TI FRE (COPI

NAVAL POSTGRADUATE SCHOOL

*Monterey, California
.LO

|" 6
MODELING THE BEHAVIOR OF THE
LINEARLY TAPERED SLOT ANTENNA
0
I!
by

1
Thomas P. Kelly

THDecember sI
1988

Thesis Advisor: Ramakrishna Janaswamy

Approved for public release; distribution is unlimited

~E DT!C
MAR1 4198
 UNCLASSIFIED A 5
Form Approv.ed
REPORT DOCUMENTATION PAGE OMB fC 0 704 0188
"
1a PEP0P S CA-, C_ ASS -CA ,ON P;SQ 1 ,V %GS
UNCLASSIFIED
2a SECURITY CLASS F:CA' ON ATrOR TY 3 DSTR 3B ON A' AB v OF PPP"
Approved for public release;
2'bDECLASSIF'CA'CN DO\GRAD:,NG SCihEDuLE distribution is unlimited
T
4 PERFORMING ORGANZA-;ON REPORT NUMBER(S) S MONiTORG ORGAN ZA ON REPORT %_%(B- S

6a NAME OF PERFORM!NG ORGANIZATION 6b OFFICE SYMBOL 7a NAME OF WON'TORNC'G ORGAN ZA- ON

(If applicable)
Naval Postgraduate School 62 Naval Postgraduate School
6c ADDRESS (City State, and ZIP Code) 7b ADDRESS (City State and ZIP Code)

Monterey, California 39943-5000 Monterey, California 93943-5000

"
8a NAME OF .JND
G SPONSOR %G SO OFFICE SYMBOL 9 ;PC,( EN.N NSR..M DEN CA O*. '
ORGANIZAON, (If appcable)

BC ADDRESS(City State, and ZIP Code) In SOiPCE OF FNDIG 'i BEdS

PROGRAM I=OE;C- ,
jS , .
E__ _ _ __ro 0 NO -__CSS 0N NO

11 TiTLE (Include Securit) Classification)

MODELING THE BEHAVIOR OF THE LINEARLY TAPERED SLOT ANTENNA

12 PEPSOA_ A7.-HORIS,
KELLY, Thomas P.
1
3a 'YPE OF REPOT 3b TIME COVERED 14 DATE 0c REPOR' (Year Monh Day) 5 PAC- ,.
Master's Thesis FROV O_ 1988 December 142
1 SUPPEEN'APv NO-ATU'4 he views expressed in this thesis are those of the
author and do not reflect the official policy or position of the Depart-
ment of Defense or the US Government
'7 - COSATI CODES 18 SUBJECT TERMS (Continue on reverse if necessary dnd identify tb block numner)
"E.D ,1
POP SUB-GROUP LTSA; Flare Angle; Substrate; Dielectrical;
Antenna; Planar Antenna; Radiation Pattern;
Moment Method; The Linearly Tapered Slot Antenna
'
9 ABSR.AC :Continue on reverse if necessary and identify by block number)
The Linearly Tapered Slot Antenna (LTSA) had been investigated and devel-
oped experimentally; its applications have primarily been based on
empirical designs. An accurate theoretical model based on Moment Methods
(MM) is developed here to study the radiation characteristics of the
LTSA. Using the MM solutions to the reaction integral equation, this
thesis presents an analysis to model and explain the LTSA behavior. The
effects of variable design parameters on radiation patterns are studied.
Discussion is augmented by relating preducted radiation patterns to cal-
culated current distributions on the antenna surface. Conclusions are
iade regarding optimum designs for the LTSA. Relevant observations are
made concerning the extensive computational tasks and the computer
resources required for the MM model. - / It

20 D'STP R:J-fO, A, 1 45. Y Q ABSorAC' 2' ARSu9,ACr S( 1 " (4 A, - (,4

CR JNCLASSIF'ED .,NL'M"ED El SA%'E AS PP AC jSEPS
1" UNCLASSIFIED
27 a 7AM; O P7POS -B '.D /:DUAL I2t ELL.P-1ONE (lriciue Aed Code) i v :,
JANASWAMY, R. 408-646-3217 62Js
DD Form 1473, JUN 86 Previous editionsare obsolete P ,V (-AS I CI , ',

.3/N ) (I2-L]V-(; i4-flJ UNCLASSI FIED

i
 Approved for public release; distribution is unlimited

Modeling the Behavior of the

Linearly Tapered Slot Antenna

by

Thomas P. Kelly
Major, United States Army
A.A.S., Delaware Technical and Community College, 1971
B.S., United States Military Academy, 1976
Submitted in partial fulfillment of the
requirements for the degree of

MASTER OF SCIENCE IN ELECTRICAL ENGINEERING

from the

NAVAL POSTGRADUATE SCHOOL

December 1988

Author: A o -Thjmas P. Kelly

Approved by: _ m_

Ram swamy, Thesis Advisor

Michael A. M&tgan, Second Reader

Jolm-_. Powers, Chairman, Department of

Electrical and Computer Engineering

Gordon E. Schacher
Dean of Science and Engineering

ii
 ABSTRACT

The Linearly Tapered Slot Antenna (LTSA) had been

investigated and developed experimentally; its applications

have primarily been based on empirical designs. An accurate

theoretical model based on Moment Methods (MM) is developed
here to study the radiation characteristics of the LTSA.
Using the MM solutions to the reaction integral equation,
this thesis presents an analysis to model and explain the
LTSA behavior. The effects of variable design parameters on
radiation patterns are studied. Discussion is augmented by
relating predicted radiation patterns to calculated current
distributions on the antenna surface. Conclusions are made

regarding optimum designs for the LTSA. Relevant

observations are made concerning the extensive computational
tasks and the computer resources required for the MM model.

