

For all the configurations presented a critique is provided, as the


philosophy of "there is no such thing as a free lunch", still applies, even here.
For all the configurations presented in this section, the degree of complexity
has increased. However, having said this, I believe that the rewards evident in
the results obtained far out weigh this increase in structural complexity.

231
5.6-2 Stacked Shorted Patches

5.6-2.1 Introduction

As presented in Chapter 3, a common technique to enhance the


bandwidth of a printed antenna is to parasitically couple another printed
radiator to the driven element. A coplanar version of this was presented in
chapter 3, with a shorted patch and an annular ring. Although this small
antenna had a bandwidth greater than 10 %, the increase in surface area may
not be acceptable in some applications. When the parasitic element is in the
vertical direction, the technique is often referred to as stacking. Importantly,
as shown in chapter 3, when stacking printed antennas, the surface area is not
greatly increased. In this section we apply this philosophy to a short patch
antenna.

5.6-2.2 Configuration and Theory

Figure 5.6.1 shows a schematic diagram of a stacked probe-fed


rectangular shorted patch. Here the driven patch of length, L 1 and width. WI is
etched on a substrate of thickness. d l and with a dielectric constant. ErI. The
driven patch is fed by a coaxial probe of radius, rOlocated at (xp' yp) from the
center of the patch. Importantly, the shorting post of radius, rO s' positioned at
(x ps' yps) is connected to and extends through the first printed conductor and
is soldered to the second patch of dimensions Lz and W2. A dielectric layer of
height, d2 and dielectric constant. Ea separates the two patch conductors.

" - - -1.. ~

Figure 5.6.1 Schematic Diagram of a Stacked Shorted Patch

To investigate the geometry shown in Figure 5.6.1. the commercial


FDTD code, XFDTD was used. Details of its implementation can be found in
[42]. This analysis was incorporated due to its flexibility and ease of which it
can model electrically complicated antenna structures such as shown in Figure
232
5.6.1. For the printed antenna configuration presented here, a voxel cube of
2.5 mm was utilized. It should be noted that the infinite ground-plane option
was implemented in the code such that the results could be compared to
previous shorted patch configurations.

5.6-2.3 Results and Discussion

Although there are a variety of stacked printed antennas, there are


several commonalities that are evident when trying to broaden the impedance
bandwidth. As was shown in Chapter 3 for aperture coupled patches, large
bandwidths can be achieved if a combination of low dielectric constant
material (say, RT Duroid 5880, Er = 2.2) and foam (Er = 1.07) is used. The
broad banding effect can be attributed to the similar sizes of the patches and
thus the extent of coupling between these elements.

Consider a stacked shorted patch designed for operation near 1.8


GHz. Incorporating the above-mentioned approach, 5 mm RT Duroid 5880
was used for the bottom dielectric layer and a 7.5 mm layer of foam was
utilized between the shorted patches. A thin layer of 0.25 mm RT Duroid
5880 was used to etch the top patch conductor and this was adhered to the
foam. The shorted rectangular patch conductors were both 17.5 mm x 15
mm. The shorting post was located 2.5 mm from the edge of the patches and
the probe feed was a further 5 mm closer to the center of the structure. It
should be noted that the relative distance between the two posts is
significantly greater than that for a conventional shorted patch: in this case by
a factor of four. The relative distance between these posts is critical in terms
of repeatable manufacturing of such antennas. The greater the distance, the
less sensitive the antenna is to positional errors. The connection of the
shorting post to both patch conductors is quite easy. After the post-holes were
drilled in the materials, the 12.5 mm long pin was soldered to the ground-
plane and the lower patch. The upper material was then positioned on top of
this and the pin was soldered to this upper patch.

Figure 5.6.2 shows the predicted and measured input impedance of


the stacked shorted patch. As can be seen from this figure, there is good
agreement between these results . The predicted and measured 10 dB return
loss bandwidth of the antenna is 9.6 % and 10.1 %, respectively. This
bandwidth is significantly more than previously obtained for shorted patches
and approaches that reported in [43]. Importantly here, the surface area of the
new antenna is still small. The measured E- and H-plane co-polar and cross-
polar radiation patterns are shown in Figure 5.6.3. As can be seen from
Figure 5.6.3, these patterns are very similar to a conventional shorted patch

233
shown in Section 5.2. The asymmetry in the E-plane co-polar plot can be
attributed to the presence of the shorting post. A gain of 3.8 dBi was
measured using a ScientificAtlanta 12-1.7 standard gain hom.

Figure 5.6.2 Predicted and Measured Input Impedance of Stacked Shorted Microstrip
Patch

110

Figure 5.6.3 Measured E-plane and H-plane Co-polar and Cross-polar Radiation
Patterns

5.6-2.4 EnhancedBandwidth Stacked Shorted Patch

The stacked version of the shorted patch antenna presented above


showed enhanced bandwidth behavior compared to a conventional shorted
patch. The stacked version did not increase the surface area required for the
antenna and also eased the manufacturing tolerances on the position of the
shorting post with respect to the feed. Although the stacked shorted patch
showed improved impedance performance, it was not broadband. In this
subsection we are going to see how much we can increase the impedance
bandwidth by, albeit at the expense of a larger shorted stacked patch .

234
As stated in Chapter 3, the key to obtain a wide-band impedance
characteristic, is to ensure that a 'mutual resonance' has occur. This
phenomenon yields a loop in the impedance locus when plotted on a Smith
chart. A hint to why this was not obtained in the previous subsection can be
found by examining the Smith chart of the results (Figure 5.6.2). As can be
seen from this figure, the shorted patches are over-coupled, producing a
somewhat large 'coupling loop', similar to a conventional stacked patch when
the patches are located to close together. To achieve a wide-band impedance
response the shorted patches must be located further apart.

Figure 5.6.4 shows a schematic of the broadband stacked shorted


patch. The new configuration has an extra shorting pin located between the
two patches. This pin (radius, rod gives an extra degree of freedom in the
design of the antenna and in particular allows greater control of the
impedance locus. The lower patch is fed by a coaxial probe (radius, ro)
located at (xp, yp) from the edge of the patch and a shorting pin (radius rOsl) is
connected to and extends through the first patch conductor and is soldered to
the second patch. To analyze the stacked shorted patch shown in Figure 5.6.4,
the software package Ensemble 5.1 was utilized.

Figure 5.6.4 Schematic Diagram of Broadband Shorted Slacked Patch

The design procedure for a broadband stacked shorted patch is


relatively straightforward. As in Chapter 3, the driven patch must be over-
coupled to the feed and so in the case of a shorted patch, this means the probe
and shorting post must be located far apart. Also the lower patch must be
mounted on an electrically thin substrate, typically O.01Ao. The bandwidth of
the stacked shorted patch is primarily governed by the properties of the
material residing between the two patches. A low dielectric constant material
(foam, s, = 1.07) that is electrically thick (say 0.09 Ao) tends to give wide
bandwidth. Of course the height of this material will be dependent on the
height of the lower dielectric layer, that is, the thicker the lower layer, the

235
thinner the upper layer must be made if good control of the impedance locus
is desired. Thus there is a compromise between bandwidth and impedance
control. The size of the patches in the initial design phase is not too important
as the relative dimensions will be modified through the course of the design to
ensure the loop in the impedance locus is center around 50 Q. A typical
starting point is a length and width for each patch of Arj8.

Once the resonant loop has been formed in the impedance response as
a result of the choice of materials used, its location on the Smith chart can be
controlled by changing the relative dimensions of the patches. Making the
bottom patch larger than the top patch lowers the real part of the input
impedance. The opposite effect can be achieved by reversing the size trend.
The position of the feed probe and the second shorting pin provide means of
fine tuning the impedance locus.

Figure 5.6.5 shows the predicted and measured input impedance


response of a stacked shorted patch (refer to the figure caption for the relevant
dimensions). Here the second shorting pin is located directly above the feed
probe and so for the experiment, the feed pin was simply extended and
soldered to both patches. The predicted and measured 10 dB return loss
bandwidth was 32 % and 34 %, respectively. This bandwidth is significantly
larger than the previously considered variations of the shorted patch and is in
fact also greater than for a conventional probe-fed stacked patch antenna (see
Chapter 3). The latter can be attributed to the extra degree of freedom offered
with the newly proposed antenna and thus the better impedance control
obtainable.

Figure 5.6.5 Predicted and Measured Input Impedance Response (Parameters: e,.1 =ea =
1.07, d l = 2.0 mm, d2 = 18.0 mm, L I = 36.0 mm, WI = 23.0 mm, ~ = W2 = 23.0 mm, xp = Xpa2
= 15.0 mm, ro = r0a2 = 0.325 nun, Xpsl = 1.0 rom, rOsI = 0.6 mm)
236
The measured E- and H-plane Eo radiation patterns are shown in
Figure 5.6.6 and are similar to those presented in the previous sub-section.
The gain of the antenna was measured as 3.5 dBi across the relevant band of
frequencies. It should be noted that for all measurements presented in this
paper, the ground-plane was large, 50 x 50 em. A further improvement in the
impedance behavior may be achieved similar to the cases for a monopole and
PIFA [12] if the handset was included in this study. The overall dimensions
of the microstrip antenna presented here are 4 x 4 x 2 em, of which can be
reduced by using the technique outlined in Chapter 3 at the slight expense of
bandwidth.

,.
Figure 5.6.6 Measured Eo Radiation Patterns at 1.9 GHz

5.6-2.5 Small Square Dual Spiral Printed Antennas

In Section 5.5, small shorted spiral printed conductors were


presented. A significant reduction in real estate was achieved by maximizing
the current path for the printed antennas. The major drawback for the antenna
configurations presented was that it exhibited very narrow bandwidth,
typically less than 1% making it not suitable for most wireless application. ' In
this section, we present a small printed antenna that consists of two shorted
interleaved spiral printed radiators.

The antenna configuration is shown in Fig. 5.6.7. Here, the spiral


patch combination has an overall length L and width Wand is printed on a
substrate of thickness d, with a relative dielectric permittivity s, The two
spirals are interleaved as shown in Fig. 5.6.7 to reduce the overall area
occupied by the antenna. Each of the spirals has a printed conductor width W f
and contains a shorting post positioned at (xpsh ypsi) relative to the center of
the overall area (L x W) of the patch. Each spiral has its shorting post in close

237
proximity to the feed pin to ensure that the size of the conductor is reduced as
mentioned before. The two spirals are located close to each other with the gap
spacing Wg between the radiators being 0.3 mm. It is interesting to note that
by decreasing We, the resonant frequency of the antenna decreases; however,
the limit of this reduction is set by the diameter of either the shorting pin or
the feed pin.

+- L -..

I
r

( ~-r.,
( • p.'

w -~1
~ I .+
~y )1
~ (,d.y~)-

1 ~I w
t
1. w,
T
Figure 5.6.7 Interleaved Shorted Printed Spiral Antenna

As mentioned previously, the key to reducing the size of the antenna


is to maximize the current path of the printed conductor. We have found that a
spiral incorporating a single shorting post and probe-fed achieved a
significant reduction in size. The bandwidth for a single shorted spiral is
however very narrow, typically less than 2 % making it not suitable for most
applications. To achieve a return loss bandwidth comparable to [32], we
propose an interleaved spiral configuration antenna. The concept of this
antenna is based upon the principal of any dual resonance antenna, namely,
that if a mutual resonance can be formed, the overall impedance variation of
the antenna will be minimized enhancing the impedance bandwidth of the
shorted patches. A thin layer of Taconic (TLC30: s, = 3, d = 0.5mm) was used
to etch the patch conductors.

To couple power into the antenna, the printed conductors were


connected by probes (xpsit ypsi) relative to the center of the overall size L x W,
onto the feed network located beneath the ground plane. This is done here by
connecting pins through the dielectric material from the feed network ports
onto the antenna ports labeled (xpit ypi)' The printed spirals were designed for
an input impedance of 50 Q, so transformers were required on the feed
network to transform the input port into the radiators. A dielectric constant

238
material (RT/Duroid 6010, e, = 10.2) was used for the feed network to
maintain the small size of the antenna. A photograph of the manufactured
dual spiral printed antenna is shown in Figure 5.6.8.

t,

Figure 5.6.8 Photograph of Small Dual Printed Spiral Antenna

Figure 5.6.9 shows the measured return loss bandwidth of the


proposed shorted spiral antenna. The 10 dB return loss bandwidth of the
antenna is 9.2%. This value is slightly higher than a standard circular and
rectangle probe-feed shorted patch mounted on the same substrate, 6.7% and
7% respectively. The larger bandwidth is due to the previously mentioned
mutual resonance phenomenon between the spirals that is evident in the plot
of Figure 5.6.9. To put these results into perspective, for a fixed frequency,
the proposed spiral antenna is 28% smaller in physical length compared to a
shorted circular patch (radius = 7 mm). A greater reduction is achieved (-
33%) when compared to a standard rectangular shorted patch (15 x 10 mm)
using the same dielectric materials. An interesting factor to note is that
although the antenna is smaller, the bandwidth has not been compromised due
to the two radiators interacting. It should be further noted that a larger
reduction in size could have been readily achieved, however at the expense of
bandwidth and ease of fabrication. The far field radiation patterns were
measured at 2.35 GHz. It can be observed from Fig. 5.6.10 that the antenna
has an omni-directional pattern (note that broadside is 0 degrees in this plot).
The cross polarization levels evident in Fig. 5.6.10 are typical for electrically
small printed antennas mounted on small ground-planes [36]. Here the
ground-plane extends 8 mm past the printed spiral conductors. The measured
gain of the antenna was 0.2 dBi.

239
!g"0
.
o

-- r-,
\
I
I
/
,;
,2·15 \
E
\v

"
e".20
·25

-30
1.8 1.9 2.1 2.2 2.3
V
2.4 2.5
frequency , GHz

Figure 5.6.9 Measured Return Loss of Shorted Dual Spiral Printed Antenna
(parameters: e, =1.07, d = 10 nun, L =10 nun, W =9.5 nun, WI =0.6 nun, wg =0.3 nun, TO =
0.3 nun, ro. =0.15 nun, xpJ =-2 nun, YpJ =0 nun, X p2 =2 nun, Yp2 =0 nun, xpsJ = 1 nun, YpsJ =1,
x psl = -1 111Ip, Yps2 = -1)

110

Figure 5.6.10 Measured Radiation Patterns of Dual Spiral Printed Antenna at 2.3 GHz

5.6-3 Reducing Cross-polarized Radiation of Shorted Patches

5.6-3.1 Balanced Shorted Patch Antennas

A fundamental problem of the shorted patch is the high level of cross-


polarized fields generated in the H-plane (cjl = 90j. For applications such as
mobile communications base station antennas where polarization purity is
important (as opposed to handset terminals where it is not an issue), the
shorted patch in the states presented cannot be used.
240
A common technique to reduce the magnitude of the cross-polarized
fields of a conventional microstrip patch is to utilize a balanced feed [9]. This
approach can also be applied to a shorted patch configuration. Figure 5.6.11
shows a schematic of the balanced fed shorted patch. Here a circular
microstrip patch of radius, R is excited by two probe feeds at symmetrically
0
opposite locations (Xph ypl) and (X p2, yp2)' Here feed 2 is excited 180 out of
phase with respect to feed 1. The patch is also loaded by two shorting pins
located in close proximity to the two feeds at (Xpsh ypsl) and (Xps2, YPS2),
respectively . The balanced shorted patch is mounted on a grounded substrate
of height, d and dielectric constant, e,

I<--_-+-~x

Figure 5.6.11 Schematic of Balanced Shorted Microstrip Patch

To qualitatively understand the basic impedance behavior of a shorted


patch and therefore its design and the subsequent design of a balanced
configuration, simple circuit theory is appropriate . For the case of a probe fed
microstrip patch incorporating a single shorting post, the typical tank RLC
network representation is loaded with a series LC circuit . The LC
combination represents the effect of the shorting pin on the input impedance
of the antenna. When the shorting pin is located near the feeding pin, a strong
capacitive coupling or loading results. This effect counters the usual
inductive nature of a patch below resonance. Thus depending on the amount
of capacitive coupling, the size of the shorted radiator can be dramatically
reduced. As the shorting pin is moved away from the feed, the capacitive
coupling decreases and the feed sees the shorting pin as an inductive
reactance, similar in form to that typically used to represent the effect of the
feed itself. This inductive effect increases the resonant frequency of the
shorted configuration, or increases the size of the patch conductor for a
constant operation frequency . This postulation is also valid for a shorted
patch incorporating a second post located on the opposite side of the patch to
that of the feed. Hence the presence of the second shorting post and related
feed, required to achieve a balanced configuration, must be taken into account
when designing the shorted balanced radiator. The consequence of one feed
241
seeing the other shorting pin is to increase the physical size of the shorted
patch.

To design the balanced configuration a full-wave spectral domain


analysis was used [25]. Here attachment modes are used to accurately model
the currents on all the feed and shorting pins as well as the discontinuities
created by the connecting these pins to the patch conductor. Unlike the
conventional patch with a balanced feed, where the design of the balanced
configuration is a simple extension of the single feed, here the presence of the
second shorting pin must be taken into account.

Consider the case where the printed antenna is to be mounted on foam


(e, = 1.07) of height 5 mm. To achieve an input impedance of approximately
50 Q at 1.9 GHz at each driven port, the patch radius must be 20.6 mm, and
the shorting posts located at Xpsl = 15.8 mm and the feeds positioned at Xpl =
18.6 mm. The dimensions of a conventional shorted patch designed for the
same specification are: R = 12.52 mm, xp = 9.2 mm and xps = 10.52 mm.
There are two important ramifications of the balanced design: the patch
conductor size has increased, by a factor of 2.7; and the gap between the
shorting pin and feed pin has also significantly increased, by a factor of 4.7.
The patch conductor size is now approaching that of a traditional quarter-
wave patch . Associated with this increase in size is an increase in bandwidth
(comparable to that of a conventional microstrip patch, mounted on the same
material) and gain. The second ramification implies easer fabrication of the
antenna .