Accession For
I-NTIS G7A&I
DT IC .. ! -

r
¢"
I
I----
A "',,,
t~v:
Ility Codes
1 n nd/or
Lb t I ,,? J
 TABLE OF CONTENTS

I. INTRODUCTION ........................................ I

II. THE MOMENT METHOD ................................... 4

III. LTSA MODEL .......................................... 25

IV. PREDICTED PATTERN BEHAVIORS ......................... 39

V. CONCLUSIONS OF MODELING STUDY ....................... 57

APPENDIX A LTSA RADIATION PATTERNS ....................... 60

APPENDIX B LTSA RADIATION CHARACTERISTICS AS FUNCTIONS OF

THE LENGTH AND HEIGHT ......................... 86

APPENDIX C LTSA CURRENT DISTRIBUTION ..................... 96

LIST OF REFERENCES ........................................ 136

INITIAL DISTRIBUTION LIST ................................. 137

iv
 I. INTRODUCTION

The many practical advantages of the Linearly Tapered

Slot Antenna (LTSA) are based on its low cost, light weight
and relative simplicity. Specific designs of LTSA arrays
have been implemented for millimeter wave imaging systems and
more applications are inevitable. However, since the

refinement of an accurate theoretical model has lagged its

development, LTSA uses have been based on empirical designs.

Based on the proven Moment Methods (MM) model, developed

by Janaswamy [Ref. 1], this paper presents an analysis of the
LTSA radiation pattern. The model permits a parametric study
of the antenna. The physical antenna variables used in the

LTSA modeling are illustrated in Figure 1.1.

Since the method of this research is the MM model, a
general review of MM and a specific discussion of the LTSA
model are included in Chapters 2 and 3 respectively. It is

intended that these chapters address both the accuracy of MM

as well as the limitations, particularly with respect to

solution by computer. By way of examples, Chapter 2 is

demonstrative of the solution by the Moment Method. The MM

concept is then advanced in Chapter 3 to explain how the

reaction integral equation is solved in the LTSA model.

Chapter 3 further serves to discuss the model in terms of
computer tasking considerations and approximations which
 Copper plate -

..... ....

D..e.e.....

Feed Gap...
.L.........
.....
......
Copperp.....e
g... .......

Figure
Obliqu.View.of
11 e.L..
..

2.......
alleviate computational burden with little or no sacrifice in
accuracy.
Chapter 4 focuses on the LTSA modeling results and

analyzes the observed effects of changing beamwidths and

sidelobe suppression in the E- and H-planes. Of particular

interest are the pattern changes which result from various

combinations of antenna length and height, but effects of
changing other design parameters are likewise investigated.
corresponding to the radiation pattern changes, surface
current distribuLions are used to explain some phenomena
which appear to synergistically create the different
patterns. (Appendix C contains current distribution graphs
for the respective radiation patterns illustrated in Appendix

A).
Conclusions are summarized in Chapter 5. The thrust

therein is to provide a credible spectrum of design

considerations for the LTSA. It is fully expected that this

work has provided the necessary data and rationale for

relevant interpolation as well as extrapolation. Moreover,

the methods used in this study may well serve similar studies
of other antennas and scatterers.

3
 II. THE MOMENT METHOD

A. CONCEPT

Although the Moment Method (MM) is a widely used

technique in solving electromagnetics problems, only the

advances in computer technology of recent years enable the

cost-effective use of MM to pursue such rigorous solutions.

Specifically, a linear operator equation with an unknown

function is solved by expanding the unknown function and

casting it into a system of linear equations which is then

solved by known techniques. The thrust of MM is to maintain

the exactness of the original equation and to solve it

nu,erically or by other means. This is normally done at the

expense of additional computer central processing unit (CPU)

time. However, MM generally provides a more accurate

solution in comparison to other techniques which reduce and

approximate equations solely to minimize the computational

effort.

The prescription for solving a problem by MM is best

illustrated by an example. Two such examples are chosen for

this purpose. The first problem is to solve for f(x) in the

Fredhom integral equation of the first kind

1
f(Y)cos(xy)dx = (1-cos y)/y 2 . (2.1)
0

• a i i i4
The exact solution to the above equation is

f(x) = 1-x, O<x<l. (2.2)

While exploring this elementary problem in detail, it is

intended to demonstrate the MM technique and to highlight

both its potential advantages as well as its shortcomings.

The second sample problem is to solve for the current
distribution, I(z') on a short dipole using Pocklington's

integral equation [Ref. 2:p. 307]:

- -L/2 I( ,r aE~' z
3 o.f-L/2
I-~
(z') L
L
2
a
+)3
+z') 2
, '
zz
, =
-Ei(z)(23)
z

The development and analysis of this equation can be found in

most antenna text books and only the application of MM is of

current interest.