The S-parameters of the balanced shorted patch were measured using


a Wiltron 3608 network analyzer. Figure 5.6.12 shows the predicted and
measured S11' As can be seen from this Smith Chart, very good agreement
between the results was achieved . The slight discrepancies can be attributed
to a thin layer of dielectric required to etch the patch conductor. In this
instance the thin layer is RT-DuToid 5880 with a thickness of 0.254 mm and a
dielectric constant of 2.2. The 10 dB return loss bandwidth measured at each
feed port is 4.4%, compared to a theoretical value of 4.2%.

Figure 5.6.13 shows the predicted and measured E- and H-plane co-
polarized radiation patterns for the balanced shorting post patch configuration,
at 1.9 GHz. To provide the 1800 phase difference between the two feeds, a
Norsal 4114 coupler was used. As can be seen from this graph good
agreement between theory and experiment was achieved. The measured
cross-polarization levels were greater than 20 dB below the co-polar fields in
both planes. Theoretically, the cross-polarized fields should be more than 100

242
dB lower, however, this discrepancy could be due to diffraction from the
finite ground plane (in this case 7cm2) . It should be noted that the measured
values in the H-plane are still more than 20 dB below that for a typical shorted
patch. The theoretical gain of the balanced configuration is 5.0 dBi compared
to 3.5 dBi for a typical shorted patch mounted on the same material.

Figure 5.6.12 Predicted and Measured SI1 of a Balanced Shorted Microstrip Patch
(parameters: e, = 1.07, d = Smm, R = 20.6mm, Xpi = lS.8mm, Xpsi = 18.6mm, Xp2 = -lS.8mm,
Xps2 = -18.6mm, rOi = r02 = 0.32Smm, rOsi = rOs2 = O.6mm)
o

270 1---r--~~~_.c::+:-:----l-- -l 80

1110
Figure 5.6.13 Predicted and Measured Radiated Fields from the Balanced Shorted
Microstrip Patch at 1.9 GHz

5.6-3.2 Compact Dual Concentric Ring PrintedAntennas

One issue with the previously considered balanced configuration is


the increase in patch conductor size: the real estate had increased from an

243
eighth of a wavelength for a conventional shorted patch using a single
shorting pin to approximately a quarter-wavelength. The increase in antenna
real estate was attributed to each excited probe 'seeing' the shorting post used
for the other feed. This 'inductive' loading somewhat cancels the 'capacitive'
effect of the closeness of the excited probe and its associated shorting post.
The more shorting pins located on the patch conductor the greater the size of
the patch. Therefore a CP version utilizing 4 shorting pins and feeds
(required to ensure good cancellation of cross-polarized fields) would give
very little size reduction compared to a conventional microstrip patch antenna.
A key problem here is that the shorting pins must be electrically isolated if
more than one is to be used on one patch conductor.

This electrical isolation issue can be overcome using a dual


concentric shorted ring printed antenna. Each ring has a localized/truncated
dielectric layer below it to give control of its resonant frequency and therefore
enabling a relatively broadband small printed antenna. Importantly too, the
proposed antenna allows for polarization control depending on the feed
network and phase distributed to the excitation ports of each antenna. Using
such a concentric ring configuration provides the spatial symmetry necessary
for a 2Dd order synchronous subarray. The shorted patch geometry can be
extended to a general multi-ring configuration having n dielectric materials
and rings to provide more flexibility if desired.

A generalized version of the proposed printed antenna is shown in


Figure 5.6.14. Here, a shorted printed annular ring antenna of inner radius al
and outer radius a2 is etched onto a dielectric material with dielectric constant
frl' Please note the substrate is truncated at the inner and outer edges of the
ring. A second annular ring of inner radius b l and outer radius b2 is
concentrically located outside the inner annular ring and below the second
ring is a dielectric layer with a dielectric constant of Ea. As with the first ring,
and as is depicted in Figure 5.6.14, the substrate for the second ring is
truncated such as not to extend beyond the conductors of the patch in the
lateral directions. Each ring is fed by a coaxial probe of radius roi located at
(xpi. ypi) from the geometrical center of the ring and has a shorting post
located in close proximity to the feed positioned at (xpsi. ypsi) relative to the
center of the antenna. As in all cases of shorted patches the close proximity
of the shorting pin to the feed is to reduce the overall size of the antenna. A
small gap (rg) exists between each of the concentrically located rings. The
thickness of all the substrates used here is d, as shown in Figure 5.6.14,
however a more generalized version of the proposed antenna could have this
parameter as another variable.

244
y
0..
2
h. III
X

d
Er2 Er l

Figure 5.6.14 Schematic Diagram of Concentric Shorted Printed Rings

The fundamental reason as to why different dielectric materials are


used under each shorted printed ring is to ensure that the resonant frequency
of each radiator can overlap. It is not possible to achieve this mutual
resonance between the lowest order modes of two concentric rings if the same
dielectric material is utilized under both rings. Thus the inner ring must have
a higher dielectric constant substrate under its conductor, compared to the
outer ring.

To accurately predict the performance of the proposed antennas


would require a numerical analysis tool that can support truncated substrates
such as a Finite Element Method or Finite Difference Time Domain
procedure. Such sophisticated tools were not utilized here. As a starting
point, the inner annular ring was designed using Zealand IE30™ [44]
assuming an infinite grounded substrate without the presence of the outer ring
or its associated dielectric material. The design procedure outlined in [45]
was adopted. After a reasonably well-matched design was achieved, the outer
ring was then designed to operate at a similar frequency. Such an approach
will ensure that the bandwidth of the entire small printed antenna will be
enhanced once the patches are incorporated together. As mentioned before,
the inner ring of the antenna has a higher dielectric constant than the outer
ring (in the cases considered Erl = 2.2 and Er2 = 1.07). Importantly too, a gap
must exist between the two rings to ensure the problems outlined in the
previous sub-section are not encountered when more than one shorting pin are
located on the same electrical conductor. The design of the radiating elements

245
for both the balanced feed configuration and the CP version of the shorted
concentric rings shared this common procedure.

Figure 5.6.15 shows the balanced dual concentric rings printed


antenna and it's associated feed network. The antenna consists of an inner
shorted annular ring microstrip patch mounted on RT/Duroid 5880 (e, = 2.2)
with a thickness of 10.3 mm. A second shorted annular ring (outer patch) is
mounted on Rohacell foam (Er = 1.07) with a thickness of 10 mm, A thin layer
of RT/Duroid 5880 (thickness of 0.254 mm) was used to etch the outer ring
patch conductor. The gap between the concentric rings is approximately 1
mm. The shorted patches are probe-fed at symmetrically opposite locations as
shown in Figure 5.6.15. The feed network is located beneath the ground plane
of the antenna and shares this ground-plane. To minimize the size of the feed
network, a high dielectric constant feed substrate was used (RT/Duroid 6010,
e, = 10.2). A relatively thick feed substrate was chosen (2.5 mm) to ensure
that the widths of the feedlines required were not too small, thereby
improving the robustness of the proposed antenna. The feed pins, labeled port
1 and port 2 in Figure 5.6.15, were connected through the feed material to the
aritenna ports labeled (Xph 0) and (X p2' 0), respectively. It should be noted that
port 2 is excited 1800 out of phase with respect to port 1 to ensure a balanced
configuration. The extra phase was achieved using a simple extended
microstrip line, as can be seen in Figure 5.6.15. Also apparent in this figure
are the impedance transformers necessary to ensure that the antenna is well
matched at its input port. Figure 5.6.16 shows a photograph of the
manufactured balanced dual concentric ring printed antenna.

1 /
Port 2

Figure 5.6.15 Schematic Diagram of Balanced Dual Concentric RingPrintedAntenna


andFeed Network

The measured return loss bandwidth of the proposed balanced dual concentric
ring antenna is shown in Figure 5.6.17 (refer to the figure captions for the
locations of these probes as well as the appropriate shorting pins and the
246
dimensions of the rings). The 10 dB return loss bandwidth of the antenna is
8.9%. It should be noted that the simulated results for the inner ring alone is
4.4% whilst for the outer ring a bandwidth of 3.6% was simulated. The
proposed antenna has a higher bandwidth due to the concept based upon the
principal of any dual resonance antenna, namely that if a mutual resonance
can be formed, the overall impedance variation of the antenna will be
minimized, enhancing the bandwidth of the antenna. Importantly too, the
proposed antenna has a larger bandwidth than that presented before, namely
4.4%.

Figure 5.6.16 Photograph of BalancedPrinted Dual Concentric Shorted Rings


o
- -.....
<, I
/'

\
-5

,/
<> \
~10
rii
III
.915
\
E
.a
!!20

-25

-30
1.B 1.9 2 2.1 2.2
frequency, GHz

Figure 5.6.17 Measured Return Loss of Balanced Dual Concentric Shorted Ring Antenna
(Parameters: al =10 mID, az =7 mID, £,.. =2.2, d = 10.3 mID, bl = 14 mID, bz =11 mID, E,.z =
1.07, d = 10 mID, rol =0.2 mID, rosl =0.3 mID, roz =0.325 mID, fosZ =0.6 mID, Xpl =7.5 mID, Ypl
=0 mID, Xp2 =- 11.5 mm, Yp2 =Omm, Xpsl =9.5 mm, Ypsl =0 mm, Xps2 =-13.3 mm, Yps2 =0
mm)
247
Figure 5.6.18 shows the radiation patterns for the balanced dual
concentric shorted ring antenna measured at 2.1 GHz. In Figure 5.6.18, 0°
corresponds to broadside. The measured cross polarization levels were
greater than 15 dB below the co- polarized fields in both E- (tP = O~ and H-
planes (tP = 90~. A significantly lower cross polarization level was achieved
for the H-plane measurement, and this discrepancy may be due to the
presence of the connectors and cables used to record the radiation pattern not
being calibrated out of the measurement. These cables will have a more
adverse effect when scanning in the E-plane. As can be seen in Figure 5.6.18,
the patterns are fairly symmetrical around broadside. It should be noted that a
greater degree of cross-polarized field reduction could have been achieved if a
larger ground plane was used; here the ground-plane was only 7 em x 5 em.
The measured gain of the antenna was ~. O dBi.

180

Figure 5.6.18 Measured Radiation Patternof BalancedDual Concentric Ring Antennaat


2.1 GHz

5.6-4 Small Circularly Polarized Microstrip Patches

5.6-4.1 Introduction

As mentioned in Chapter 2, the advantage of using a sequential


rotation technique to achieve CP is its wide impedance and axial ratio
bandwidth. As mentioned before, the array typically utilizes linearly polarized
elements rotated sequentially to achieve spatial symmetry and fed with the
appropriate phases to each of the antennas. One of the drawbacks of using this
248
method is the overall size of the antenna. Once again, the size of the antenna
is critical to have an aesthetically pleasing GPS handset. Therefore, in this
section, we will present three antenna configurations utilizing the sequential
rotation technique to achieve CP and most importantly, these antennas are
electrically small and compact in size. The antennas considered include using
a 2 x 2 synchronous subarray of rectangle shorted patches, 3 shorted
equilateral triangles and dual concentric rings printed antenna. As will be
shown later in this section, the proposed antenna configurations demonstrate
the impedance and axial ratio bandwidth satisfies the GPS requirements and
are small in size.

5.6-4.2 4 elementsynchronous subarray ofshortedpatches

As mentioned before, one of the CP printed antennas utilizes a 4


element synchronous subarray of shorted patches. It has been shown in the
past that good axial ratio bandwidths can be achieved using a synchronous
subarray of elements that individually do not radiate CP [46]. Figure 5.6.19
shows a schematic diagram of the top view of the proposed antenna. Here,
each of the 4 rectangular shorted patches of length, L and width, W are fed by
a probe located at (Xph Ypi) relative to the center of the corresponding ilb patch.
A shorting pin is located at (xpsi> ypsi) relative to the center of each patch. The
subarray is mounted on grounded substrate with a dielectric constant, e, and a
thickness, d.
~

s~ortin1
pm
+~
[w,
L3
Feed pin

Figure 5.6.19 Schematic Diagram of 4 Element Synchronous Subarray of Shorted


= = = = = = = =
Patches <Parameters: L; 25 mID, Wi 17 mID, Xpl X p2 Xp3 xp4 -10.0 mID, Xp.1 Xp.2
Xps3= =
X ps4 = =
-10.0 mID, Ypi Yp.i 0)

249
To ensure the overall size of the antenna is small, the 4 shorted
patches are located in close proximity to each other, with the gap between the
patches approximately 1 rom. It should be noted that the 4 patches must not
be in direct contact as this will substantially increase the size of the antenna
for a set operation frequency due to the inductive loading effect of each
additional shorting pin. To maximize the impedance bandwidth of the shorted
patches a low dielectric constant, thick substrate (Rohacell: Er = 1.07, d = 9
rom) was utilized. A thin layer of RT/Duroid 5880 (e, = 2.2, d = 0.254 rom)
was used to etch the patch conductors.

Figure 5.6.20 shows a schematic diagram of the feed network for the
synchronous subarray. As can be seen in this diagram, additional lengths of
microstrip line are used to provide the required 900,1800 and 2700 phase shifts
at 3 of the excitation ports (labeled as 2 - 4 in the layout). Also apparent in
Figure 5.6.19 are the impedance transformers needed to ensure the antenna is
well matched. To minimize the size of the feed network, a high dielectric
constant substrate was used (RT/Duroid 6010, Er = 10.2). A relatively thick
feed substrate was chosen (d = 1.9 rom) to ensure the widths of the feedlines
required were not too small, thereby improving the robustness of the proposed
antenna. To connect the patch conductors to the feed network, pins of radius
0.45 rom were used. Similar pins were soldered to the 4 patch conductors and
the shared ground-plane between the feed and the printed antenna to provide
the required short circuits. Finally a SMA connector was soldered to the input
port shown in Figure 5.6.19. Both antenna and feed substrates were truncated
such that the area of the CP antenna was 4.5 x 4.5 em. A photograph of the
manufactured 2 x 2 synchronous subarray is show in Figure 5.6.21.

2
3

1
4

Input port

Figure 5.6.20 Layout of feed network for 4 Element Synchronous Subarray of Shorted
Patches
250
Figure 5.6.21 Photograph of 4 Element Synchronous Subarray of Shorted Patches

The measured return loss of the proposed CP antenna is presented in


Figure 5.6.22. The 10 dB return loss bandwidth of the antenna is 8.5 %. This
value is slightly higher than predicted for a single shorted patch antenna
fabricated on the same material (6.5 %) due to the mutual coupling between
the elements reducing the impedance variation of the antenna, as well as the
effect of the phase shifts in the feed network canceling unwanted reflections.
The axial ratio of the printed antenna as a function of frequency is also shown
in Figure 5.6 .22. As is evident from this plot, the 3 dB axial ratio bandwidth
is greater than the impedance bandwidth of the antenna, here 11.3 %. This
can be attributed to the fact that the bandwidth of the synchronous feed (in
terms of providing the correct phase and amplitude to each antenna port) is
greater than the antenna elements. The 3dB axial ratio beamwidth was
greater than 120° for each frequency across the 10 dB return loss bandwidth
and the gain was greater than 2.5 dBc.
o

--- -, V- r---
10
9

I
·5 8

m
~ \ -, 7

I
-10 6
:l
.9 ~
~
5
E 4 0"
·15
a:
\j
::>
iii ~
a: 3 ~
-20 2

~ 0
1.1 1.15 1.2 1.25 1.3 ' 1.35 1.4 1.45 1.5
Frequency (GHz)

Figure 5.6.22 Return Loss and Axial Ratio of 4 Element Synchronous Sub array of
Shorted Patches

251
It should be noted that there are techniques that can be incorporated to
further reduce the size of the proposed antenna including placing 'slits' in the
printed conductors (as in [47]) and also deforming the printed conductor
around the handset (as in [48]). If the symmetry of these configurations can
be maintained, then similar results to those presented in Figure 5.6.22 should
be expected.

5.6-4.3 3 element synchronous subarray ofshorted patches

It is interesting to note that of the many sequentially fed techniques


(for example, 2 elements fed 0° and 90° respectively, and so on), 3 elements
fed 0°, 120° and 240° respectively gives better VSWR and axial ratio
bandwidth that the other configurations [40]. The only possible drawback
pointed out in [40] of the 3 element synchronous subarray appropriate to the
application here is the more complicated feed network to give the required
power split.

Consider a small printed antenna consisting of three shorted


equilateral triangle patches sequentially rotated and excited by 0°, 120° and
240°, respectively. The feed network providing the correct phasing is located
below the ground-plane of the shorted patches and is etched on a high
dielectric constant material to reduce its size. A schematic diagram of the
sequentially fed subarray of shorted equilateral triangles is shown in Figure
5.6.23. Shorted equilateral triangles were chosen as the radiating element due
to the smaller size than rectangular and circular patches.

I
I. <x.,3· YpJ)
'-'J .K
(xpo!'
\
Ypa)
';..
L
2

I <J,yps3) ( t ~
IA~ ~'YP2 : ~

(~I·Ypol)
~ «,r- Yp , )
.K
- - - L 1- - -
x
Figure 5.6.23 Schematic Diagram of 3 Element Synchronous Subarray of Shorted
Triangular Patches

252
Here, each of the 3 equilateral triangle patches of length, L is fed by a
probe of radius ro located at (Xph ypi) relative to the center of the ilb patch (i =
1, 2, 3). A shorting pin of radius ro s is positioned (Xpsh ypsi) from the center of
each of the triangular shaped patch conductors. The subarray is mounted on a
grounded foam substrate (e, = 1.07) to maximize the bandwidth of the shorted
patch configuration [25]. The height of the substrate (d) is 10 mm, A thin
layer of Taconic TLC30 (e, = 3.0, d = 0.5 mm) is used to etch the patch
conductors. The patches are sequentially positioned and fed in a 0, 120 and
2400 phasing arrangement and so the entire antenna demonstrates symmetry
about boresight. A small gap (1 mm) exists between the apexes of the
triangles. The feed network is similar to that presented for the 4 element
synchronous subarray, although here a 3-way power splitter was developed to
ensure the correct magnitude distribution to each port. As for the 4 element
synchronous subarray, to achieve the required phasing to the second and third
patches, additional lengths of microstrip line were used. To minimize the size
of the feed network, a high dielectric substrate was used (RT/Duroid 6010, e,
= 10). A relatively thick substrate (2.5 mm) was chosen to ensure the widths
of the feedlines are relatively thick, thus making the antenna robust. In a
similar arrangement to that given for the 4 element synchronous subarray,
pins of radius 0.6 mm were used to connect the short patches to the feed
network. Shorting pins (ros = 0.325 mm) were soldered between the patch
conductors and the common ground-plane. An SMA connector is the input to
the feed network and the substrates of both the antenna elements and the feed
were truncated to give an antenna with the dimensions of 4.5 em x 4.5 ern. A
photograph of the manufactured 3-element antenna is presented in Figure
5.6.24.

Figure 5.6.24 Photograph of 3 Element Synchronous Subarray


253
The measured return loss of the proposed 3 element synchronous
subarray is shown in Figure 5.6.25. The 10 dB return loss bandwidth of the
configuration is 31 %. This is significantly greater than that recorded for the
4 element synchronous subarray, namely 8.5 %, although a direct comparison
is probably not appropriate as thicker materials relative to the operating
frequency were used here. Despite this, the bandwidth is remarkably large
for such a small antenna. These results are however consistent with the
findings in [40], that is, 3 element synchronous subarrays improve the VSWR
bandwidth by a factor of 2 or so essentially due to mismatch cancellation.
The axial ratio of the antenna at boresight was measured from 1.6 - 2.5 GHz
and the results are also displayed in Figure 5.6.25. The measured 3 dB axial
ratio of the printed antenna is approximately 27.5 %. Once again the
somewhat larger than expected AR bandwidths are consistent with the
findings in [49] where it was postulated that the multiple reflections set up by
the patches and feed mismatches radiate mainly reference polarization. It
should be noted that the AR and impedance bandwidths do not exactly
overlay, which is typical for most CP antennas. The common bandwidth is
approximately 22 %.
o
- '\
J r\ /
10

lr- B

! e!
:
-10
v-.. r-, /
.3 -15
I
E
... I
~ ·20
IX:

II 2
-25

~ 0
1 1.2 1.4 U 1.8 2 2.2 2.4 2.8 2.B 3
Frequency (GHz)

Figure 5.6.25 Return Loss and Axial Ratioof 3 Element Synchronous Subarray
(Parameters: ~ = 21 mm, Xpl = 0, Ypl = -8.1 mm, Xpsl = 0, Ypsl = -6.1 mm,Xp2 = 7.0 mm, Yp2 =
4.1 mm,Xps2 = 5.3 mm, Yps2 = 3.1 mm, xpJ = -1.0 mm, YpJ = 4.1 mm, xpsJ = -5.3 mm,YpsJ = 3.1
mm)

Figure 5.6.26 shows a measured spinning linear far-field radiation


pattern of the antenna at 1.9 GHz. It is evident from this plot that the
truncated, symmetrical printed antenna provides excellent AR in most
directions. This characteristic reduces the possible detrimental affect of a user
in close proximity to the antenna on its overall performance [48]. The gain of
the antenna at 1.9 GHz was measured as 3.3 dBc.

254
110

Figure 5.6.26 Radiation Pattern of 3 Element Synchronous Subarray at 1.9 GHz

5.6-4.4 Shorted concentric annular rings generating CP

Thus far in this section we have shown that a small printed antenna
with good impedance and axial ratio characteristics can be developed using a
subarray of shorted patches with sequential rotation. The synchronous
subarray procedure presented earlier was applied to both 4-element and 3-
element shorted patch subarrays and yield 10 dB return loss and 3 dB axial
ratio bandwidths in excess of 15 %. Although good impedance and radiation
characteristics were obtained for these configurations, the overall size of the
antenna was still significantly larger than the individual shorted patch due to
the need for the other elements in the subarray. It should be noted that a 2-
element synchronous subarray has not been considered using conventional or
shorted patches due to the lack of spatial symmetry that is necessary to
achieve good CP qualities.

The concept of utilizing concentric, shorted printed ring antennas was


introduced earlier in this chapter. Here, we present a dual concentric shorted
ring printed antenna to achieve CPo Each ring has a localized/truncated
dielectric layer below it to give control of its resonant frequency and therefore
enabling a relatively broadband small printed antenna. Importantly too, the
proposed antenna allows for polarization control depending on the feed
network and phase distributed to the excitation ports of each antenna. Using
such a concentric ring configuration provides the spatial symmetry necessary
for a 2Dd order synchronous subarray.

Figure 5.6.27 shows a circularly polarized dual concentric rings


printed antenna with the required feed network. As for the balanced
configuration presented earlier, the inner ring is mounted on RT/Duroid 5880
255
and the outer ring mounted on Rohacell Foam. It should be noted that the
design procedure used here is the same as that discussed for the balanced
configuration. As can be seen from Figure 5.6.27 there is spatial symmetry
between the proposed 2nd order synchronous subarray. Figure 5.6.27 also
shows a schematic of the feed network to provide the necessary 90° phase
shift to give CP. Once again, high dielectric constant material is utilized for
the feed to minimize its footprint. The ground-plane for the proposed CP
antenna was 7 em x 5 em, A photograph of the small printed CP antenna is
shown in Figure 5.6.28.

/
Port 2

Figure 5.6.27 SchematicDiagram of CP Dual ConcentricRing ShortedPatch Antenna

Figure 5.6.28 Photographof CP Dual ConcentricRing Shorted Patch Antenna

Figure 5.6.29 shows the measured return loss bandwidth of the


proposed CP dual concentric ring antenna. The 10 dB return loss bandwidth
of the antenna is 13.3%. The enhanced bandwidth compared to the balanced
256
configuration could be due to imperfections in the construction of the two
antennas. The radiation performance of the antenna was also measured across
its matched impedance bandwidth. Figure 5.6.30 shows three measured axial
ratio patterns for the dual concentric shorted rings using a spinning linearly
polarized hom antenna at 1.7 GHz, 1.8 GHz and 1.9 GHz. The 3 dB axial
ratio beamwidth at each frequency was 20°, 70° and 40°, respectively. The
reason for the reduction in 3 dB axial ratio beamwidth at the outer edges of
the frequency band is simply because the axial ratios at broadside at these
frequencies, are higher, 2.5 dB and 2 dB, respectively, compared to 1 dB at
1.8 GHz. The 3dB axial ratio bandwidth of the proposed antenna is 16.2 %.
The somewhat larger than expected axial ratio bandwidths are consistent with
the findings in [46] where it was postulated that the multiple reflections set up
by the patches and feed mismatches radiate mainly reference polarization. It
should be noted that the axial ratio and impedance bandwidths do not exactly
overlay, which is typical for most CP antennas. The measured gain of the
antenna was 2.0 dBc.
o
r-., ~~
,..,...,
-5

-10 i\ I
ID
'tJ
"15
\ / /
.2
520
\ / 1\, I
~
I
",
f·25
·30

-35
1.6 1.7 1.8 1.9 2 2.1
frequency, GHz

Figure 5.6.29 Measured Return Loss of CP Dual Concentric Shorted Ring Antenna
= = = = =
(Parameters: a, 10 mm, az 7 mm, £'1 2.2, d 10.3 mm, b J 14 mm, b z 11 mm, E,z = =
= = = = =
1.07, d 10 mm, rol 0.2 mm, rosl 0.3 mm, r o2 0.325 mm, ros2:: 0.6 mm, Xpl 7.5 mm, Ypl
= = = = = =
0 mm, Xp2 0 rnm, Yp2 11.5 mm, Xpsl 9.5 mm, Ypsl 0 mm, Xps2 0 mm, Yps2 13.3 mm) =
The CP version of the shorted concentric ring antenna is significantly
smaller in size than the other printed subarray antennas mounted on similar
substrates and operating at the same frequency band. Table 5.6.1 gives a
comparison between the three configurations, in terms of size, impedance
behavior and radiation characteristics.
257
110 110

(a) (b)

110

(c)

Figure 5.6.30 Spin Linear Radiation Pattern of CP Printed Antenna: (a) Measured at 1.7
GHz; (b) Measured at 1.8 GHz; (c) Measured at 1.9 GHz

Synchronoussubarrays Size (LxW) Impedance Axial


Bandwidth (%) Bandwidth (%)
4 elements 44mmx44mm 8.5 11.3
3 elements 38 mmx38 mm 31 27.5
2 elements 28mmx28mm 13.3 16

Table 5.6.1 Comparison of Synchronous Subarray Techniques

5.6-5' Improving the radiation efficiency of a shorted patch

5.6-5.i introduction

It was shown in Section 5.2-2 that significantly better impedance


bandwidth and radiation performance for shorted patch antennas can be
achieved when the antenna is mounted on foam substrates rather than on

258
microwave laminates. However, using such a material (foam) jeopardizes the
ease of manufacturing and the robustness of the antenna, contrary to the often-
touted advantages of printed circuit technology.

Like most patch antennas, the performance of a . shorted patch is


compromised as the dielectric constant of the material is increased. One
reason can be attributed to the excitation of surface waves. As was discussed
in chapter 4, these electromagnetic waves propagate along/near the surface of
the substrate and are radiated to free space at the truncation of the dielectric
substrate. The radiated surface waves can significantly distort the radiation
pattern of the antenna. Several schemes can be found in the literature to
suppress these surface waves including the use of photonic bandgap (PBG)
structures, as shown in Chapter 4.

Several types of antennas with these PBG structures as ground planes


have been investigated, including vertical monopoles, horizontal wire
antennas and the conventional microstrip antenna [50 - 51]. However, a
shorted patch antenna has not been investigated. In this section we present
two shorted patch antennas fabricated on the same microwave substrate, RT
Duriod 5880 (e, = 2.2). This material was specifically chosen as it is a
common substrate used in microstrip antennas operating in the S-band. One
shorted patch has been manufactured with a high impedance electromagnetic
surface surrounding the patch conductor while the other is without, in order to
ascertain the usefulness of the proposed antenna. It will be shown that the
radiation performance of the new antenna has been dramatically changed,
yielding a printed antenna with radiation characteristics similar to a
conventional patch antenna, but a fraction of the size. The two shorted
microstrip patch configurations have been investigated theoretically, however
for the sake of brevity only measured results that include input impedance,
radiation patterns and gain are presented.

5.6-5.2 Antenna and psa Configuration

Figure 5.6.31 shows a schematic diagram for the two shorted patch
antennas. The antenna consists of a coaxial probe of radius r, located at (xp,
yp) and two shorting posts of radius ros located at (Xpsh YPls) and (Xps2' yps2)
from the center of the patch conductor. The shorted patch antenna has been
constructed with two shorting posts as it reduces the fabrication tolerance
between the probe and shorting pins, as was shown earlier this chapter. The
patch conductor has a length L I and width WI shown in Figure 5.6.31a. In
order to investigate the effectiveness of PBG structures on shorted patch
antennas a second antenna shown Figure 5.6.31b was constructed. This
shorted patch antenna has identical dimensions as the previously mentioned
259
shorted patch only it is surrounded by a PBG structure. Both shorted patch
antennas are mounted on a grounded substrate of thickness d and dielectric
constant Er (not shown in Figure 5.6.31). The PBG structure surrounding the
shorted patch antenna consists of periodic square metallic pads of length L 3•
width W3 and a lattice constant a, located on the same surface as the patch
conductor. Each metallic pad has a via in its center, with a radius rOY
connecting the pad to the ground plane. The presence of the nearby metallic
pads and vias tend to raise the resonant frequency of the shorted patch antenna
since the effective cavity volume is reduced [50]. Given that it is important
the resonant frequency of the antenna remain unchanged in order that the
antenna operates within the band gap of the structure, a small ring of bare
substrate around the patch conductor was incorporated [51]. Thus the metallic
pads are separated from the patch conductor by a distance gx (r-direction) and
gy (y-direction) as shown in Figure 5.52b. It is important to note that the
overall dimensions (L, x W2 X d) for both shorted patch antennas are the
same. For more detail on the PBG design refer to Chapter 4.
)' 1

(a) (b)
Figure 5.6.31 Schematic diagram of (a) Conventional Microstrip ShortedPatch Antenna
and (b) proposed paG shortedpatch.(Configuration 1 parameters: e,= 2.2, d = 9.51 mm,LJ =
12 mm,WJ =6 mm, fa = 0.6 mm, f oo =0.6 mm, Xp = 3 mm, Yp = 0 mm, XpaJ= 5 mm, YpsJ =1.5
mm, Xpa2 = 5 mm, Yp02 = -1.5mm, Lz = 69mm, W2 = 69mm) (Configuration 2 parameters: e, =
2.2, d = 9.51 mm, L = 12 mm, W = 6 mm,r, = 0.6 mm, ro• = 0.6 mm, x p = 3 mm, Yp = 0 mm,
xpsJ = 5 mm, YpoJ = 1.5mm, Xps2 = 5 mm, Yps2 = -1.5 mm, Lz = 69mm, W2 = 69mm, L, = 9mm,
W3 =9mm, a = 10 mm, rOY =0.5mm, gx =2.5 mm, gy =3.5mm)

5.6-5.3 Results and Discussion

Figure 5.6.32a shows the measured return loss (SJI) of the probe-fed
shorted patch antenna on an ordinary metallic ground plane and the other with
the PBG structure. Although the presence of the PBG has raised the resonant
frequency of the antenna slightly, it still remained within the bandgap and so
260
no measures to compensate for the increase in resonance was undertaken. The
measured 10 dB return loss bandwidth for the antenna surrounded by the PBG
structure is narrower compared to the antenna on the conventional ground
plane. The bandwidths are 3.1% and 5.9% for the shorted patch with the PBG
ground plane and the conventional shorted patch, respectively. This is a
characteristic common to an antenna with higher efficiency as surface waves
(or any loss for that matter) tend to mask/improve the return loss of the
antenna. On the other hand, a significant improvement in the broadside gain
and gain bandwidth were observed as shown in Figure 5.6.32b. The maximum
gain for the ordinary shorted patch is 1.2dBi at 2.97 GHz with a IdB gain
bandwidth of 1.3 %, while the maximum gain for the shorted patch antenna
with the PBG ground plane is 4.4 dBi at 2.9 GHz with a 1 dB gain bandwidth
of 4.5 %. This gain enhancement is equivalent to a 109 % increase in effective
radiated power (ERP) at broadside.
o
.. ,
. . .
,

. . ('
1\
,
,
H ~"- '==WIUtPBO I v VV\
,
,
, y
,,
JV , I~
... ..
'. ~1
... ,.tett .5 0 i '
.~-:.I .t I.
• 8hor1ed
~ hloh wlh PSO
~ I. ,, ':
"t\:
'!. V\(
·1
~
.
I~· ·"",,~, "
"
t-
-2
";:'.:\ fJ' -. ".:
-3

-<lO -4
\f"
2.8 2.7 2.8 U
Frequency (GHz)
, 3.2 :1.8 2.7 2.8 2.' ' .1
Frequency (GHz)

(a) (b)
Figure 5.6.31 Comparison of Performance of Conventional Shorted Patch and PBG
Based Shorted Patch: (a) Return Loss; (b) Gain

Figure 5.6.33 shows the measured E- and H- plane normalized


radiation patterns for both shorted patch antennas. The measured data was
taken at 2.89 GHz where both antennas have an identical return loss of -21
dB. The co-polar radiation patterns in the E- and H-planes show an
improvement in radiation pattern for the shorted patch formed on the PBG
ground plane. Please note that there is a slight normalization error shown in
the conventional shorted patch patterns presented in Figure 5.6.33a,
approximately 0.5 dB. This can be attributed to experimental error associated
with measuring patterns of low gain antennas. From Figure 5.6.33 it can be
seen that the PBG shorted patch has a much smoother radiation pattern than
the conventional form, suggesting a certain degree of surface wave
suppression. This is particularly evident in the large E- plane co-polar ripple
being removed (Figure 5.6.33a). The H-plane cross polarization levels for the
ordinary shorted patch antenna shown in Figure 5.6.33b are extremely high.
261
This is common for shorted patch antennas and is a result of the discontinuity
on the patch conductor associated with the shorting posts and also the image
currents formed on the conventional, electrical ground-plane. The addition of
the PBG has significantly lowered the H-plane cross polarization level.
o 0

180 180

Ant E-p1a"" (eo-polar) AnI H-pla"" (eo-polar)


• - _ . Anl&PBG E-plano (co-polar) • •• •Ant & PBG Hopla"" (eo-polar)
(a) (b)
o o

180 180
AnI E-plo"" (erOM-polar) - Ant H-pI_ (ero_lat)
••• •Ant&PBG E-pI.... (_.o-polor)
- - - . Ant & PBG Hop..... (ero_lar)
(c) (d)

Figure 5.6.33 Measured Radiation Patterns of Conventional Shorted Patch and PBG
Shorted Patch (E-plane: x-z plane in Figure I , H-plane: y-z plane in Figure 1)

5.7 Performance of Shorted Microstrip Patch Antennas for


Mobile Communications Handsets at 1800 MHz

5.7-1 Introduction

The increasing use of radio based portable communications devices


has created public concern about their safety despite reviews finding no
262
substantive scientific evidence of a long-term public health hazard. The
proximity of these devices to the body during their operation leads to highly
non-uniform exposure and so it is inappropriate to specify limits based on
electric and magnetic fields or power density. Limits are therefore specified
in terms of the actual rate of radio frequency energy absorption (specific
absorption rate or SAR) with an averaging mass of 1 g or 10 g of tissue in the
shape of a cube. Government agencies have announced plans to regulate
portable radio transmitting devices based on these limits and this has led to a
need for effective means of demonstrating compliance either through direct
measurement of SAR or through electromagnetic (EM) computational
methods. Presently direct measurement methods are preferred due to the
ability to quantify the accuracy of the system and the difficulties of modeling
the intricate details of individual handsets.

In this section we investigate the performance of a shorted patch


antenna and a stacked shorted patch antenna in terms of SAR and other
aspects. Results are presented at 1800 MHz for three antennas: ')., /4
monopole, shorted patch and stacked shorted patch, in each of three cases: on
a handset in isolation, a handset near the realistic head model and with the
inclusion of a block model of the hand.