B. SOLUTION OF THE FREDHOLM INTEGRAL EQUATION

In solving the Fredholm integral equation, the analytical

approach is somewhat awkward. However, MM is used with

principles of linearity and superposition to reduce the

functional equation to a matrix equation of the form [Ajxi =

[B] [Ref. 3]. Then known techniques are used to determine a

unique solution for [x) = [A]-'[B]. Equation 2.1 in terms of

a linear operator, L, is written as

L(f) = g (2.4)

5
where,

1
L = J
J0
cos(xy)dx (2.5)

g = ('-cos y)/y 2. (2.6)

The first step is to expand f into a set of basis

functions, {fl, f2 ...fm}. N determines the size of matrix to
be used and implies both the degree of accuracy and the
required computational time. The selection of f,. is more

crucial and should be selected to approximate the expected

behavior of f(x). Figure 2.1 shows typical functions with
which to represent f, where N = 5 and

N
f = a f (2.7)
n=1

The problem reduces to solving for "best fit" coefficients,

a,, for each f,,. Substituting equation (2.7) into (2.4)

yields

N
E a..L(f,) = g. (2.8)
n=1

There are now essentially N unknowns, and the number of

linearly independent equations must also equal N for a unique

solution. This is achieved by selecting a set of weighting
functions, {wi, w 2 ...w,} in the range of L and equating the
weighted averages of equation (2.8) for each w. ,
6
0 X1
I
x2
8(x-x2)
x3 x4 x5
(a) Dirac delta function

P(x-Xl)

oX1 x2L0 x4 x5
(b) Pulse function

()Stepapproximnation 41

01

0xl x2 03 x4 x5

(e)piec al sins function.

7
 N
Z a.*<Wm, L(f.)> = <wm,g>. (2.9)
n=1

Here < > denotes the inner product, defined as

1
J f(x) g(x) dx. (2.10)

For M equations and N unknowns, equations (2.9) is

expressed as

[lmi][a] = [g.] (2.11)

where,

<wl, Lf,> <wj, Lf2> ...<wi, Lf,>

[imn] = <W 2 , Lf,> <W 2 , Lf>. .. (2.12)

<wm, Lf,> ............... <wrn, Lfm>

[a.] = [a., a 2 ...a,.] T (2.13)

1g-] = [g1., g2 ...g.] . (2.14)

Thus, the unknown coefficients are determined by

[a.] = [l..-1(g.]. (2.15)

A point not to casually dismiss is the selection of

weighting functions. It is essential that both f. and w, be

linearly independent sets and that the resulting "1" matrix

8
have an inverse and be well conditioned. The infinite
possibilities for f,, and w, allow considerable latitude to
accommodate (a) the degree of accuracy desired, (b) ease of
computation and (c) an acceptable size matrix for inversion.

Choices for w. should be generally based on these criteria

while selections of f,,, are to help approximate the expected
or assumed behavior f(x).
For illustrative purposes, the above Fredholm integral

equation is solved with N pulse functions as basis functions.

The pulse function P(x) is defined as

1 lxi < 1/2N

P(x) = . (2.16)
0 lxI > 1/2N

Thus, a linear combination of

fn = P(x-xn,) (2.17)

gives a step approximation to f(x).

Similarly, N weighting functions are chosen as Dirac

delta functions:

wM = 8(x-x .. (2.18)

with xm taken as the midpoint of each subinterval. This

choice of weighting functions is also known as the point-
matching technique. Physically this choice implies that the

original integral equation (2.1) is now imposed at the

discrete points x.. (the locations of the delta functions),

9
rather than in the continuous interval, O<x<l. These
selections serve primarily to easily demonstrate the
computations rather than to faithfully reproduce f(x).

Figure 2.2 shows a pictorial representation of the basis and

weighting functions.
For a manageable demonstration, set N=M=3. The Fredholm
integral equation (2.1) is expressed in the form of equation
(2.9):

32
E a.8<6(y-xm), Plx-xn) cos(xy)dx>
n=1

= <6(y-xm), (l-cos/y 2 > (2.19)

where the inner product is computed using dummy variable, y.

Therefore, from equation (2.12), each matrix element

1 1
imr = J 00
6(y-x.) J P(x-x.) cox(xy)dxdy

J 1
8(y-x) Jn/3 cos(xy)dxdy
0 (n-l)/3

n/3
cos(x-x.)dx
(n-1)13

10
cf
n n

xlI x2 x3 * 0
]IN 2/N 3/N

Figure 2.2 Pulse Basis Functions to Represent f(x), with Delta

Test Functions Enforced at x M.
 - I X.3
sin
nxm
- - sin
(n-l1)xm

3j
I. (2.20)

In the same manner,

g 6(y-x) - C y 1 Cos Xm (2.21)

Solving equations (2.20), (2.21) and (2.15) yields

.32 .3321 .3301j

[3318 .3226 .30451 (2.22)
.91 .3038 .2553

[g,] = [.4988 .4897 .47171- (2.23)

[an] = [.8319 .5647 .1034]T. (2.24)

This solution is depicted with the exact solution (f(x) =

1-x) in Figure 2.3.

Figure 2.3 seems to suggest that the accuracy of the
numerical solution may be improved by increasing the number
of basis functions. While this may be true in general, it

needs a high degree of precision to actually achieve it. For

a given N, the percent error is less if the precision is

increased. Conversely, for a given precision, increasing N

eventually results in an increase in percent error. The
accuracy will also be affected by moving the matching point,

12
 f(x)

1-

Numerical solution
2/3 -

1/3 -

Exact solution

x
1/3 2/3 1

Figure 2.3 Numerical (by MM) and Exact Solutions for the
Fredholm Integral Equation.

13
 i.e., by shifting the weighting functions 6(x-xm) within the
corresponding subinterval

m-i m
< xM < -. (2.25)
N N

In another approach known as Galerkin's technique, the

set of weighting functions is the same as the set of basis
functions, i.e., w,, = f,,. This results in a symmetric NxN

matrix and reduces the required number of computations needed

for filling the matrix.