5.7-2 FDTD Technique Used

The implementation of the FDTD technique used in the XFDTD code


will not be reviewed as it is described in detail by Kunz and Luebbers [42].
The COST244 handset dimensions, 120 x 42.5 x 22.5 mm, are multiples of
2.5 mm so this was selected as the voxel size of the model that for stability
leads to a time step of 4.12 ps. The antennas were excited with a gap voltage
source specified with a 50Q series source impedance to simulate a real
voltage generator and reduce the number of time steps needed for
convergence. Transient excitation for 3500 time steps using a Gaussian
derivative pulse was used to calculate the wideband input impedance with a
specified pulse width of 410 time steps (1.71 ns) which leads to a frequency
spectrum which is more than 120 dB down at 2.2 GHz and reduces aliasing
effects in the biological materials. Antenna patterns and SAR results were
calculated using a steady state sinusoidal excitation of 1800 MHz and were
run for 4000 time steps (30 periods) to ensure converged results. The antenna
patterns were corrected for the inclined head so that the patterns presented
here represent the antenna on a handset in the intended position of use.
Second order Liao outer radiation boundary conditions were used with a
minimum 20 cell border established around the model.

5.7-3 Intended position of user head model


263
A wide variety of computational phantoms have been used for SAR
assessments of mobile communications type exposures ranging from
homogeneous or layered spheres to millimeter resolution anatomically
realistic models. A 3 mm voxel resolution head and shoulders mesh based on
data from the Visible Human Project was purchased from Remcom Inc. The
mesh was resampled with 2.5 mm voxels and the shoulders were truncated to
fit the available computer memory. The values of the tissue dielectric
parameters at 1800 MHz were derived from recent work. In some cases due
to the seven tissue groupings used in the model the parameters for several
tissue types are averaged to assign to the FDTD mesh materials. The details
of the tissue parameters are shown in Table 5.7.1. The tissue densities are
those provided with the mesh.

FDTD Mesh Tissue Gabriel tissues [52] 1800 MHz

e, cr(S/m) p
(kg/nr')
cartilage cartilage 40.22 1.29 1000
muscle muscle 53.55 1.34 1020
eye sclera, cortex, nucleus, vitreous 50.58 1.49 1000
humour,
nerve, brain cerebellum, cerebro spinal 50.11 1.85 1050
fluid, white matter
skin skin (dry) 38 .87 1.19 1000
fat, bone averaged infiltrated fat and 11.40 0.23 1200
cortical bone
blood blood 59.37 2.04 1000

Table 5.7.1 FDTD Mesh Tissue Group ings and Corresponding Tissues Averaged from
[51] to Provide Dielectric Properties

The CENELEC draft MTE standard specifies the intended or normal


operating position in terms of a reference plane and a reference line. The
reference plane is defined by the auditory canal openings and the center of the
closed mouth. The reference line is defined to be the "tangential" line which
connects the center of the ear piece with the center bottom of the case. The
phone is positioned so that the reference line lies in the reference plane and
with the center of the ear piece in line with the entrance of the auditory canal.
The handset is also positioned such that an angle of 80" exists between the
reference line and the line connecting both auditory canal openings. To avoid
staircasing effects on the handset the head model was rotated forward rotation
by 74° and to the right of 10" to orientate the phone in the CENELEC
intended position of use relative to the head. The handset was then placed on
the right hand side of the head model and aligned with the FDTD mesh.
264
Figure 5.7.1 shows this relationship and the orientation of the coordinate
system.

x~
Figure 5.7.1 Perspective View Showing Effective Orientation of Handset Relative to
Head and Coordinate System

The resampling of the head model for 2.5 rom voxels and the
rotations were done on the original head mesh to minimize anatomical errors,
however any distortions are expected to be within the range of human
variation. The major anatomical features were cross-checked with an atlas of
sectional anatomy before and after the resampling and rotation so that gross
distortions would be detected. The block model of the hand wraps around
three sides of the lower half of the handset and consists of a core of bone
surrounded with a layer of muscle similar to that used in. Details of the hand
model can be seen in Figure 5.7.2.

Muscle

Figure 5.7.2 View of the handset showing the location of the shorted patch antenna and
the dimensions of the block model hand

265
The resulting mesh has dimensions of 147x161x148 (= 3.4million) cells
requiring approximately 105 Mbytes of RAM. The analysis was run on a HP
715/100 with 128 Mbytes of RAM and took about 8 hours for a transient or
steady state excitation. The results for the antennas on the handsets in
isolation used a space of 65x70x120 cells and took just over 1 hour to
calculate.

5.7-4 Monopole and shorted patch models

The COST244 monopole is specified to be A/ 4 with a diameter of


2.5 mm and a feed gap of 2.5 mm. The antenna is positioned in the center top
of the handset and is modeled as a column of perfectly conducting cells as it
has been reported that for the present cell size and the specified antenna
diameter the column approach yields the most accurate simulations. At 1800
MHz, A/ 4 corresponds to an overall antenna length of 16.7 cells that is
rounded down to 15 cells plus gap. The distance between the feed point and
the head is specified as 1.5 cm. The gap voltage source feeds the monopole
on a comer that results in some asymmetry in the calculated antenna patterns.

A shorted patch and a stacked shorted patch presented previously


were modified to suit the rectangular FDTD grid and then analyzed. The
single shorted patch takes the same shape as the stacked version (Figure 5.6.1)
but only has a single dielectric layer and radiating patch. In the general case
of the stacked patch the driven patch of length, LI and width, WI can be
etched on a substrate of thickness, dr and with a dielectric constant, e-i - The
driven patch is fed by a coaxial probe of radius, f 0 located at (yp ,zp) from
the center of the patch. The shorting post of radius, fo s ' positioned at
(yps,zps) is connected to and extends through the first printed conductor and
is soldered to the second patch of dimensions Lz and Wz. The dimensions
for each of the patch antennas are listed in Table 5.7.2. For the stacked patch
antenna a dielectric layer of height, dz and dielectric constant, frZ - separates
the two patch conductors.

5.7-5 Comparison of antenna performance

In this section the principle E and H plane antenna patterns as well as


/slll are plotted for each of the antennas on a handset .in isolation, near the
rotated realistic head model and with the addition of a block model hand. In
order to get a clearer picture of these antennas during use, the antenna patterns

266
with the head present are 'corrected' for the rotation of the head so that they
represent the antenna patterns with the handset in the intended position of use.
The antenna radiation patterns for the antennas on the handset in isolation are
not transformed as it was observed that handsets are generally positioned in
an upright orientation when not in use.

Slnale Patch Stacked Patch


Wl=W2 25mm 15mm
L 1 =Lz 17.5 mm 17.5 mm
dl 10mm 5.0mm
d2 N/A 7.5mm
(YP' Zp) (0,1.25) mm (0 _1.25) mm
(Yps - Zps) (0,3.75) mm (0 _6.25) mm
rO = rO s 0.6mm 0.6mm
erl- sl 1.07, lxl0·4 Slm 1.07, lxl0·4 Slm
er2 ,s2 N/A 2.2, lxl0· 4 Slm

Table 5.7.2 Dimensions of Single Layered and Stacked Shorted Patch Antennas

Examining the azimuth patterns for the antennas on the handset in


isolation, Figure 5.7.3, we see the essentially omnidirectional monopole
pattern, the reason for the slight deviation is the non-central excitation of the
antenna. The two shorted patch antennas exhibit similar patterns with about a
5 dB front-to-back ratio (FIB) for the single patch and 2.5 dB for the stacked
patch. When the head and hand are introduced the patterns, Figure 5.7.3,
become much more complicated with many smaller lobes however on average
each of the antennas offers similar performance.

~ \. ~

~f#" """
tti,
~ ... ~ l:t:e: ~ 't ~ :\ 1"" '\ J!
.D' fU" r'\
,t """ ·'0 ':'; .i .,.
r~ "'Jlf- IJ '11.."
I 1\
"
· 10

...il \ If) j ~
."
II \
\ r, j
-20
'J
·20
v. ~
u

.-.. ."
ri\ r"\
I I)
"'
· 10 ·30
o ~ ~ ~ ~ 00 1m 2~ ~ m D m ~ o » ~ ~ ~ ~ 1m 2W * m m m ~

Df~U DfCl'!'rJ
Pp~po ~ -O-Mcu_ _ po~

....... ..ItJ.. .. clt paid -0- a..ta-t_11t pll C:1lo


....... Epll.Htadlld,..".. M tU-I 16cb4 pak.

(a) (b)

267
./,~ '\

,
0
~ \ r\ )tj~
4 v:
1. -
I ~ \f",
·5

"r
~r<1
(j \If rJ
· '0

T; ~ ~ lJ~ r..,
0;
-e
· 15

-:lll
l~
ft"" ~ \ i
18

-25
r-.t
"V If
· 30
l
o 30 00 so l :lll ' ''' 1110 2'0 24/l 270 3CIl 3JO 360
D.S"'''
Ephi-Ja:lDOpo1e -o- B.bcta-cro. opo1c
- 0 - Epbk .a: 1ep~b -0- &betH.,1e, .ceil
____ Epbktltbd p.tcb -*- Bbcta.. ,.bd p-.h

(c)

Figure 5.7.3 Azimuth (9 =90°) Patterns for Monopole, SingleLayered and Sacked
ShortedPatches on: (a) Vertical handsetin isolation; (b) Handset Adjacent to Head Rotated to
CENELEC normal Position of Use;(c) Including BlockModelHand

In the monopole elevation patterns, Figure 5.7.4a on the handset we


see the downtilted (peak gain of 4.12 dBi at 13(0) and split main lobe which
typically occurs on handsets which are long relative to the wavelength.
Microstrip antennas on infinite ground planes will exhibit almost complete
coverage over 1800 in the broadside direction. When mounted adjacent to the
upper edge of the handset this leads to diffraction of the fields over the top
edge of the handset and splitting of the main beam with a downtilted lobe
(single patch peak gain 3.24 dBi at 1500 , stacked patch peak gain 3.64 dBi at
15(0) similar to the monopole. The patch antenna patterns peak
approximately 4 dB lower than the monopole in the front (side towards the
face) 1800 of the handset which implies less RF current flow on this surface.
The introduction of the rotated head manifests itself in the patterns as a
general reduction in level in the half-space occupied by the head and the
raising of the cross-polar components. The monopole main lobe is quite
broad and peaks at 1.45 dBi at the angle 7(1) so that it is orientated close to
the horizon with the handset in the normal position of use. The main lobe of
the patch antenna patterns is directed to 13(1) which is somewhat below the
horizon but this may not be significant in a multipath environment. There is
also a secondary peak at about 2900 due to diffraction over the top edge of the
handset and around the head. There is remarkable similarity between the co-
polar elevation patterns of the three antenna types when the . hand is
introduced. The presence of the hand tends to reduce the variation of the
patterns due to absorption of the RF currents flowing on the handset.

268
J
IP'" _I-
~
"I Ij ~ ?f-~ If
..,;,;
V ~l -
"'" ~
~ l.P ~ r-... I ...

'l:. II
.J

I' t-,~ \J f- 1-
>...:
~~ If
~
·'0 II( · 10

1\
~
il
-e T I\..J ~ ~ I~ ~
~ 1'\h WII'(J
.1$ . 1$

·lll
~
~ iK\
I ~

1-. ~ D I"::f"--o. .,.


. 1j

·1O
loA" r-... ~ l1 r0o- F::: . J)
o ]() so :wo xo :m

... .
o 30 60 to 120 l'O 110 210 )10 l?D )CO )JO )60 90 120 ISO 110 210 VO )00

DC&ITU Dcpul
~_aopo lll -o- E1lltla-nc>MJlClII: 1f'W-~ -o- BJKt.· goaopok
~.I p. kiI -o- E'a' la-JM... ". td ....o- r"bH.... ' .l(b -o- Bbd•.•• ' RP-CdI
-r-~.ucbol .. Ie' ~ .lH tac bcl "d ...... 1l'...... lKkDcI ,.tch _abCII' -ltad,,,4p.r.c.b

(a) , (b)
_h<

-
'rF'
l..il: W ~ ~b.. l,., ~
....
"V
·'0
~ \ II Ir -
.....
II \
.....
·U
v r
;rl-
h ~
r1 M'I'I
f--
·lll
v,\
. 1j

·lO
o
f--
~
)0 ro 90 120 1.50 110 210 NO
\ ,II
no 300 3JO )6()

Del~U

flo"--.op:>}, -1UwU·OPft09Ok
-'- lVW-... ,1o p4 l o;b ~Bheta"'''. 10 '''cb
-+- fl>bl-.uc b4 p' lch Bbeta-atac.Lt.ip.lcb

(c)
Figure 5.7.4 Elevation <Ill =00 ) Patternsfor Monopole, SingleLayeredand Sacked
ShortedPatches on: (a) Vertical handsetin isolation; (b) Handset Adjacent to Head Rotatedto
CENELEC normalPositionof Use; (c) Including BlockModel Hand

The antenna ISul


plots, Figure 5.7.5, are interesting in providing
further insight into the interaction of the RF .current flows on the handset with
the head and hand. The monopole shows the widest -10 dB bandwidth of the
three antennas in all three situations. On the handset in isolation the
monopole has a -10 dB bandwidth of 35.5%, the single patch 10% and the
stacked patch 11.7%. The single patch is not centered on the nominal
frequency of 1.8 GHz due to the limitations of the 2.5 mm grid. A more
refined grid would allow the patch length to be slightly increased to improve
the match but the present model is sufficient to illustrate the relative trends. It
should be noted that the larger bandwidth of the stacked patch configuration
was achieved in 60% of the patch area and 75% of the volume occupied by
the single layered version. Larger bandwidths could have been achieved
using thicker material, however a compromise with respect to ergonomic
considerations would need to be made. When the head is included the two
patch antennas show an upward shift in the frequency of the VSWR minimum
while the monopole shifts downwards. The -10 dB bandwidths are now
269
32.2% for the monopole, 10% for the single patch and 10.5% for the stacked
patch. The block hand causes an overall reduction in the degree of the match
for all the antennas. The three antenna types show a downward shift in the
position of the maximum return loss relative to the situation of the handset in
isolation so that all the antennas are matched at 1800 MHz. The bandwidths
are 37.8% for the monopole, 8.2% for the single patch and 9.4% for the
stacked patch.
0.---- -.-- - . - - -..-- - ,
.1.f-~~~..,....-__1_--_t_--_i -s .f---=.....,+=''''-- -+-- --+--- -j

~ . IS f--=...~-~_t_---:~+--___j

!a. -20 t - - --I-- - ="ki ---+-- - --j !a. ·20 t -- - -I--"<-- --+-&d=---t-:;r-- ---j
-25 t -- - - I -- --t------''''r- - --j ·25 . f - -- 4-- .:s,-__1_- +-t-- - -j
. )1) t----I----t----+---"O;~'-i . )1) f---_t_- -1r/--f-- -t-- --j
.lS ~~~+____~~.........
~~........~~~

... 0 1.70 1.10 1.90 2.00 1.70 1.80 1.90 2.00


FrtfjllfOCY (GIIz) Frtfjllfncy (GIIz)

(a) 0.-- - ...--- - -.-- - .-- - -, (b)

·'0 .f-----"-,.--1-~.......-+___pl~..+"'--_j

~ ·'1 t---'f""=~~+----+----;

ia. ·20 t - - --I-- - 4-",,-O;;.-__1_-- -;


.2S t - -- -I-- - + -- --+""'"-"""--;
. )1) t -- - -I-- - 4 -- - __1_- - -i
-lS +-----~+____~~.........~~........~~~
... 0 ' .70 1.10 1.90 2.00
FrtfjllfOCY (GIIz)

(c)

Figure 5.7.5 Isul plotsfor Monopole. Single Layered and Sacked ShortedPatches on:
(a) Vertical handset in isolation; (b) Handset Adjacent to HeadRotated to CENELEC normal
Position of Use; (c) Including BlockModel Hand

Table 5.7.3 and Figure 5.7.6 compare the SAR distributions for each
of the antenna types. A previous study which modeled a ')../4 monopole at
1900 MHz close to an upright or 300 forward tilted head found peak 1 g
SARs of 8.88 and 8.64 Wlkg respectively for 1 W of time averaged radiated
power. The lower SARs in our study are explained by the finite return loss of
270
the antennas and the rotation of the head placing the antenna further from the
surface of the head. The peak spatial SARs for the shorted patch antennas are
about one third of those for the monopole. This is better than the reduction
seen at 900 MHz with back mounted planar inverted-F antennas and can be
attributed to the better directivity achievable at higher frequencies. For each
of the antennas the inclusion of the hand in the present position reduces the
efficiency by about 10%. However if the hand is moved further up the
handset the efficiencies of the shorted patch antennas is expected to degrade
faster than that of the monopole due to current flows and possible physical
obstruction of the antenna. The SAR contours of Figure 5.7.6 show the
monopole to be exposing a wide area of the side of the head whereas the
shorted patch antennas show distinct volumes of absorption in the area of the
ear and the cheek. The shorted patch antennas produce higher SARs in the
hand than does the monopole due to the greater current flows on the handset
case.

No Hand With Hand


1 gSAR n 19SAR 11
Monopole 4.80Wlkg 60.3 % 3.92 Wlkg 50.5 %
Slnale patch 1.50Wlkg 79.0% I.00Wlkg 69.0%
Stacked patch 1.53 Wlkg 79.6% 0.92Wlkg 69.3 %

Table 5.7.3 Comparison of Maximum I g SAR Values in Head, Normalized to 1 Wof


Input Power, and Efficiency for Each of Antennas with and without Block Model Hand

(a) (b)

271
(c) (d)

(e) (0

Figure 5.7.6 Unaveraged SARContours Normalized to 1.6W/kg for 1 W of Input


Powerfor EachAntenna Adjacent to Realistic HeadModel Rotated to the CENELEC Intended
Position of Use withandwithout BlockModel Hand

5.8 Chapter Summary

In this chapter size reduction techniques of microstrip patch antennas


were proposed and investigated. A thorough investigation of shorted
microstrip patch antennas was presented and the design procedures outlined
and the characteristics summarized. These printed antennas are typically .half
the size of a conventional quarter-wave microstrip patch antenna. A further
size reduction procedure utilizing strategically positioned notches in close

272
proximity to the shorting post was then presented and verified. The resulting
antenna was 75% smaller in surface area than a conventional microstrip patch.