Reviewing a few different MM solutions to the Fredholm

integral provides greater insight. Various MM approaches and
results are itemized in Table 2.1. For this problem and the
limited trials indicated, a few conclusions are made:
* The noticeably low value of N is the largest which can
be used before the matrix becomes ill-conditioned. This
is attributable to the available limits of precision in
the mantissa.
* Use of Galerkin's technique does not appear to have a
dominant effect on accuracy.
* Selection of apparently suitable basis functions does
not guarantee convergence, since any N too large may
result in an ill-conditioned matrix.
* Shifting the weighting functions in the point matching
technique does not significantly affect the solution for
[a.].

The ambiguous tone of these conclusions alludes to the

obvious point that few solutions offer the neatness of f(x) =

1-x. What is to be gleaned from this example is that there
are limitations to the MM process. No exact guidelines exist

14
 TABLE 2.1 REPRESENTATIVE SELECTIONS OF BASIS AND
WEIGHTING FUNCTIONS TO SOLVE THE FREDHOLM
INTEGRAL.

Basis Weighting N Precision Solution[a. 1 1

Function Function
Pulse Dirac Delta enforced 3 Single [.8319 .5647
@ center of sub- .10341
interval
Pulse Dirac Delta enforced 3 Single [.8343 .5611
@ left of subinterval .1046]

Pulse Dirac Delta enforced 5 Double [.9622 .6283

@ left of subinterval .5167 .3288
.06411

Pulse Pulse Galerkin 3 Single [.8449 .5451

.1101]

Pulse Pulse Galerkin 3 Double [.8379 .5557

.10641

Powers of x Dirac Delta enforced 3 Single [ .9778

(x°,x'L ... @ x.=(2m-l)/N -. 8841
x" - 1 ) .1072]

Powers of x Dirac Delta enforced 2 Single [ .9991

(X0 ,x' ... @ xm=(2m-I)/N -.98811
xM- 1)

Powers of x Powers of x(Galerkin) 2 Single [ .9974

(x0 ,x'... @ xm=(2m-l)/N -.9946]

15
for the best choice of expansion and weighting functions for
a given integral equation. In spite of this shortcoming, MM
is much more useful in solving an integral equation,
particularly when it cannot be solved in a closed-form, e.g.,
A-cos (wt). This point is reinforced in the next example.

C. SOLVING FOR CURRENT DISTRIBUTION ON A DIPOLE

Consider now an electromagnetics application of MM to
solve for current distribution on a linear dipole. The
motivation for finding current distribution on any antenna is
to eventually solve for the radiation pattern and/or the
input impedance. Since this is the idea of the LTSA model
specifically discussed in Chapter 3, an example of MM to
solve the Pocklington integral equation is worthwhile.
Using the conventional assumptions for thin wire antennas
and illustrations of Figure 2.4, the problem statement is to
solve for the current, I(z') on a short dipole with length, L
= .a oand radius, a = .005X 0 . Assume a 1 volt center-fed
excitation.
In a simpler form, equation (2.2) is written as

1
WEo fL/ - i
[L/2 (Z') K(z,z') dz' = -Ez (2.26)
JoL/2z

where the kernel,

o2 (z,z')
K(z,z') - + B2 2(z'Z) (2.27)

16
 z

Observation
point I
- a

tt
Js

-L

Figure 2.4 Linear Dipole Illustrated as a Wire With Surface

Current Density J s and Observation Point on the Surface.

17
and the free space Green's function is

e- :M
#(z,z') = - . (2.28)
4nR

Equation (2.25) is expanded and simplified for integration:

K(z,z') = [(1 + jBR)*(2R 2 - 3a 2 ) + a 2 B 2 R 2 1.(2.29)

4nR2'

It is readily apparent that the required complex integration

is best done numerically and is well-suited for a modest

computer program. However, for present purposes only, the MM

concept and results need be explained.

Following the same procedure as in the previous example,

the steps are simple: (a) expand I(z') in a set of basis

functions; (b) obtain linear equations by weighting functions
and computing inner products; and (c) use matrix algebra to
solve for the unknown coefficients of the basis functions.
For simplicity sake, N=5 is chosen to limit the matrix
dimensions. For interpretive convenience, I(z') is
approximated by a series of pulse functions (see Figure 2.5):

5
I(z') = E IF.(z') (2.30)
n=1

where I. is a complex unknown number and

I1 for z' in z'1

F (z') =(2.31)
-( 0 elsewhere.

18
 I

Azi I Az AIz Az Az' \

-.05 -.03 -.01 0.3.5

Figure 2.5 "Staircase" Approximation to an Actual Current

Distribution.

19
substituting equation (2.30) into (2.26) yields

1 5 r12 i
E I, F.(z')K(z,z')dz' - E (z). (2.32)
jWEo n=l J-/2 z

Selection of weighting functions is influenced by the desire

to make computations simple. Therefore, the point-matching

technique is used. The weighting functions are

W(z) = 6(Z-Zm). (2.33)

Taking the inner products in equation (2.32) produces

1 5 ['/2 i
E n F (Z')K(Z ,Z')dz' - E (zm). (2.34)
jWE(, n=l1 -L/2 z

The subscript, m, is considered the match point index because

it is associated with the observation point at which the m" ''

equation is valid. The source point index, n, is so called

because it refers to the field from the current on the n'"

segment. For mathematical convenience, the locations of

enforcement are at the centers of each segment. (It is

important to recognize that while mathematically simple,

point matching actually relaxes the boundary conditions

i s
E + E = 0 (2.35)
z z

20
between selected points. Such simplification should be
questioned before hastily using this technique in other
antenna problems).
Equation (2.34) is solved as a matrix equation:

[Zmn][In] = [V-] (2.36)

where

1 rrlL/2
Zn = F,(z')K(z,,,z')dz' (2.37)
j OJ (.,-I)Lt/2

i
V. = -E (z,). (2.38)
z

Note the mathematical convenience of (2.36) which is

analogous (but not equivalent) to Kirchhoff's network

equations. Because of the analogy, however, [ZmJ,[In] and

IV,] are respectively referred to as generalized impedance,

current and voltage matrices.