A winged shaped printed antenna was presented in Section 5.4. A


brief study of the parameters affecting the resonance frequency of the antenna
was given. The wing antenna is smaller by a factor of 26 when compared to a
shorted edge-fed antenna in terms of conductor area. The 10 dB return loss
bandwidth of the winged antenna was 7.4%. An integrated version of the
winged antenna, incorporated on a mobile communication handset was
examined and its performance was presented and discussed.

In Section 5.5 electrically small spiral printed antennas were


presented. The designs of the antennas were based on the philosophy of
maximizing the current path for a given surface area, yielding spiral-like
conductor patterns. Size reductions greater than a factor of 110 were achieved
using this technique.

Section 5.6 was dedicated to enhancing the performance of the


shorted patch. Firstly improving the impedance bandwidth of the shorted
patch was investigated. Bandwidths greater than 30 % can be achieved using
a stacked configuration. A bandwidth enhancement technique for very small
shorted patches was presented using two interleaved shorted spiral printed
antennas: planar coupling. The measured 10 dB return loss bandwidth of the
dual spiral printed antennas was 9.2%. The dual spiral printed antennas is 28
% smaller in physical length compared to a shorted circular patch.

This chapter presented a balanced feed technique to reduce the level


of cross polarization of shorted printed antennas. Shorted annular rings
printed antennas were concentrically loaded with each ring have a
localized/truncated dielectric layer below it to give control of its resonant
frequency was one example given. The 10 dB return loss bandwidth of this
antenna was 8.9%. The measured cross polarization levels were greater than
15 dB below the co-polarized fields. Importantly too, the antenna is compact
in structure.

Synchronous subarrays of small printed antennas have been presented


in this chapter. These antennas consist of linearly polarized small printed
antennas to produce circular polarization. Several subarrays were presented,
including a 4-shorted rectangular patches sequentially rotated and fed with a
90 degrees phase differential. The 10 dB return loss bandwidth of the 4
element shorted rectangular patches was 8.5%. The 3 dB axial ratio
bandwidth is 11.3%. The overall dimension of the antenna was 44 mm x 44
mm. A synchronous subarray of shorted equilateral triangle patches was
273
utilized and presented. The antenna consisted of 3 elements and sequentially
rotated at an angle of 120 degrees. The 10 dB return loss of the 3 element
shorted equilateral triangle patches was 31%. The 3 dB axial ratio bandwidth
of the antenna was 27.5%. The overall dimension of the antenna was 38 rom x
38 rom. An antenna utilizing two-shorted annular ring printed antennas was
presented. The antennas were concentrically loaded with each ring having a
localized dielectric material. The 10 dB return loss bandwidth was 13.3%.
The 3 dB axial ratio bandwidth of the dual concentric ring printed antennas
was 16.2%. The dual concentric ring printed antenna is compact in structure
and smaller in size than the 4- element and the 3- element synchronous
subarray. The overall dimension of the antenna was 28 rom x 28 rom.

A shorted microsrip patch with a PBG ground plane was designed,


tested and the results compared to a shorted patch of identical dimensions on a
conventional ground plane. The use of a PBG significantly improves the gain
and reduces the cross polarization levels for a shorted patch antenna.

Finally the performances of variations of the shorted patch were


investigated on a handset terminal in the presence of the user. These antennas
offer better efficiencies than monopole antennas at 1800 MHz through
directing more of the phone's radio transmissions away from the users head
so that peak spatial 1 g SARs are reduced by approximately 70%.

Bibliography
[1] J. E. Padgett, C. G. Guntherand T. Hattori, "Overview of wireless personal
communications," IEEECommunications Magazine, pp. 28-41,Jan. 1995.
[2] T. S. Rappaport, "Future trendsof mobileand personal communications," (Invited
paper) SBMOIIEEE MIT-S IMOC'95 Proceedings, pp. 387-395.
[3] R. Prasad,"Overview of wireless personal communications: Microwave
Perspective," IEEE Communication Magazine, pp. 104-108, April 1997.
[4] J. L. Finol and lG Mielke,"Past and futuredirections in cellar telephony," (Invited
paper) IEEEAP-S, pp. 7-13,1998.
[5] T. Ojanpera and R. Prasad, "An overview of third-generation wireless personal
communications: A European Perspective," IEEEPersonal Communications, pp. 59-
65, Dec. 1998.
[6] D. M. Pozar, "An overview of wireless systems and antennas," IEEEProc. AP-S, pp.
566-569,2000.
[7] K. Hirasawa and M. Haneishi, Analysis. design. and measurement of small and low-
profileantennas, ArtechHouse, Inc., 1992.
[8] K. Fujimotoand l R. James,MobileAntennaSystems Handbook. Norwood, MA:
ArtechHouse, 1994.
[9] C. A. Balanis, AntennaTheol)': Analysis and Design, secondedition,John Wiley &
Sons Inc., 1997.
[10] R. C. Johnsonand H. Jasik, AntennaEngineering Handbook, 2nd Edition,McGraw-
HilI,1984.

274
[11] Z. D. Liu, P. S. Hall and D. Wake, "Dual-frequency planar inverted-F antenna," IEEE
Trans. Antennas & Propagat., vol. 45, pp. 1451-1458, Oct. 1997.
[12] M. A. Jensen and Y. Rahmat-Samii, "Performance analysis of antennas for hand-held
transceivers using FOID," IEEETrans. Antennas & Propagat., vol. AP-42, (8), pp.
1106-1113, Aug. 1994.
[13] P. Salonen, M. Keskilarnmi and M. Kivikoski, "Single-feed dual-band planar
inverted-F antenna using V-shaped slot," IEEETrans. Antennas & Propagat. , vol.
AP-48, (8), pp. 1262-1264, Aug. 2000.
[14] C. R. Rowell and R. D. Murch, "A capacitively loaded PIFA for compact mobile
telephone handset," IEEETrans. Antennas & Propagat., vol. 45, pp. 837-843,1997.
[15] R. E. Munson, "Conformal microstrip antennas and microstrip phased arrays," IEEE
Trans. Antennas & Propagat., vol. AP-42, (1), pp, 74-78, Jan. 1974.
[16] 1. W. Howell, "Microstrip antennas," IEEE Trans. Antennas & Propagat., vol. AP-
23, (1), pp. 90-93, Jan. 1975.
[17] K. R. Carver and 1. W. Mink, "Microstrip antenna technology," IEEE Trans.
Antennas & Propagat. , vol. AP-29, (I), pp. 2-24, Jan. 1981.
[18] T. K. Lo, Y. Hwang, E. K. W. Lam and B. Lee, "Miniature aperture-coupled
microstrip antenna of very high permittivity," Electron. Len; vol 33, pp, 9-10, Jan.
1997.
[19] G. A. Kyriacou and J. N. Sahalos, "Analysis of a probe-fed short-circuited microstrip
antenna," IEEETrans. Vehicular Tech., vol. 45, (3), pp. 427-430, Aug. 1996.
[20] A. Boag, Y. Shimony and R. Mittra, "Dual band cavity-backed quarter-wave patch
antenna," IEEEAP-S Digest, vol. 4, pp. 2124-2127, 1995.
[21] R. B. Waterhouse, "Small microstrip patch antenna," Electron. Lett., vol. 31, pp. 604-
605, April 1995.
[22] K. -L. Wong and Y. -F. Lin, "Small broadband rectangular microstrip antenna with
chip-resistor loading," Electron. Lett., pp. 1593-1594, Sept. 1997.
[23] 1. -H. Lu, "Single-feed dual-frequency rectangular microstrip antennas with pair of
step-slots," Electron. Lett ., vol. 35, (5), pp. 354-355, Mar. 1999.
[24] D. M. Pozar, Microwave Engineering - Second Edition, John Wiley and Sons Inc.,
New York, 1998.
[25] 1.T. Aberle, D.M. Pozar and C.R. Birtcher, "Evaluation of input impedance and radar
cross section of probe fed microstrip patch elements using an accurate feed model",
IEEETrans. Antennas Propagat., AP-39, pp. 1691-1697, December 1991.
[26] M. Sanad, "Effect of the shorting posts on short circuit microstrip antennas", Proc.
IEEEAP-Symp., pp. 794-797, June 1994.
[27] I. Park and R. Mittra, "Aperture-coupled small microstrip antenna", Electron. Lett.,
Vol. 32, pp. 1741-1742, Sept. 1996.
[28] R. B. Waterhouse and S. D. Targonski, "Performance of microstrip patches
incorporating a single shorting post", Proc. IEEEAP-Symp., pp. 29-32, July 1996.
[29] 1. Huang, Personal Communication.
[30] D. H. Schaubert, "A review of some microstrip antenna characteristics", Microstrip
Antenna Design, IEEE Press, pp.59-67, 1995.
[31] Ensemble 5.1, Ansoft, 1998.
[32] R. B. Waterhouse, S. D. Targonski and D. M. Kokotoff, "Design and performance of
small printed Antennas", IEEE Trans. Antennas & Prop., vol. AP-46, pp. 1629-1633,
Nov.1998.
[33] J. T Rowley and R. B. Waterhouse, "Performance of shorted microstrip patch
antennas for mobile communications handset at 1800 MHz," IEEE Trans. Antennas
& Prop. Vol. AP-47, pp. 815-822, May. 1999.
[34] M. A. Jensen and Y. Rahmat-Samii, "EM interaction of handset antennas and a
human in personal communications", Proc. IEEE, vol. 83, pp. 7 -17, Jan. 1995.

275
[35] D'Inzeo, "Proposal for numerical canonical models in mobile communications" in
Biomedical Effects of Electromagnetic Fields - Reference Models in Mobile
Communications, D. Simunic, Ed. Rome, Italy: COST244, pp. 1-7, 1994.
[36] R. B. Waterhouse, "Small printed antenna easily integrated into a mobile handset
terminal," Electron. Lett., Vol. 34, pp. 1629-1631, Aug. 1998.
[37] F. Ali and J. B. Horton, "Introduction to special issue on emerging commercial and
consumer circuits, systems, and their applications," IEEE Trans. Microwave Theory
Tech., Vol. 43, pp. 1633-1638, July 1995.
[38] R. B. Waterhouse, "Small printed antenna easily integrated into a mobile handset
terminal," Electronics Letters, vol. 34, pp. 1629 -1631, Aug. 1998.
[39] R. B. Waterhouse, 1. T. Rowley and K. H. Joyner, "A stacked shorted patch,"
Electronics Letters, vol. 34, pp, 612 - 614, April 1998.
[40] 1. Rashed, and C.-T. Tai, "A new class of resonant antennas", IEEE Trans. Antennas
& Propagat, vol. 39, pp. 1428 - 1430, Sept. 1991.
[41] P. S. Hall, Private Communication.
[42] XFDTD (1997), User's manual for XFDTD the X-Window Finite Difference Time
Domain Graphical User Interface for Electromagnetic Calculations. Version 4.03,
Remcom Inc., June 1997.
[43] R. B. Waterhouse, "Printed antenna suitable for mobile communication handsets",
Electron. Lett.• Vol. 33, pp. 1831-1832, Oct. 1997.
[44] Zealand Software Inc., IE3D™ Version 6.01.
[45] D. M. Kokotoff, 1. T. Aberle and R.B. Waterhouse, "Rigorous analysis of probe-fed
printed annular rings," IEEE Trans. Antennas & Prop., vol. AP-47, Feb. 1999.
[46] J. Huang, "A technique for an array to generate circular polarization with linearly
polarized elements," IEEE Trans. Antennas & Propagat., vol. 34, pp. 1113-1124,
Sept. 1986.
[47] C. -Yo Huang, 1. -YoWu and K. -L. Wong, "Broadband circularly polarized square
microstrip antenna using chip-resistor loading," lEE Proc. Micro. Antennas &
Propagat., vol. 146, pp. 94-96, Feb. 1999.
[48] M. A. Jensen and Y. Rahmat-Sarnii, "Performance of circularly polarized patch
antennas for personal satellite communications including biological effects," Proc.
AP-S, Newport Beach, Califoma , USA, pp. 1112-1115, June 1995.
[49] P. S. Hall, 1. S. Dahele and J. R. James, "Design principles of sequentially fed, wide
bandwidth, circularly polarized microstrip antennas," lEE Proc. H, vol. 136, pp. 381-
389, Oct. 1989.
[50] D. F. Sievenpiper, L. Zhang, R. F. Jimenez Broas, N. G. Alex6polous, E.
Yablonovitch , "High-Impedance Electromagnetic Surfaces with a Forbidden
Frequency Band," IEEE Trans. Microwave Theory and Tech., vol. 47, pp. 2059-
2074, Nov. 1999.
[51] R. Coccoli, F. Yang, K. Ma, T. Itoh, "Aperture-Coupled Patch Antenna on UC-PBG
Substrate," IEEE Trans. Microwave Theory Tech., vol. 47, pp, 2123-2130, Nov.
1999.
[52] C. Gabriel, "Compilation of the dielectric properties of body tissues at RF and
microwave frequencies", Brooks Air Force Base, report no. AUOE-TR-1996-0037,
1996.

276
Chapter 6 Direct Integration of Microstrip
Antennas

6.1 Overview of Requirements for Integration

As the popularity of wireless communications continues to grow and


the demand for more sophisticated applications and therefore required
bandwidth increases, it is imperative that technical solutions are found to
ensure these needs can be met. There have been a variety of methods that
have helped overcome some of the issues related to the transmission of higher
bit rates/bandwidth such as using higher order modulation techniques and
smart base stations, however, most of the proposals are relatively short-term
in nature. Fundamental to any advancement in wireless services is to increase
the frequency of operation of the communication link thereby allowing more
useable bandwidth and information to be transferred. For this reason there
have been many investigations into wireless systems at higher transmission
frequencies, well into the millimeter-wave ranges (26 - 110 GHz). Such
systems should overcome the inherent limitations and issues associated with
the lower microwave frequency spectrum.

Of course there are issues associated with utilizing higher frequencies


for wireless communications and these include higher atmospheric attenuation
and therefore the requirement for substantially more base stations for an
equivalent coverage area. Having more base stations also requires a
sophisticated distribution network and this is where optical fiber networks can
help resolve some of these issues. The transmission properties of fiber-optic
systems, as well as the capacity and latency abilities of these networks makes
them extremely useful for unconventional applications such as wireless
communications.

One critical issue associated with the deployment of high radio


frequency 'fiber-wireless' communication systems is the cost associated with
the base station module. To make such systems viable the cost of these base
stations must be minimized. At frequencies beyond 20 GHz, an integrated
approach of all the photonic, microwave and radiation components must be
incorporated to maximize the efficiency of the link. Of course there are
several means to implement this integration, each with their advantages and
disadvantages and the objective of this chapter is to review these techniques
with respect to printed antenna technology. In particular we will focus on
direct integration: where some of the functions of the module are integrated
within the same material. This can be extremely difficult for microstrip patch

R. B. Waterhouse, Microstrip Patch Antennas: A Designer’s Guide


© Springer Science+Business Media New York 2003
technology, as single layered patches are. notoriously inefficient when high
dielectric constant material is used.

6.1-1 Indirect Integration

The indirect integration approach is really a sub-set of the Hybrid


Integration technique [1]. Here each photonic/RF/antenna function is
designed and developed preferably using planar technology on an optimized
combination of materials. These 'islands' of optimized functions are then
interconnected via small sections 50 Q transmission lines and a series of bond
wires. Some of the advantages of the indirect integration technique are
relatively obvious but it is important to spell them out. Importantly the
indirect integration approach allows for each functional block of the module
to be developed relatively independently of the other components of the base
station. For example the filters/diplexers can be designed independently of
the antennas or photo-diodes as long as each component has input (and/or
output) ports matched to a given impedance, typically 50 Q. Such a
procedure allows for well-established practices to be undertaken for the
design and development of the individual components. Figure 6.1.1 shows a
photograph of an example of a broadband integrated photonic-antenna
interface developed using the indirect integration approach [2].

palch
ar1ema

Package dirrcnslons - 4.5 em x 2.0 em x '.0 em

Figure 6.1.1 Example of Indirect Integration Approach

As pointed out before, each element was designed individually, for


example the antenna was designed using a full-wave spectral domain analysis,
whereas the diplexer was designed using transmission line theory. Such an
approach allows for efficient design practices to be incorporated for the
individual sub-eomponents of the base station. Also it allows for out-sourcing
of components: in this case the photodiode. There are however some
278
drawbacks of this technique including the larger size of the terminal
(compared to direct integration), the labor-intensiveness of the development
process (due to the more elaborate integration and packaging required for in-
direct integration) and the loss associated with the interconnections, namely
the bond-wires and the small sections of transmission lines. This loss is not
trivial at higher frequencies (beyond 20 GHz) and can cause degradation in
the system performance as well on a sub-eomponent level. One such loss is
associated with the excitation of leaky waves that are unguided modes within
the substrate of the material. Typically when the sub-components of a base
station are designed, the mounting structure in which they reside is not taken
into consideration (for example, the diplexer in Figure 6.1.1 sits within a
recessed cavity). These structures can strongly couple to a typically weakly
excited leaky wave thereby causing spurious radiation and reduced efficiency.

6.1-2 Direct Integration

The second procedure is once again a sub-elass of hybrid integration.


Here all the functions utilize a common material and so the issues associated
with the development of small, independent transmission line sections
discussed earlier are avoided. However, unlike in the concept of OEIC,
required aspects of each sub-eomponent are not entirely accomplished by
designs using that common material. For example, consider an integrated
antenna-photonic module incorporating Lithium Niobate (LiNb03) as the
material for the FlO converter. To use this material in an OEle approach
would yield a very inefficient antenna, simply because of the high dielectric
constant of LiNb03• However, by using a two element radiating structure,
such as in [3], where one radiator resides on the high dielectric constant
material and the other on a low dielectric constant laminate, highly efficient
and broadband antennas can be formed. Thus only part of the functionality of
each sub-eomponent is provided by the common material. This approach also
allows for the utilization of 'flip-chip' techniques [4] for the development of
any RF amplifiers required for the up and/or down-links of the base-station
module on materials unsuited for this function, once again providing a more
flexible efficient design. Figure 6.1.2 shows a photograph of two antenna-
photonic modules developed on LiNb03 where other laminates are integrated
into the design to improve the overall efficiency of the link.

Although direct integration overcomes some of the issues associated


with indirect integration there are still some problems with this technology.
Firstly, typically large wafers are required to accommodate for the foot-print
of the entire base-station module. This includes all amplifiers, filters,
photonic components and antennas. It is very difficult to make wafers of any
opto-electronic material of the order of 4 x 4 inches and so some form of
279
indirect integration is still necessary. Secondly, the design procedures for
each sub-component now become difficult as the designer has lost one of their
degrees of freedom, namely the choice of material (dielectric constant and
thickness). For example, developing RF passive components and antennas at
millimeter-wave frequencies on very thin laminates with dielectric constants
greater than 10 is very difficult to realize. Despite these drawbacks and with
the development of sophisticated micro-machining procedures (for example
[5]), direct integration could provide the most cost effective procedure for
integrated antenna-photonic modules. As mentioned previously, this chapter
explores how an antenna designer can develop efficient printed antennas that
can be integrated with microwave and photonic devices.

Figure 6.1.2 Example of Direct Integration Approach

One motivation for looking at antennas suitable for integration with


OEICs and MMICs for the development of compact base stations in optically
distributed radio communication networks, as mentioned before [6]. Other
applications include fibre based antenna remoting, phased arrays, and quasi-
optical power combining systems [7]. Printed antennas have compatible
structural properties with these integrated module approaches; namely, they
can be implemented on the same material used to fabricate the microwave and
photonic devices. However in their conventional form, printed antennas
fabricated on high permittivity OEICIMMIC materials suffer from a very
small impedance bandwidth (as was shown in Chapter 2). Also, surface wave
effects can severely degrade the efficiency of the printed antenna, diminishing
the performance of the overall fibre-radio network. The excitation of surface
waves not only reduces the efficiency of 'the radiator, but can also cause
unwanted coupling between the devices within the module.

280
6.1-3 Single Layer Printed Antennas on High Dielectric Constant
Material

Figure 6.1.3 represents a traditional printed antenna located on a high


dielectric constant substrate. This basic structure was simulated and
optimized on HFSS [8] to be a benchmark for comparison to other
configurations introduced in this chapter. The high dielectric substrate (e, =
10.4), which simulates the high permittivity OEICIMMIC materials, had a
height of 1.925 mm, The patch element dimensions were 9.4 x 20 mm. The
return loss of the benchmark antenna is represented in Figure 6.1.4. A 10 dB
return loss bandwidth of approximately 2.3% was observed. Also note how
frequencies away from resonance do not rapidly approach 0 dB, especially at
frequencies higher than the resonant frequency.

Edge feeding

Patch conductor

~ High dielectric constant


grounded substrate

Figure 6.1.3 Conventional EdgeFed Microstrip PatchAntenna on HighPermittivity


Substrate
o
--.... ~ r:
..... -5
m
~ ·10
\, I
:I
o -15
..J

-
E
~
-20

CD
a: ·25

-30

-35
3.5 4 4.5 5
Frequency (GHz)

Figure 6.1.4 Return loss of Benchmark Antenna

281
Figure 6.1.5 show the E- and H-plane radiation patterns for the
benchmark antenna. A wide beamwidth in both planes is evident, which is a
typical trait of printed microstrip patch antennas. The backfire radiation level
is low, exhibiting a front to back ratio of 24 dB. Figure 6.1.6 plots the
magnitude of the E-fIeld for a two dimensional E-plane cut through the
benchmark antenna. A propagating wave is evident on the microstrip line,
showing field concentrations in the high dielectric constant substrate between
the strip and ground plane. Also note the high intensity field in the high
permittivity substrate under the patch conductor, implying that a large portion
of the field is trapped within the substrate.
o

90

- H-plane
E-plane

180

Figure 6.1.5 Co-polar Radiation Patterns of Benchmark Antenna

Patch element Microstrip feed line


Max

_".,"C--'"
1 High


pennittivity
T substrate
Min

Ground plane

Figure 6.1.6 E-fieldintensity plot of Benchmark Antenna (E-planecut)

282
6.2 Slot-coupled Procedures and Solutions

6.2-1 Aperture Coupled Antennas

As previously stated, aperture coupling energy from a feed line to a


printed antenna element has several advantages over conventional direct
contact techniques, such as microstrip edge feeding or probe feeding. The
geometry of a conventional printed antenna element that employs aperture
coupling is given in Figure 6.2.1. As was shown in Chapter 2, the
configuration uses two dielectric substrates separated by a common ground
plane.
. - Patch substrate
......-~-+--- Patch clement

r='O"'~ .-- Feed substrate

~-Z~~~~;~~~~~-COUPl i ng aperture

Figure 6.2.1 Aperture Coupled Patch Antenna

In this section, aperture coupling is investigated as a method to


alleviate the drawbacks of using high dielectric constant feed materials in
printed antennas. Section 6.2-2 covers the microstrip fed aperture stacked
patch configuration, and the effect of decreasing the thickness and increasing
the permittivity of the feed substrate on the behavior of the antenna. The
Coplanar Waveguide (CPW) fed aperture stacked patch antenna is then
introduced in Section 6.2-3 as a wideband element that demonstrates the
desirable characteristics for OEICIMMIC integration. Techniques to further
improve the suitability for integration of the CPW fed ASP by reducing the
level of back radiation are detailed in Section 6.2-4.

6.2-2 Microstrip Aperture Stacked Patch on high dielectric material

As mentioned in Chapter 3, exceptionally wide impedance


bandwidths of 50 to 70% can be obtained from the aperture stacked patch
configuration [9]. However, this is not an optimal structure for integration
into OEICIMMIC modules. The primary hindrance the feed material used
was of low permittivity (e, = 2.2), not being compatible with the high
dielectric constant OEICIMMIC wafer materials. To attain the superior
bandwidth attributes of the aperture stacked patch configuration, the structure
needs to be altered to incorporate a high permittivity feed material for
283
integration. The process of performing this alteration, and the consequences
on the behavior of the aperture stacked patch antenna are detailed in this
section.

Figure 6.2.2 for convenience, displays the standard aperture stacked


patch configuration. A microstrip reactive power divider feeds the resonant
coupling aperture, which is mutually coupled to the two parasitic patch
elements. The effect of gradually increasing the permittivity while
maintaining the thickness of the feed substrate was investigated using
Ensemble [10]. At each increment of permittivity the structure was optimized
for maximum bandwidth, but the basic geometry of the aperture stacked patch
antenna shown in Figure 6.2.2 was maintained. The microstrip line widths
were also adjusted at each stage to ensure the efficient operation of the
reactive power divider.

_ _~_ _ Pan..itic patch


element 2

Parasi tic patch


clemen t I

Feed substra te

.!.KL-~-- Coupling aper ture

1-- Microsirip feed line

Figure 6.2.2 Aperture Stacked Patch Configuration

As the permittivity of the feed substrate is increased, there is a decline


in optimal bandwidth obtainable from the aperture stacked patch
configuration. This is shown in Figure 6.2.3. The reduction in bandwidth is
primarily due to an increase in the quality factor (Q) of the resonant coupling
aperture. To illustrate this, the apertures and microstrip feed network from
the aperture stacked patch antennas with feed substrate permittivity of 2.2 and
10.2 were analyzed. The input impedances of the two single substrate
aperture antennas are compared in Figure 6.2.4. Distinctly narrower resonant
peaks are observed for the aperture antenna with the high dielectric constant
substrate, compared to those of the lower permittivity material. This increase
in Q equates to a reduction in the bandwidth of the aperture element. Hence,

284
when the aperture is mutually coupled to parasitic patch elements, the overall
aperture stacked patch antenna bandwidth is diminished.
70
:.........
65
.........
r-,
---
60
::::l!
0 ~ '\..
55
J:
'C
<,
<,
50
j
'C

<,
C 45
lU
lD
40

35
-,
~
30
2 3 4 5 6 7 8 9 10 11

Relative Permittivity

Figure 6.2.3 Optimized percentage bandwidthversus permittivity of feed substrate


600 .---,.--.,..--.,....--..,..--...,.---.----r--...,

-
CII
g 400 \---

a 500 \---f--+--+---f-+---+-i'--f---+-:-
:,
f --t-::,---+-++----Hr--t---Hft---l
;;
--l

'.
I'll
'0
8. 300 \--- f -- H I -+---:r+--+'tt--t---tf-t---l
.5
,,
'5 200 f - - I ' t - -+++-+--i++---f..I.& --- t - ---lf--l--i
Q.
c

,,
.1

5 10

Figure 6.2.4 Input Impedance Comparison for High and Low Feed Substrate
Permittivity ApertureAntennas

The peak gain within the impedance bandwidth of the antenna is also
diminished with increasing permittivity, although it still remains at a typical
level for microstrip patch antennas. This is shown in Figure 6.2.5. The
minimum gain remains relatively constant. It should also be noted that the
width of the 100 .Q sections of microstrip line in the reactive power divider
became extremely thin (80J.lIIl) when the permittivity of the feed material
reached 10.2. The level of back radiation remained constant.

285
Due to the very thin microstrip lines required to realize the reactive
power divider of the antenna structure of Figure 6.2.2 when a high
permittivity feed substrate is employed, a single feed line aperture stacked
patch antenna was developed. A schematic of the single feed line aperture
stacked patch antenna with a feed substrate dielectric constant of 10.2 is given
in Figure 6.2.6. For OEICIMMIC integration purposes, the thickness of the
high permittivity feed substrate (hi) is excessive. A theoretical study of the
consequences of decreasing hi was undertaken using [10]. Once again, the
structure was optimized for maximum operational bandwidth after each
iteration of hi.

1.5
:....
...... 1'\..
. . . Maximum Cain
1-. - MInimum Cain
r-
I
........
~
~
iii 7.5
........... Ie
:!:!.
c 7
'iii

-- .. .. .- -- -. . . .. .. .. ..
C) 1.5

I
:-- -
~
5.5

5
2 4 5 I 7 I a 10 11
Relative Permittivity

Figure 6.2.5 Maximum and Minimum Antenna Gain versusPermittivity of Feed


Substrate

- - - Patch substrate (h..)

_-,.e...,t''----- - Parasitic patch


element 2 (I,. w,)

Parasitic patch
element I (I" wr)
_ High E, feed substrate (h i)

~~~t; ",,::::.,:r.L-- - Coupling aperture (I.. wJ

L....- Microstrip feed line (wm, loel

Figure 6.2.6 SingleFeed ASP Antenna

Figure 6.2.7 shows that a decrease in hi also leads to a reduction in


the achievable bandwidth of the aperture stacked patch structure. This is due
to the aperture element having a lower bandwidth with smaller values of hI.
In general, the bandwidth of a printed antenna element is proportional to the
volume it occupies, explaining the observed trend. As before, when the

286
aperture is mutually coupled to parasitic patch elements, the overall aperture
stacked patch antenna bandwidth is diminished with decreasing hi.
42

40 »:
38 /
~
..
~
.t: 38
V
·"i 34
./
"III
C 32
./
ca
30
»"
28
... /
26
0.2 0.4 0.6 0.8 1.2 1.4 1.6

Thickness (mm)

Figure 6.2.7 Optimized Percentage Bandwidth versus Thickness of Feed Substrate

As can be seen from Figure 6.2.8, the peak gain within the impedance
bandwidth of the antenna is relatively consistent as hi is varied, with slight
fluctuations around 8 dBi. The minimum gain displayed some minor
variations, but maintained a level above 6 dBi. The front to back ratio
exhibited improvement for lower values of hi.
8.5
I,'"
....- -
Mlllmum G.ln

----... ~ ,.
- • • Minimum G.ln . /

~ 7.5

.... ..... ... ,


C

~ 7

.... ..... ----


8.5
,,
,
8
0.2 0.4 0.6 0.8 1.2 1.4 1.8

Thickness (mm)

Figure 6.2.8 Maximum and Minimum Gain versus Thickness of Feed


Substrate

A generous impedance bandwidth is obtainable from an aperture


stacked patch antenna that incorporates a thin, high permittivity feed
substrate. However, the bandwidth is not of the same magnitude as the
aperture stacked patch antennas in Chapter 3 due to the effect of the feed
substrate on the resonant coupling aperture.

287
6.2-3 CPW fed Aperture Stacked Patch on high dielectric material

Coplanar waveguide (CPW) transmission structures are commonly


used in the design of microwave and optical integrated circuits, as the
connection of series and shunt components can be accomplished without the
need for vias or holes through the substrate. CPW also offers other salient
properties at higher frequencies (especially millimeter-wave frequencies)
when compared to conventional microstrip line printed circuits. To illustrate
this, approximate quasi-static expressions for microstrip line [1] and CPW
[11] are analyzed below.

60 In(8d +~)
F: W 4d
for W
d
~1

Zom = 120n

Ie{; + 1.393+ O.667In(; + 1.444)] for W


d
~1

ohms (6.1)
e, is the effective relative permittivity
d is the thickness of the dielectric substrate
w is the microstrip line width

ac=--
Rs
Np/m where R. =~CDIl'
ZomW 20-
(6.2)
R, is the surface resistivity of the conductor

Z = 30n K'(k)
where k =
ohms S
ocpw ~£'2+1 K(k) S+2W

(6.3)
S is the CPW center conductor width
W is the CPW gap width
K(k) is the complete elliptic integral of the first kind and
K'(k) = K(k')
kl=~(1-k2)
288
As shown in Equation 6.1, the impedance of a microstrip line (Zom> is
dependant on the thickness of the substrate. Hence, for a given substrate
thickness and permittivity, only a single value of conductor width can achieve
a desired Zorn' For OEICIMMIC integration applications where the substrates
materials are generally thin and of high dielectric constant, very fine
microstrip lines are required. Aside from elevating the complexity of
fabrication, decreasing the width of a microstrip line increases the conductor
loss. This is especially pertinent at millimeter-wave frequencies, as the
attenuation due to conductor loss is also escalated at higher frequency (refer
to Equation 6.2).

The approximate quasi-static solution for the impedance of CPW


(Equation 6.1) does not contain a dependence on the substrate thickness.
Hence, the desired characteristic impedance can be obtained by choosing
appropriate values of the CPW center conductor width (S) and CPW gap
width (W). It is possible to realize an impedance value with numerous
combinations of S and W, providing the value of k is maintained. This
flexibility enables the selection of Sand W to minimize the level of conductor
loss in the high microwave and millimeter-wave range, as well as controlling
dispersion and radiation losses [12].

CPW also has other advantages over microstrip transmission lines for
OEIC/MMIC integration. CPW circuits only require processing on a single
side of an OEICIMMIC wafer. Aside from the ease of connecting series and
shunt devices as previously mentioned, single sided processing requires a
lower number of fabrication steps than microstrip based circuits, reducing
manufacturing expenses. Having the center conductor and ground electrodes
of the CPW in the same plane provides a greater level of isolation between
adjacent transmission lines. Therefore, it may be possible to achieve smaller
circuit/chips sizes if CPW is used as the transmission medium, rather than
microstrip lines.

To exploit these benefits for the integration of antennas with


OEICIMMIC technology, a CPW line can be employed to feed an aperture
coupled patch antenna [13, 14]. It has been shown that the impedance and
radiation characteristics of a CPW fed aperture coupled patch antenna are
analogous to those of microstrip line fed aperture coupled patch with an
aperture length shortened by the CPW center conductor width [15].

A modified version of the wide-band aperture stacked patch


configuration [9] is investigated for OEICIMMIC integration applications.
The microstrip line feed is substituted for a CPW, and a thin, high permittivity
289
substrate is employed to emulate a semiconductor wafer. These alterations
ease the integration of the antenna into a CPW based OEICIMMIC module.

The geometry of the broadband CPW fed Aperture Stacked Patch


(ASP) antenna is given in Figure 6.2.9. The CPW feed line is constructed on
0.254 mm Alumina substrate (Er = 10.2) to simulate the high dielectric
constant materials used for monolithic microwave and optical integrated
circuits. A large aperture is used as a resonator within the operating band.
Thick slabs of Rogers Duroid 5880 and foam are used as dielectric material
for parasitic patch elements 1 and 2, respectively. The parasitic patches are
etched on the underside of the thinner RT Duroid5880 substrates.

+--7"r- - Parasi tic patch


clemcn l2

- - - Palch substrate

Parusitic patch
elemen t I

6:::,:,jz:..-- - Coupling aper ture


~~~~~~~WZ:....-._-- CPW reed line

Figure 6.2.9 CPW fed ASP antenna

The simulated and measured input impedance is shown in Figures


6.2.lOa and 6.2.10b respectively, over the frequency range of 5 GHz to 9
GHz. A VSWR < 2:1 bandwidth of 40% centered at 6.75 GHz was predicted
by the simulation. Very good agreement between theory and experiment is
obtained, exhibiting similar resonant behavior and bandwidth. Slight shifts in
the resonant loops can be attributed to lateral movement and air gaps between
the layers of the constructed antenna, as no provisions were made for this in
the simulations.

Figure 6.2.11 compares the simulated and measured far field radiation
patterns. Reasonable similarity in the shape of the radiation patterns is
observed, and front to back ratios of 13 dB and 11 dB were obtained for the
theoretical and experimental results, respectively. This high level of back
radiation is a consequence of the resonant aperture in the ground plane. A
resonant aperture element in isolation will radiate to an almost equal level at
broadside and backfire. Hence, a trade-off exists between the enhanced
290
bandwidth that can be achieved with a resonant aperture element in a CPW
fed ASP, and the level of back radiation. Techniques to reduce the back
radiation of CPW fed ASP antennas are given in Section 6.2.4. Both patterns
in Figure 6.2.11 exhibit a side lobe approximately 5 dB down from the main
lobe in the E-plane, which is a consequence of spurious radiation from the
CPW feed line.