Before solving equation (2.36), [Vn] must be defined for

the given problem. Source modeling is a subject worthy of
exploration beyond this study [Ref. 2]. Two most popular
versions are (a) the delta gap generator that corresponds to
feeding by an ideal voltage source and (b) a magnetic frill

generator that corresponds to feeding by a coax cable from a

ground plane. Here, a 1 volt center fed excitation, when
modeled with a frill generator and weighted with delta
testing functions, results in
21
 0.502 L-179.690
3.248 z--179.95 0
[V.. 70.55 z-179.99 0 (2.39)
3.248 z--179.950
.502 z--179.690

The above choice of uniform segmentation of the wire results

in an impedance matrix that is Toeplitz symmetric. (In a

Toeplitz matrix, Z., = Zx,,1 -#, for m 2, nl1.) Thus, the

first column (identical to the first row) is all that is
needed to fill Zn.

679.2 . 90.010
292.4 /--90.030
[Zmz = 10- " 33.01 /-90.270 (2.40)
9.74 L-90.01(
4.24 /L-92.080

Solving (2.36) for [I,] yields

.81 L89.54 0
1.54 z-89.640
[1,] = 10' " 2.44 z-89.750 (2.41)
1.54 z-89.640
.81 zL89,540

The solution for [I,,] is consistent with expected

results. The current distribution is generally triangular.

The input impedance is largely capacitive with a real part

less than 2 ohms. Nevertheless, a more accurate solution for
[I,,] is obtainable with a larger N. Figure 2.6 shows how a

22
 L = 0.47 X
a=0.005X
0.014
N = 60

N =50

0.007 N = 20

U
.

0.
-0.235 0.235
L/X

Figure 2.6 Current Distributions on a Half-wave Dipole for Various

Numbers of Basis Functions and a Unit Volt Source [Ref. 2: p. 3 14].

23
solution for [I,, converges by increasing N in a similar

problem [Ref. 2:p. 3141.

The purpose of this example has been to demonstrate the

MM application to a simple antenna problem. It is necessary

to qualify its obvious merit by a few key points which

generally apply to all MM problems.

* The matrix dimension N (number of expansion modes) must
be taken sufficiently large to get a convergent result
but not so large as to cause numerical instability.
* Small numerical errors due to round-off or truncation
usually have catastrophic effects if they cause apparent
singularities or ill-conditioning in the matrix.
choice of appropriate basis functions not only improves
the accuracy but reduces the matrix size needed for a
convergent solution. In the case of a short dipole, the
expected triangular current distribution suggests that
piecewise linear or triangle basis functions should give
a better representation of I(z').
* Galerkin's technique, (chosen expansion and weighting
functions are the same) is computationally convenient on
a computer but does little to affect accuracy. If
pulse-pulse Galerkin method is used on the dipole
example, convergence is achieved with a different N.

24
 III. LTSA MODEL

In order to examine the various parametric effects of

LTSA design on radiation patterns, a valid antenna model is
required. While LTSA applications are primarily based on
empirical designs, recently published works have advanced the
necessary theory to describe the LTSA behavior. Among them
is a paper by Janaswamy [Ref. 11. His model enables the
prediction of pattern behavior which agrees closely with LTSA
experimental results. It is Janaswamy's MM model,
algorithmic steps and Fortran programs which are used in this

study. Accordingly, this chapter summarizes the purpose and

process of this new model.
Analogous to the linear dipole example, the immediate
objective is to solve for the current distribution on the
surface of the antenna for a given excitation. The current

is determined by solving the reaction integral equation which

states that [Ref. 1]

Jim(J''R'-M"'H')ds =-ffvJ(J1'E'-MI'HT)dv (3.1)

where (E',HT*) is the free space field of an electric test

source Jr placed on the surface S. (J',Ml) are the impressed

sources (excitation) bounded by volume V. (JM ,M" ) represent

the unknown total surface currents on the antenna. Using the

25
concept of surface impedance, the currents JO and M" can be

related as
M. = Z.J. x n. (3.2)

The surface impedance concept is useful in simplifying an

otherwise difficult case of the dielectric supported antenna.

We recognize that M. R 0 on perfect conductors (air

dielectric). However, on dielectric supported antennas, the

presence of the thin dielectric slab is treated by means of
the surface impedance Z,. The surface impedance Z is a

function of the relative dielectric constant and the

substrate thickness in wavelengths (See Table 3.1).

According to equation (3.2), a non-zero value of Z. results

in additional electric surface currents. In either case,

equation (3.1) is solved for the single unknown JO.