(a) (b)

Figure 6.2.10 Input Impedance of CPW fed ASP: (a) Predicted; (b) Measured

The theoretical gain of the CPW fed ASP was in excess of 8 dBi for
the majority of the impedance bandwidth of the antenna,remains above 7.5
dBi. The gain remained extremely close to the directivity of the antenna,
differing by no more than 0.2 dB. This implies that the antenna is
exceptionally efficient, and its performance is not being degraded by the
excitation of surface waves.
o

-90 I-w-t--+~iE-+--h41--1 90

180
1:l0
1- II·pIane
-E-pIoDt
I tOO

(a) (b)

Figure 6.2.11 Radiation patterns of CPW fed ASP Antenna


291
Figure 6.2.12 compares the E-field magnitude for a two dimensional
E-plane cut through a conventional microstrip patch antenna and the CPW fed
ASP antenna. The two antennas use a 0.254 mm thick RT Duroid 6010.2 feed
substrate (e, = 10.2). In both cases, a propagating wave is evident on the feed
line, showing field concentrations in the high dielectric constant substrate.
However, the CPW fed ASP configuration does not show the high field
concentrations in the feed substrate that are evident under the driven patch
element of the conventional patch antenna, The strong mutual coupling of
the resonant elements in the CPW fed ASP draws the field out of the high
permittivity substrate towards the parasitic patch elements. This is believed to
be the mechanism that reduces the surface wave excitation in feed substrate,
and allows the highly efficient operation of the CPW fed ASP. For
comparison, the conventional patch antenna exhibited a bandwidth of 0.8 %
and a gain of 3 dBi, which was 3.8 dB below the directivity indicating very
poor surface wave efficiency. Hence, the CPW fed ASP structure provides a
solution to both the impedance bandwidth and surface wave excitation
concerns that arise when integrating an antenna element into an OEICIMMIC
module.
Patchelement Microstrip feed line

1 High
permittivity
T substrate
Groond pione

(a)

Patch clement Patch clement

l Uigh" .
PCmutUVl ty
T substrate

Coupling aperture CPW reed line

(b)

Figure 6.2.12 E-fieldIntensity Plot of (a) conventional microstrip patch antenna; (b)
CPW fed ASP antenna

A single parasitic patch element can also be employed in conjunction


with a CPW fed resonant aperture to obtain an antenna with the appropriate

292
properties for OEICIMMIC integration. Using a substrate structure consisting
of a high permittivity feed substrate and foam parasitic patch dielectric, low
surface wave excitation and an ample impedance bandwidth can be achieved.
This single parasitic patch element antenna has a lower profile than the CPW
fed ASP configuration, but has a reduced, although still generous bandwidth
(approximately 20 - 25%) due to the decrease in the number of resonators.

6.2-4 Back Radiation Elimination

For some applications, the level of radiation in the rear hemisphere of


the CPW fed ASP may be considered high. This is simply due to the aperture
in the ground plane being of resonant length as was pointed out for any large
slot coupled patch in Chapter 3. This rear directed radiation from the antenna
can introduce difficulties in packaging, or cause coupling to components or
other antennas mounted behind, as in the case of sectoral coverage ARUlbase
stations. There are several techniques to reduce the level of back radiation
from antennas using resonant aperture coupling. A few of the more popular
methods are explained below.

6.2-4.1 Shielding plane

The use of a shielding plane separated from the rear of the antenna by
a foam dielectric can reduce the backward radiated power. This shielding
plane configuration is shown in Figure 6.2.13, attached to a single parasitic
patch element CPW fed ASP. However, the presence of this shielding plane
and the CPW ground plane can promote the generation of parallel plate
modes, reducing the efficiency/gain of the antenna. Parallel plate mode
excitation can also lead to impedance abnormalities, and pattern degradation
due to diffraction at the edges of the parallel plates.

4 -- - Patch substrate (E, =2.2)


_ Foam (E, = 1.07)
- - r -r -- -Parasitic patch (1,_ w,)

_ Feed substrate (f, = 10.2)

' # - _- Coupling aperture (I., w . )


7F--r-r-- CPW feed line

, .- - Merallic shielding plane

Figure 6.2.13 CPW fed ASP with shielding plane


293
The measured return loss of a single parasitic patch element CPW fed
ASP with a shielding plane affixed to the rear is displayed in Figure 6.2.14.
The effect of parallel plate mode excitation is clearly identified. The return
loss dip indicated in Figure 6.2.14 was not present when the shielding plane
was removed. Although the effect of parallel plate mode excitation can be
alleviated with the use of shorting pins, it remains an inherent concern of this
configuration.
o
·5
, , ~

iii ·10
\
~ \ f"'\ I'U
I ,
,~
Gl · 15
"C
:::l I
~ ·20
l:
CI Due to parallel
C'II ·25
== plate mode
·30 excita tion
·35

-40
4 5 e 7 e
Frequency (GHz)

Figure 6.2.14 Measured Return Loss of CPW fed ASP with Shielding Plane

6.2-4.2 Cavity backing

Adding a cavity to enclose the rear hemisphere of an aperture fed


antenna can reduce back radiation without parallel plate mode excitation and
this was implemented in an example in Chapter 3. However, manufacturing
and machining issues can be critical factors in the viability of using cavities.
These issues are especially pertinent as the operating frequency increases.
The cavity becomes physically small, and the degree of difficulty in realizing
accurate dimensions escalates. Also, the cavity tends to diminish the
achievable impedance bandwidth.

Another concern unique to using a cavity for a CPW fed ASP is that
the lateral metallic bounds of the cavity can form a section of conductor
backing in the CPW feed line. The conductor backed CPW section will then
have a contrasting characteristic impedance to the ordinary CPW line,
resulting in an impedance mismatch.

294
6.2-4.3 Reflector Patch element

Microstrip patch elements can also be used to decrease back radiation


from CPW fed ASP antennas. A 'reflector' patch element (similar to those
used in Chapter 3) for microstrip line fed ASP antennas) mounted at the rear
of the antenna on a foam dielectric can produce a field that cancels the rear
directed radiation of the resonant aperture. This method does not have the
detrimental effects of parallel plate modes, or any major manufacturing
difficulties.

In this section, the use of patch elements at the rear of a CPW fed
ASP on a finite size ground plane to reduce the level of back radiation is
experimentally analyzed. A comparison between a standard CPW fed ASP
with and without a reflector patch is presented. The effect of varying some of
the parameters of the reflector patch to tailor the rear hemisphere radiation
pattern is also considered.

A schematic diagram of the single parasitic patch element CPW fed


ASP being examined is given in Figure 6.2.15. It consists of a CPW fed
resonant aperture formed on a 0.254 mm thick alumina substrate. Alumina
was chosen to simulate the high permittivity materials commonly used for
MMIC and optical circuit fabrication. A 3 mm thick foam dielectric separates
the aperture and the parasitic patch conductor. Also shown in Figure 6.2.15 is
the reflector element that uses the same ground plane as the CPW feed line,
and is mounted at the rear of the antenna on a 5 mm thick foam substrate.
Both the parasitic patch and the reflector elements are etched on 0.254 mm
Duroid 5880 laminates, as it is not possible to deposit conducting material
directly on the foam material. The ground plane and substrate size was 60 x
60mm.
_ Patch substrate ( E, = 2.2)

=7'1-- - Parasitic patch (I,. "'p)

_ Feed substrate (r, =10.2)


-~= ~~~~-- COuPling aperture (I•• "'.)
~~~~~6r-- Cpw feed line
_ - - Foam (E, = 1.07)
_ Reflector substra te (E, = 2.2)

- - - Reflector (I, . "',)

Figure 6.2.15 CPW fed ASP with reflectorpatchelement (lp = wp = 16 rom, I. = 11 rom,
w.= 0.2 mm.L« 27.3 rom, wr= 0.2 rom)

295
As in Chapter 3, the reflector element is designed to operate well
above it's first resonance, producing a radiated field approximately 1800 out
of phase with the field radiated in the backward direction from the aperture.
With the appropriate selection of reflector dimensions and substrate thickness,
the magnitude of these radiated fields can be made almost equal. This results
in the cancellation of the total radiated field in the backward direction. In
Chapter 3, an infmite ground plane was used in the analysis, producing almost
ideal cancellation of the backfire fields. In a more realistic scenario, the work
here uses a small, finite ground-plane, hence the cancellation effects are also
influenced by edge diffraction. The reflector dimensions used were quite
different when compared with a frequency scaled version of the reflector used
in Chapter 3, highlighting the impact of the finite ground plane.

Figure 6.2.16 depicts the return loss for the antenna, with and without
the reflector element. A 10 dB return loss bandwidth of 23% is achieved. As
the thickness of the foam reflector substrate is relatively large, the reflector
element has little effect on the input impedance of the antenna. The E and H
plane radiation patterns shown in Figures 6.2.17, 6.2.18 and 6.2.19 highlight
the reduction in field radiation in the rear half plane of the antenna when the
reflector element is added to the CPW fed ASP. The antenna and reflector
dimensions are the same as those given in Figure 6.2.15. Reduction of the
total power radiated into the rear hemisphere of the antenna is evident across
the entire impedance bandwidth, emphasizing the broadband ability of this
technique. The major field cancellation occurs in the region around backfire
(180°), which is particularly noticeable in the E-plane. In the forward
hemisphere of the H-plane the total radiated power has increased, and the
indentations in the pattern are smoothed over. Minor alignment errors were
experienced in the measurement of the radiated fields.
o

"\ ./

,
·5
/
--8
iii'
'C
·10

·15
1
r <, I
=C · 20
::I
~,f I
Cl
III
== ·25
r I
·30
~ - without reflector
• • •• with reflector
II
·35
5 5.5 8 8.5 7 7.5
Frequency (GHz)

Figure 6.2.16 Return Loss of CPW fed ASP with and without Reflector Patch Element
296
(a) (b)
Figure 6.2.17 Measured Far Field Radiation Patterns of CPW fed ASP with and without
Reflector Patch Element at 6.1 GHz: (a) E-plane; (b) H-Plane

(a) (b)
Figure 6.2.18 Measured Far Field Radiation Patterns of CPW fed ASP with and without
Reflector Patch Element at 6.7 GHz: (a) E-plane; (b) H-Plane
o

110 100

(a) (b )
Figure 6.2.19 Measured Far Field Radiation Patterns of CPW fed ASP with and without
Reflector Patch Element at 7.3 GHz: (a) E-plane; (b) H-Plane
297
A 2 nun offset of the reflector element in the positive direction along
the y-axis produces a reduction in the rear directed radiation in the H-plane, as
seen in Figure 6.2.20. This effect is particularly evident in the sidelobes away
from backfire. The E-plane pattern remains relatively unchanged.

1&0 1&0

(a) (b)

- H-plane
••••• E-plane
1&0

(c)

Figure 6.2.20 Measured Radiation Patterns of CPWfed ASP with + 2 mmy-axis Offset:
(a) at 6.1 GHz; (b) at 6.7 GHz; (c) at 7.3 GHz

If the desired field cancellation is away from backfire, the reflector


can be modified to tailor the resulting field pattern. Figure 6.2.21 shows the
measured E- and H-plane radiation patterns for an antenna with a reflector
element 5 nun wide. All other parameters of the antenna and reflector are as
in Figure 6.2.15. Severe dips in the pattern can be seen away from backftre in
both principle planes. The effect is most pronounced in the E-plane.
Adjustment of the reflector width can shift the location of these dips.

298
Theoretical trends infer that an increase in the width of the reflector patch
shifts the E-plane dips away from backfire.
o

110 110

(a) (b)

_ H-plane
110 ••••• E-plane
(c)

Figure 6.2.21 Measured Radiation Patterns of CPW fed ASP with 5 rom Wide Reflector
Element: (a) at 6.1 GHz; (b) at 6.7 GHz; (c) at 7.3 GHz

6.2-5 Dual polarized version of slot-coupled solutions

6.2-5.1 Introduction

In mobile communication systems incorporating single linear


polarized antennas, polarization loss is encountered if the mobile unit
polarization varies from that of the Antenna Remote Unit (ARU) or base
station. A propagating wave can be de-polarized in certain transmission
media due to environmental factors such as rain or ionospheric conditions.
Random scattering from objects and obstructions in urban/suburban
environments can lead to polarization cross-coupling. For these reasons,
picocellular mobile architectures, as well as other applications, may utilize
299
dual polarized antennas in the ARUlbase station to reduce the effect of
handset orientation. This requires simultaneous emanation and reception of
orthogonal linear polarizations from the ARUlbase station antenna.

Dual polarized antennas that are suitable for integration with


microwave and photonic devices enable a simple, small, lightweight and
robust ARUlbase station to be designed, with enhanced wireless performance
over a single linear polarized system. For large-scale fiber-radio networks
with numerous ARUlbase station modules, such improvements to the can lead
to a substantial reduction in infrastructure costs.

There has been the occasional dual polarized antenna design reported
that has the desired characteristics for integration with high permittivity
MMIC or OEIC substrates. Hettak et al. have proposed CPW fed aperture
coupled antenna designs on a single layer of high dielectric constant material,
for both simultaneous orthogonal linear polarization [16] and polarization
diversity [17] applications. However, these structures display a very narrow
impedance bandwidth (approximately 2%) that could be a limiting factor to a
communications system design. Stotz et al. demonstrated the use of silicon
nitride membranes to suspend a parasitic patch above microstrip line fed
coupling apertures on a Gallium Arsenide substrate for both single and dual
polarizations [18]. These configurations also had a narrow impedance
bandwidth, and a complex fabrication procedure. None of the aforementioned
designs made reference to the gain or efficiency of the antenna to determine
the influence of surface wave excitation.

Circular Polarization (CP) can also be employed to reduce multipath


fading. A right-hand circularly polarized wave changes sense upon complete
reflection to a left-hand circularly polarized wave, and vice versa. Hence the
reflected wave is cross-polarized to the transmitted wave [19]. CP can also
alleviate polarization mismatch, scattering and de-polarisation as a circularly
polarized wave contains all the information of a linearly polarized wave of
arbitrary sense, at reduced amplitude [20]. However, it is generally a more
complex method of achieving increased performance, compared to dual
polarization.

Consequently, CP transmission systems often have a greater


probability of connectivity when compared to a single linear polarized
arrangement. The reflective and transmissive behaviour of linear polarized
waves is dependant on the grazing angle (angle of arrival), while CP waves
can provide better coverage into buildings/vehicles, into shadowed areas
caused by hills or other obstructions [20]. As with dual polarized antennas, a
circularly polarised antenna that can be integrated into MMICIOEIC
300
structures can potentially provide an efficient, small, low cost ARUlbase
station solution for fiber-radio architectures.

6.2-5.2 Single feed dual polarized (SFDP) solution

Figure 6.2.22 illustrates the feed configuration used to achieve a dual


linear polarized CPW fed ASP with a single feed line. A diagonal CPW line
terminated with an open circuit stub excites orthogonally crossed resonant
apertures. The SFDP antenna structure, consisting of the feed and parasitic
patch element, was simulated and optimized on HFSS [8].

Figure 6.2.22 Dual Polarized Single Feed Configuration

The simulated and measured results for the return loss of the SFDP
antenna are given in Figure 6.2.23. The measured results show a slight shift
in the lower frequency resonance. This shift could be caused by tolerances in
the substrate permittivity (to.25 for Duroid 6010.2), or small air gaps
between substrate layers in the fabricated antenna. Otherwise, reasonable
agreement between theory and experiment is achieved. A measured
bandwidth (defined as >10 dB return loss) of 19% was observed.
o
--- .. .... .
............
I\. "
~~
·5

-
in 'Y
~ ~
\ "
·10
~
.3 V
, 'I,'
· 15
c
:; ,
OJ
. L
·20
VI
a: I , ' ~
••••
Simula led
Measured
·25
I
It

·30
i
5 5.5 I 1.5 7 7.5 I
Frequency (GHz)

Figure 6.2.23 Return Loss of SFDP Antenna

The co- and cross-polarized radiated fields for the SFDP antenna in
the ep =0° , 90° and 45° planes at 6.5 GHz are represented in Figure 6.2.24a, b
301
and c, respectively. For most angles within ±45° from broadside, the co- and
cross-polarized fields are of approximately equal magnitude for all three
planes depicted. This implies that the antenna is capable of receiving
arbitrary polarization angles at a relatively constant level. Such a
characteristic can alleviate polarization loss concerns at a mobile cellular base
station, where the incoming radiation from the handset may vary in
polarization due to the orientation of the handset with respect to the base
station antenna. Minor alignment errors may have been incurred during
measurement. Also note the relatively low front to back ratio of less than 10
dB, common to resonant aperture antennas (see Chapter 3). The radiation
characteristics presented in Figure 6.2.24 remained fairly constant over the
operational bandwidth of the antenna. The theoretical antenna gain is above 7
dBi across the entire return loss bandwidth, and the antenna exhibits high
efficiency.

'OIl 'OIl

(a) (b)

-90 1-lr+-+-+-?lE-+-f-j-t;~

Co -polarization
Cross-polarization

110

(c)

Figure 6.2.24 Measured Radiation Patterns of SFDPAntenna at 6.5 GHz: (a) cjl = 00
plane; (b) cjl = 90° plane; (c) cjl = 45° plane

The bandwidth of the SFDP antenna can be expanded by stacking an


additional parasitic patch with a foam substrate on to the top of the standard
configuration. After optimization, a return loss bandwidth of 33% for the
302
stacked arrangement was achieved. Figure 6.2.25 shows a comparison of the
simulated and measured return loss at the feed of the stacked antenna. Again,
there is a slight discrepancy in the position of one of the resonances.
However, the predicted and measured bandwidths are equal. The stacked
SFDP antenna displayed similar radiation and gain characteristics to the
single parasitic patch element SFDP antenna, with the exception of a couple
of frequency points where the cross-polarized radiation at broadside was 1 or
2 dB lower than the co-polarized field in one of the planes. This could be the
effect of alignment errors in the fabrication of the antenna and in the antenna
range measurements.
0

·5
iii'
...
~

0
· 10

-J
C ·15
~CD
a:
·20

5.5 8 8.5 7 7.5


Frequency (GHz)

Figure 6.2.25 Return Loss of stacked SFDP antenna (aperture lengths = Ilmm, aperture
widths = lmm, CPW strip width =O.255mm, CPW gap width =O.l7mm, open circuit stub
length = l.lmm, foam thickness =3mm, patch size = 16.6mm2, additional foam thickness =
=
3mm, additional patch size 14.6mm2)

6.2-5.3 Dualfeed dual polarized (DFDP) solution

Polarization diversity applications require the independent control of


orthogonal polarizations to enable the enhancement of wireless links over a
fixed frequency band. The feed structure shown in Figure 6.2.26 uses two
CPW lines to feed separated orthogonal apertures to attain this independent
control. The 'T' configuration of resonant apertures provides high isolation
between the two feeds [21], but results in an asymmetry of the aperture
positions under the parasitic patch element. HFSS was again used to analyse
the DFDP antenna structure that consisted of the feed layout, parasitic patch
element, and their related substrate materials.

The simulated S-parameters for the DFDP antenna are displayed in


Figure 6.2.27a. The asymmetry in the feed aperture positions under the
parasitic patch element causes the variation in the SII and 5 22 responses, as the

303
dimensions of the apertures are identical. The magnitude of 5 21 demonstrates
the high level of isolation achieved by arranging the apertures in the 'T'
configuration. The measured 5-parameter results shown in Figure 6.2.27b
confirm the predicted isolation, remaining below -29 dB across the region of
interest. The simulated and measured 5 11 results show excellent agreement in
resonant frequencies and bandwidth. However, the 5 22 results show a reduced
bandwidth due to the influence of a strong parasitic effect at approximately
7.25 GHz caused by the interaction of the aperture and the edges of the small
ground plane used. This parasitic effect was predicted by HF55 at around 7.6
GHz, but it did not impact on the impedance bandwidth performance of the
antenna. A similar effect can also be seen in the measured 5 11 results at 7.9
GHz. It was found both analytically and experimentally that varying the size
of the ground plane and feed substrate can alter the position and severity of
the parasitic effect. The unaltered configuration exhibited a measured
bandwidth of 14% common to both ports of the DFDP antenna.

PorI 2

Port I
Figure 6.2.26

-
Dual Polarization Dual Feed Configuration

~ /' ~

...
. , .......... ,..
V

\
~

\ .
"
, - ".'
.",
\
..
I
~' .
' .

a ~5 I
Frequency (GHz)
(a)
~5 7 7.S

'8]"•••••••• Su
_ . - S21

iii'
~ -10
~
~I r<-. "
V· ,, .
\.,....-: ~

,.
.§ ...
~
... ...
IV
" " .
~
~ ... .
.' ~,\., ; .v,"
E .IIj, ..
:""
:J
;! -so
... I ~5 U J U
Frequency (GHz)
(b)
Figure 6.2.27 Simulated and Measured S-parameters of DFDP Antenna

304
The co- and cross-polarized radiated fields for port 1 of the DFDP
antenna in the H- and E-planes are represented in Figure 6.2.28a and b,
respectively. Port 2 results are given in Figure 6.2.28c and d. Cross-
polarization levels at broadside are more than 20 dB below the co-polarized
levels, giving a high level of polarization isolation . This is not only attractive
for polarization diversity applications, but also the generation of circular
polarization could be easily achieved by attaching a 900 hybrid to the two
input ports. Away from broadside the cross-polarization level increased to
approximately 13 dB down from the co-polarized level, implying that the
antenna is more suited to point-to-point links in its current state. As
mentioned previously, minor alignment errors may have been present in the
measurement of the radiation patterns . Relatively low front to back ratios are
again observed as a result of the resonant apertures. The cross-polarization at
broadside remains very low over the bandwidth common to both ports of the
DFDP antenna . The theoretical antenna gain is in excess of 7 dBi across
almost the entire common return loss bandwidth with high efficiency. The
distorted patterns observed at port 2 of the DFDP antenna at 7 GHz are a
consequence of the parasitic impedance effects seen in Figure 6.2.27, caused
by the aperture/ground plane interaction. The marginally higher cross-
polarization level at broadside for port 1 of the DFDP antenna at 7 GHz is of
little concern, as this frequency point is outside the return loss bandwidth
common to both ports of the antenna.

o o

(a) (b)

305
o o

180 180

(c) (d)
Figure 6.2.28 Measured Radiation Patterns ofDFD Antenna at 6.5 GHz:(a) H-plane (Port
1); (b) E-plane (port 1); (c) H-plane (port 2);(d) E-plane (port 2)

6.2-5.4 Reflector Elements for Dual Polarized Antennas

The low front to back ratio of the SFDP and DFDP antennas could
pose a problem in applications that use sectorial coverage, or where other
circuitry or support structures are mounted behind the antenna. Microstrip
reflector elements can be attached to the rear of antennas that use resonant
aperture coupling to reduce the level of power radiated into the rear
hemisphere (refer to Chapter 3). This method of back radiation reduction does
not promote the propagation of parallel plate modes or having any major
manufacturing problems, as can be the case with other methods. In this
section, the concepts for cancellation of rear directed radiation outlined before
are applied to dual polarized antennas.

SFDP ReflectorElement

The stacked SFDP antenna uses crossed apertures to achieve dual


polarization. To cancel the rear directed fields, perpendicular reflector
elements of length 27.3 mm and width 0.2 mm are arranged in a cross
formation directly beneath the apertures. A 5 mm thick foam dielectric is
used for the reflector, which is etched on a 0.254 mm thick sheet of RT
Duroid 5880. Due to the large spacing between the apertures and the reflector
element, the input impedance remains virtually unaffected. Each arm of the
reflector produces a field with approximately equal magnitude to the
orthogonal aperture, but with a phase shift of 1800 , hence theoretically
306
canceling the total radiated field at backfire if an infinite ground plane is used.
Diffraction at the edges of small finite ground planes must to be taken into
consideration when using small finite ground planes. However, with careful
design of the reflector element the total power radiated into the rear
hemisphere of the antenna can be reduced.

Figures 6.2.29a, b, c and d compare the measured far field patterns for
the stacked SFDP antenna with and without the crossed reflector element.
The power radiated into the rear hemisphere is significantly reduced with the
addition of the reflector, demonstrating a similar order of improvement to that
achieved in [22]. The results shown are from the middle of the return loss
bandwidth, at 6.7 GHz. Similar improvement is observed across the entire
33% bandwidth, exemplifying the broadband ability of the technique.
o o

180 180

180 180

(c) (d)

Figure 6.29 Comparison of Radiated Fields of SFDP Antenna with and without
Reflector Element at 6.7 GHz: (a) $ = 00 plane (co-pol); (b) $ = 00 plane (cross-poL); (c) $
=900 plane (co-pol.): (d) $ =900 plane (cross-pol.)

307
DFDPReflectorElement

As the DFDP antenna has apertures that are separated from each
other, individual reflector elements of length 27.3 mm and width 0.2 mm are
placed directly beneath, but perpendicular to the apertures, as shown in Figure
6.2.30. The foam thickness is 5 mm, and the two reflector elements reside on
opposite sides of a 0.254 mm thick sheet of RT Duroid 5880. Again, the
reflector element has minimal influence on the input impedance of the
antenna.

l.
r
On reverse side of

~~:~~~~:~fl~~rto.,?
elements
Figure 6.2.30 Reflector Element Configuration for DFDP Antenna

The effect of adding the reflector elements on the measured far field
patterns for the DFDP antenna is apparent in Figure 6.2.31a-d. Once again
the rear directed radiation is suppressed when the reflectors are attached. The
results given are at 6 GHz, which is at the lower end of the common
impedance bandwidth. Field patterns at higher frequencies show a reduction
in the level of improvement, when compared to the DFDP antenna without the
reflectors. The decrease in rear field cancellation is a consequence of the
parasitic impedance effects seen in Figure 6.2.31b, which causes the field
radiated by the apertures to change due to their interaction with the ground
plane. Hence the reflector element no longer satisfies the radiated field
criteria of similar magnitude to the apertures with approximately 1800 phase
shift. Further investigation must be undertaken to optimize performance and
the reflector elements for the DFDP antenna.
o

110 110
(a) (b)

308
- Withoutreflector
••••• With reflector

180

(c) (d)

Figure 6.29 Comparison of Radiated Fields of DFDP Antenna with and without
Reflector Element at 6.7 GHz: (a) ep =0° plane (co-pol.); (b) ep =0° plane (cross-pol.); (c) ep
= 90° plane (co-pol.); (d) ep = 900 plane (cross-pol.)

6.3 Direct Contact Procedures and Solutions

6.3-1 Introduction

The previous section showed that aperture coupled patches alleviate


problems with the excitation of surface waves reducing the antenna efficiency
and also the limited achievable bandwidth to an extent, however these patches
can suffer from layer alignment difficulties and the surface wave excitation
can still have a detrimental effect on the antenna performance [23].

Several other methods have been proposed to address the problem of


printed antenna/active device integration with, however somewhat limited
success [24, 25]. These techniques were described in Chapter 4 and are based
on the principle that if the patch is large enough, the TMo surface wave will
not be excited. Such methods can yield enhanced surface wave efficiency;
however, these proposals are typically narrow band, which limits their
application. Recently it was shown that by simply stacking the patch etched
on the high dielectric constant material with a patch mounted on a low
dielectric constant material, in this case foam, the previously encountered
problems can be resolved (see chapter 3). Bandwidths in excess of 25 % were
achieved with the surface wave efficiency greater than 85 % over this band. It
was postulated that the upper patch couples strongly to the typically excited

309
surface thereby reducing its effect on the overall radiation performance of the
antenna.

In this section we further investigate the performance of the hi-lo


dielectric constant stacked patches, focusing on the impedance and radiation
characteristics of this antenna. It will be shown that these patches yield high
efficiency solutions and low cross-polarized fields making this printed
antenna very suited to circular polarization (CP) applications. Section 6.3-2
presents the overall performance trends of the hi-lo stacked patch, comparing
its performance to single layered and other stacked configurations. Section
6.3-3 presents two simple techniques using the hi-lo stacked patch, which
result in good CP behavior over the entire impedance bandwidth. These
configurations are investigated experimentally and excellent agreement with
the predictions was achieved. Section 6.3-4 presents a simple procedure for
enhancing the bandwidth of the integrated antenna and Section 6.3-5 shows
some results of when the antenna is developed on typical OEIC material.

6.3-2 Properties of hi-lo Stacked Patches

To thoroughly investigate the performance of hi-lo stacked patches a


full-wave spectral domain analysis developed in house [26] and Ensemble 5.1
were utilized. Figure 6.3.1 shows the printed antenna configuration. Here an
arbitrary shaped patch is etched or grown on a grounded high dielectric
constant material. A second patch residing on a layer of low dielectric
constant material is parasitically coupled to the driven antenna. The lower
patch can be edge or probe fed (here shown as edge fed). As stated in Chapter
3, this technique to achieve high surface wave efficiency is independent of the
patch conductor shape. Figure 6.3.2 shows the predicted and measured input
impedance behavior of a hi-lo stacked patch configuration incorporating
rectangular patches and a probe feed (refer to the caption of Figure 6.3.2 for
the relevant dimensions). The lower layer has a dielectric constant, Erh of
10.4 and a thickness, d., of 1.905 mm (or 0.032 Ao at the design frequency of
= =
5.0 GHz). The upper layer is foam (Erz 1.07, d 2 4.5 mm (or 0.075 Ao». As
can be seen from this Smith Chart very good agreement between theory and
experiment was achieved. Here the measured and predicted 10 dB return loss
bandwidths are 27 % and 26 %, respectively. A thin dielectric layer (0.254
mm) of RT/Duriod 5880 was used to etch the top conductor and was taken
into account using the Ensemble 5.1 package. To put these results into
perspective, a single patch etched on the lower dielectric layer has a 10 dB
return loss bandwidth of 1.6 %. The measured gain of the hi-lo antenna was
greater than 6.5 dBi over the 10 dB rettJ!l1 loss bandwidth. This value is

310
comparable to a stacked patch incorporating low dielectric constant material
combinations (see Chapter 3), inferring that the surface wave efficiency is
relatively high.
Low dielectric
conslant material Patch conductor

High dielectric constant


grounded substrate

Figure 6.3.1 Hi-lo Stacked Patch Antenna

Figure 6.3.2 Input Impedance of Rectangular hi-to Stacked Patch Antenna (parameters:
=10.4, d, =1.925 mm, tan c~h =0.001, ea =1.07, d2 =4.5 mm, tan ~ =0.001, L I =9.4 mm,
Erl
WI =8.8 mm, ~ =21.4 mm, W2 =20.4 mm, xp =4.3 mm, ro =0.325 mm)

Figure 6.3.3 shows the surface efficiency, Tlsw, of two hi-lo


configurations as a function of normalized frequency (normalized to the
center of the 10 dB return loss bandwidth, fc) ' Each stacked patch uses a
lower layer with e, = lOA and d =0.032 Ac and an upper layer of s, = 1.07 and
d = 0.067Ac (where Ac is the wavelength corresponding to fc) . The first

311
configuration consists of a stacked rectangular patch combination and the
second is for a stacked annular ring scenario. As for the circular patch case
presented in [7], Tlsw is greater than 85 % over the 10 dB return loss band of
frequencies (ring: 0.895 r, to 1.105 fc; rectangular: 0.885 fc to 1.115 fc) . Note
the peak in efficiency for the rectangular case is approximately 93 % and is
near the upper edge of the 10 dB return loss bandwidth. After this peak Tlsw
gradually decreases until the two patches no longer strongly couple resulting
in the rapid decay of Tlsw. Eventually Tlsw approaches values for the case of a
single layered geometry etched on the high dielectric constant material (refer
to Chapter 2). For the annular ring case note the maximum efficiency is
slightly higher than the rectangular patches at the expense of reduced
bandwidth where the efficiency is high (above 85 %). Indeed the impedance
bandwidth of the annular ring configuration is slightly lower too,
approximately 21 % compared to 23 % for the rectangular hi-lo case. Thus
there appears to be a trade-off between maximum surface efficiency and both
return loss and Tlsw bandwidth, as for the case for a conventional stacked patch
configuration. Once again to put these results into perspective, a patch etched
on the lower dielectric material alone has a surface wave efficiency of 66 %.
It is interesting to note that for all the cases considered; including a circular
patch hi-lo structure, the maximum peak in efficiency does not match the
lowest return loss value. This is similar to the cases presented in Chapter 3,
although there is a better correlation here.
tS Ic!- _
-
ltS .-.- v ........
10
......

f:
'\.\
\.
!l70 ~

lIS I' ---I


• ring \
~
1cil
10
\:
...' ....
S5
50
45
0.1 o.t t 1.1 1.2 t.a fA t.I
Nornwllud Frequency

Figure 6.3.4 Surface Wave Efficiency of hi-lo Stacked PatchAntennas

As mentioned in Chapter 3, the hi-lo stacked patch antenna has low


cross-polarized radiation levels. Figure 6.3.4 shows the theoretical H-plane
cross-polarized levels for four probe-fed patch configurations: (1) a single
layered circular microstrip patch antenna mounted on a low dielectric constant
material (Er = 2.2); (2) a stacked circular microstrip patch configuration
utilizing a 10-10 combination of materials (Erl =2.2 and Er2 =1.07); (3) a single
layered circular microstrip patch antenna mounted on a high dielectric
constant material (e, = 10.4); and (4) a stacked circular microstrip patch

312
configuration utilizing a hi-lo combination of materials (Er! = 10.4 and Erz =
1.07). All combinations have been designed for the same center frequency
and the thicknesses of lower dielectric layers are also the same (0.03 Ao). For
the stacked cases the upper dielectric layer has a thickness of 0.06 Ao. From
Figure 6.3.4, there are several important findings. Before exploring these, it is
important to note that the main source of cross-polarized fields is the
discontinuity associated with the feed (here a probe) and the driven patch
conductor. This is why direct contact fed patches have greater cross-
polarization levels than non-contact printed antennas (such as aperture
coupled patches).
single layer (10) __--.-__

.00

Figure 6.3.4 Cross-polarization Levelsof Several Probe-fed Patches

The first trend to observe from Figure 6.3.4 is that the cross-polarized
field level for the 10-10 stacked patch combination is less than for the single
layered patch utilizing a low dielectric constant substrate (here by 5 dB). This
observation can be attributed to the presence of the parasitically coupled
patch. As there are no discontinuities associated with this second patch and
the centers of each patch are aligned, as well as the second element being
reasonably far from the discontinuity associated with the driven patch and the
probe, it can be expected that the cross-polarized radiated power will be
lower.

From Figure 6.3.4 it can be seen that the single layer patch using a
high dielectric constant substrate has significantly less cross-polarization than
the patch with e, =2.2. This is a result of the cross-polar fields being strongly
bound to the material [27]. The hi-lo stacked patch has slightly higher cross-
polarization levels (5 dB) than the single layer scenario due to the
parasitically coupled element drawing power from the lower element. Thus
there is one minor trade-off of the new configuration, the hi-lo patch has far

313
superior llsw and impedance bandwidths compared to the single layer
geometry at the expense of slightly higher cross-polarization levels. It should
be noted that the cross-polarized fields of the hi-lo stacked patch is
significantly lower (15 dB) than the conventional stacked configuration. This
makes the hi-lo structure a prime candidate for circular polarization
generation, where the cross-polarization level governs the quality of the axial
ratio (AR).

Figure 6.3.5 displays the E-field magnitude for a two dimensional E-


plane cut through a typical hi-lo stacked patch antenna. As in Figure 6.1.6 for
the benchmark antenna, a propagating wave is evident on the microstrip line,
showing field concentrations in the high dielectric constant substrate between
the strip and ground plane. However, the hi-lo stacked patch configuration
does not show the high field concentrations under the driven patch element.
The strong mutual coupling of the two printed elements draws the field out of
the high permittivity substrate towards the parasitic patch. This is believed to
be the mechanism that reduces the surface wave excitation in feed substrate,
and allows the highly efficient operation of the hi-lo stacked antenna. Hence,
the hi-lo stacked patch structure provides a solution to both the impedance
bandwidth and surface wave excitation concerns that arise when integrating
an antenna element with photonic/microwave devices on high dielectric
constant materials.
Drive n patch element Parasitic patch clement Microstrip reed line

High
permittivity
substrate

Foam
die lect ric Gro und plane

Figure 6.3.5 E-field intensity plotof Hi-lostacked Patch

6.3-3 Circularly Polarized Hi-IoStacked Patches

There are a variety of methods for generating circular polarization


from a microstrip patch antenna (see Chapter 2). One of the simplest means is
to use a single feed arrangement. Here the feed is located in the D-plane (cj) =
45°) on an almost square or circular patch to excite two orthogonal modes
with a 90° phase difference. As mentioned in Chapter 2, the phase shift is
achieved by raising the resonant frequency of mode above the other and
operating between the two resonances. This method requires no additional
phase shifting element, however, as h