The MM approach begins with selection of expansion modes

(synonymous with basis functions) used to model JO. Besides

choosing appropriate functions, this step has other

significant implications which first demand consideration.
The number of modes should be as small as possible to reduce
computation time but large enough to insure convergent

results for the radiation pattern. To that end, one helpful

simplification is to reconsider the antenna structure. Both

the number of expanded unknowns in

N
jo = I1 J,, (3.3)
n=1

26
 TABLE 3.1 SURFACE IMPEDANCES FOR TYPICAL DIELECTRIC
SUPPORTED LTSA (INDUCTIVE REACTANCE, OHMS)

Substrate Relative Dielectric Constant, e=:

Thicknes
d(Ao):

1.04 2.33 3.5 6.0 10.5

.01 11.86 11.88 11.90 11.94 12.01

.015 17.82 17.89 17.95 18.09 18.34

.021 25.02 25.21 25.39 25.78 26.51

.025 29.86 30.19 30.49 31.16 32.46

.030 35.97 36.54 37.08 38.29 40.73

.040 48.44 49.84 51.20 54.43 61.65

.050 61.33 64.22 67.13 74.55 94.28

27
and the computation time are reduced considerably when the

trapezoidal plates of Figure 3.1a are modeled with uniformly

rectangular patches as shown in Figure 3.1b. This segmen-
tation allows overlapping surface dipoles to represent each

J, as current flow along either principal direction of the

plates. Accordingly, Figure 3.1c represents the geometric
approximation for each plate where the feed gap is denoted as
2w.. (This is valid for flare angle 2a less than 400.)
If the number of segments along H and L are NI, and Nt,

respectively, the number of modes per plate is

Np = N(NH-l) + NH(N'-1). (3.4)

As the two plates comprising the antenna surface are

identical, they both have the same number of modes.

Including one mode to represent the feed dipole, the total

number of modes on the structure is

N = 2NP + 1. (3.5)

Selection of functions to represent each mode ale

illustrated in Figure 3.2 where the assumed direction of

current flow is along the z axis. For each surface dipole,

the current is represented by the Piecewise Sinusoid (PWS) in

the z direction and constant along the y direction.

Using the representations of Figures 3.1 and 3.2, a three

dimensional view of current is obtained by examining two

cross-section views. Figure 3.3a depicts Cut AA, which is

28
 2_c Height

Figure 3.1a LTSA Surface Geometry.

I
I
I
I
I I IT
I I I

- _I p I I-

J-I-- -L
I I I I I
I I I II

L., >L

Figure 3.1b Surface Dipoles Overlapping on Rectangular

Patches.

29
Figure 3.1c Equivalent LTSA Geometry for Modeling.

30
 Pulse - ----

I IS

Fiur 3.Iufc ioeRpesnainfrEc oe

I 31
 JRADt
CtAA

LL

Figure 3.3a Cross Section of Current Distribution Along

the Length Dimension of the Plate (Cut AA).

H
J RADt TRA{

Figure 3.3b Cross Section of Current Distribution Along

the Height Dimension of the Plate (Cut BB).

32
used to portray current behavior along the length dimension.
Similarly, Figure 3.3b is Cut BB, which provides a view of
current behavior along the height. Consequently, current is
analyzed using two separate plots of JTR,. and JvAD as seen in

Appendix C.
The test modes (weighting functions) for J are chosen

identical to the expansion modes for Jn, hence the method is

Galerkin. Although other choices are possible, PWS Galerkin
is best suited in the MM procedure for scattering problems in
free space and is proved to be satisfactory in this model.
Having formulated the MM strategy, equation (3.1) is

solved by substituting (3.2), (3.3) and using the familiar

form

[Zm ii[Ji,] IV.) (3.6)

where

ZM" = zo J ds -
T.J T ds
EJ (3.7)
n n

and

Vm = (J*.ET
m - T)dv. (3.8)
v

Here, Z,,, represents the mutual impedance between the mtl"

testing mode (weighting function) and the nth expansion mode

(basis function) and can be termed the inner product,

<J,,,E >. The complete discussion of equation (3.7) need not
33
be expanded here but two aspects of Z,, are significant in

this LTSA model.

* Janaswamy treats the cases of parallel and non-parallel

dipoles with different algorithms and uses closed form
expressions where appropriate. In concert with the
problem symmetry, his approach improves computational
efficiency.
Filling the impedance matrix is a time consuming task.
Table 3.2 provides typical CPU times required to fill
and perform LU decomposition on such a matrix. Owing to
aforementioned symmetry, matrices require only dimension
N', where N' = N,,+1. Therefore the matrix [Z...J has
(N')2 elements. The preponderance of LTSA modeling
problems requires N1'>130, for which a virtual memory of
1500 Kilobytes is recommended. For N' greater than 200,
2 Megabytes of memory are necessary.
As suggested in the dipole example, increasing N improves

the approximation of the current distribution. However,

practical consideration of the required computer resources
dictates a limit for N. Janaswamy has demonstrated that
convergent results are obtained for the radiation pattern
when the LTSA height and length are partitioned into four to
five segments per wavelength. This criteria was used

throughout the modeling study.

It is worth mentioning that an accurate LTSA input

impedance can be calculated only if convergence is achieved

in the solution for [I,]. Because an even larger N is
required, this task is not undertaken.

The matrix [V,,] in equation (3.3) is determined by the

type of excitation used in the model. In the present model,

a delta gap generator is used, corresponding to a

transmitting antenna. By reciprocity, the solution is also

34
TABLE 3.2 TYPICAL CPU TIME AND STORAGE REQUIREMENTS FOR THE
IMPEDANCE MATRIX OF THE LTSA MODEL (USING VS
FORTRAN AND IBM 3033)

Segments CPU Time Matrix Stor-

per Plate Dimensions of Matrix in age in Bytes
Minutes (Formatted
N,, N, N N' File)

15 2 87 44 8.0 16K

20 2 117 59 14.4 280K

25 2 147 74 22.6 440K

15 5 261 131 71.5 1.4M

23 6 495 248 245. 4.9M

35
valid for the receiving case. This choice results in Vm

equal to zero, except at the source segment where it is

unity. Due to the plate corners which would cause undue

fringing around the feed gap, the normal direct connection is

not used. Instead, the generator is superimposed on a strip

dipole centered across the feed gap, as depicted in Figure

3.4. The feed dipole 2

length and width are 2Ld and Wd

respectively. Results of this study show that the far-fields

are relatively insensitive to changes in the feed dipole

dimensions (See Figure 3.5). The apparent H-plane behavior

change shown in Figure 3.5b is attributable to the fact that

only one expansion mode is used for the source. Figures A.1

through A.4 further demonstrate more vividly that these

effects are insignificant.

In summary of the purpose and process of the LTSA model,

the MM technique provides an approximation of the current

distribution which is subsequently used to compute the far-

field radiation pattern. A point worth emphasis is the

criterion of choosing four to five segments per wavelength in

either principal direction of the plate. While this is not

enough to obtain convergent results for current distribution,

it is quite accurate for the radiation pattern. Moreover,

this model provides predicted patterns well corroborated by

empirical results. Thus, the only apparent limits on its

effectiveness are established by the computer resources

available.

36
 2Wd

Figure 3.4 Source Model of Unit Volt Excitation

on a Surface Dipole Placed across the Feed Gap.

37
 120 0

i .-.
- -. -n
100~
oLEGEND
0 0
A - 3db BeaomwLdth
C • - 1Odb BecmwLdth -20 a
S- 1st 5LdeLobe LeveL
.C in
so -30 .

C ~40 -40 C
0

20 -500a

0.1 0.2 0.3

Length of Feed OLpoLe (k)

FLgure 3.5a E-PLane Effects of OLpoLe ModeLLng.

120 0

too -10o

L
2 1
so-3

o -20-'O
60 -40 C
0

a -

A - 3db BeaomwLdth
2 0- ldb OeomwLdth -0

U- lel SLdeLobe LeveL

0.1 0.2 0.3

Length of Feed OLpoLe (

FLgure 3.5b H-PLane Effects of OLpoLe ModeLLng.

38
 IV. PREDICTED PATTERN BEHAVIORS

A. BACKGROUND

The theoretical model provides the opportunity to develop

specific parametric analysis of the LTSA. Heretofore, these
analyses have only been experimentally performed. Therefore,
the ability to accurately predict radiation pattern behavior
is a significant step in the LTSA design methodology.

Before the correlation of behavior changes to design

changes can yield useful results, a validated set of design
parameters is necessary. For reference and comparison
purposes, the following LTSA design parameters are used:
* Length, L = 3.0Ao

* Height, H = 0.9Ao

* Feed point gap, 2Wr = .02Ao

* Flare angle, 2a = 120

* Substrate dielectric constant, c, = 2.33

* Substrate thickness, d = .021Ao.

The predicted radiation pattern for the reference data is

shown in Figure 4.1.

The modeled behavior corresponds very well to
experimental results obtained with this specific design [Ref.

11. Hence, the investigation of LTSA behavior proceeds by

varying one design parameter at a time and observing the

attendant effects on the radiation patterns. This technique

39
0-

0-H -0.

0
-8~ -19 10 -9 -0 - 0-0 s -012 S 8

04
is well illustrated in Appendix A which contains radiation

pattern comparisons for the designs modeled in this study.

A more succinct analysis is obtained by de --

main beam and sidelobe properties with th- Cs

for each plane. They are:

* Half-power or 3 db beamwidths (HP7 .}-

* 10 db beamwidth,
* Level of first sidelobe relative to main bear.

Together these attributes indicate the directivity of the

antenna and provide a basis with which to compare antenna

designs. The equation used to approximate the directivity of

the LTSA is

D = 52,525/(HPzHP,) (4.1)

where HPm and HP. are in degrees and a smooth elliptical main

beam cross-section is assumed [Ref. 2:p. 396].

B. LENGTH OF THE LTSA

It has been demonstrated by earlier theory and

experiment, that for LTSA lengths (L) of three to ten

wavelengths, narrower beamwidths are obtained by increasing L

(Ref. 4]. Based on known behavior of antennas such as the

Vee dipole and horn antennas, this is a reasonable

expectation. However, results obtained with the present model

are accurate enough to offer more specific observations.

41
 Radiation patterns for L = 3 \., 4 Ao and 5Ao have been

computed for heights (H) of .25Ao to 1.5Ao. (See Figures A.5

through A.16). A synopsis of significant findings concerning
length is deduced from Figure 4.2. For a nominal height of
one wavelength, radiation pattern characteristics are plotted
as a function of length.
As L is increased the progressive decrease in beamwidth
is evident in both the E and H planes. Since the directivity
is inversely proportional to the beamwidths' product, the
improvement between L = 3, (HPn = 350, HP, = 450) and L =

5Ao (HPr = 210, HPm = 370) is approximately 3 db.

Also noteworthy is the concurrent sidelobe reduction

which is most apparent in the H-plane. Although Figure 4.2
shows only the relative gain of the first sidelobe, review of
Figures A.11 and A.12 indicates that the main beam is well-
formed and that there are no sidelobe anomalies. Suppression
of the E-plane sidelobes below -10 db is achieved with L >

4A,.
Figures B.1 through B.6 demonstrate that the observed
trends for changing L are consistent for several heights
(.25Ao - 1.5A). Thus, a further interpretation regarding
length is worth consideration. To compute an actual antenna
gain, the input reflection coefficient or the input impedance

is required. However, design for a particular directivity

can be accomplished with equation (4.1).

42
 120 0

too nn -1073
S LEGEND
A- 3db Searni~dth
CDea 0 - l0db BoarnuLdth -20--'j
S El 1stL SLdatobo LaveL -

so'i~ -30

C
aa

L2 0
20 ~-50 (

3.0 4.0 5.0

NormaLized Length (~

FLqure 4.2a E-PLone Effects of ChangLng L; H - 1.O\X

too -10 .

Sl 0
EK) -20 0

a -0

C - -

- C

o 40 . -40
Com~e Legt

Fa ue42 -~n fetso hn~qL .O,

- 3db Bamw~d3
 For example, if a directivity of 17 db is required, the

HPm'HP,, product must be less than 1048. Figure 4.2 indicates

at L = 4,o that HP, = 250 and HPH z 410. It is apparent that

a nominal antenna length of 4Ao is at least a good starting

point for this design.

C. LTSA HEIGHT
Earlier theory for the LTSA had only been developed for
large height (H : 3Ao. Experiments with an air dielectric
LTSA (L = 7.2,\o) showed that a progressively narrower E-plane
beamwidth is obtained when it was shortened to 1.5Ao, but

that at .75Ao, the HP, broadened once again [Ref. 5]. This

apparent resonant effect appears in modeling a shorter

antenna (L = 4 A.) as depicted in Figure 4.3a. When H =

1.25Ao, there is a distinct narrowing of HPE. Compared to H

= 1.5 Ao, this represents a directivity increase of about 2

db. For H decreasing below iAo, the HPr tapers to less than
200 at H = .25Ao.

In the H-plane the cited experiments showed a broadening

of HPH for decreasing height. This is consistent with the
predicted behavior shown in Figure 4.3b. However, as H is
reduced below .75Ao, HPH decreases and the 10 db beamwidth

virtually disappears. Figure A.22 shows the emergence of

large sidelobes at H = .25\o, while rigure A.23 demonstrates

that the beam is not well formed at H = .50A,. These effects

are also observed at L = 3Ao (Figure B.7) and L = 5Ao (Figure

B.9).
44
 120 0

L LEGEN

60 -

0
cc 40 -400.
aS

20 -5 em~t 1& 0d

CD
L

-C

so -30

40 -40 C-9

20 -50 -0

0 .5 1'.0 1.5
Nor-motLzed HeLght [.\,)

FLgure 4.3b EPLane Effects of ChangLng H; L - 4,

13,45
 In essence, both theory and experiment suggest that an

optimum H/Ao may exist in a range which depends on L/Ao. For

example, Figures A.20 and A.21 show that a range of optimum

height appears very narrow - about 1)o. On the other hand,

for L = 5 Ao, Figures A.30 and A.31 indicate a relatively

broad range of optimum H between 1.0/o and 1.5A.

It is necessary to qualify implications of a so-called

optimum. In this study, dimension multiples of Ao/4 are used

for L and H. Consequently, the interpolation of data points

must be judicious. It is especially significant as the

relative contributions of other fixed design parameters may

change the radiation pattern with synergistic effect.
Due to the noted pattern behavior attributed to changes

in height and length, a possible relationship to the

length/height ratio (L/H) is explored in Figure 4.4. No

discernible affirmation or refutation is immediately evident.

Consider L/H = 4 for the three modeled antenna lengths. The
characteristics of these designs are shown in Table 4.1
These comparative results generally show that a given L/H
ratio does not directly determine the radiation patterns. It

is reasonable to conclude that L/ \, and H/Ao independently

contribute to the LTSA behavior.

D. FLARE ANGLE
The effects of changing LTSA flare angle (2a) are
experienced in two competing phenomena. Consider first that

increasing the flare angle in similar traveling wave antennas

46
 120 a 0

too--0o..0

Co -20
0 A - 3db BeauiwLdth
X- 10db SeamwLdth
I~U let StdeLobe LeveL 4
so-30 .

CD
E0 C
0
4D 40 -40

20 -50

Length to HeLght Rat.Lo

FLgure 4.4a E-PLone Effects of ChanqLng the L/H RatLo.

120 0

-10 .
0 a

00 -20
A-3db BeoswLdth -2

* 1db BoomwLdth
U sltSLdeLobe LeveL
so -30..
a
-4 4m
a c
a >
20 -500

NOTE: I0db BW not cLwags observed.

6.0 11.0 16.0

Length to HeLght RatLo

FLgure 4.4b H-PLone Effects of ChorngLng the L/H RotLo.

47
 TABLE 4.1 PATTERN BEHAVIOR FOR DESIGNS WITH L/H =4.

L H HPm HP,, 10 dbv 10 db,~ 1st Sidelobe Directivity

A\0 Ao BEAMWIDTH BEAMWIDTH E(db) H(db) (db)

3 .75 30 0 620 500 1000 -8 -12 11.5

4 1.0 250 410 450 610 -9 -19 14.0

5 1.25 180 310 360 470 -17 -20 16.7

48
has demonstrated that HPr is proportional to 1/2a and that HPH is
generally insensitive to flare angle changes [Ref. 6]. These

results are justified in part by analogy to aperture antennas in

which narrower beamwidths are obtained by increasing the aperture
size. The second effect for increasing 2a is a broader beamwidth

which results from an increased phase velocity. This is

explained by noting that the fields between the plates are

weakened as the plates move farther apart. The phase velocity

relates inversely to the energy stored in the fields. Therefore,

as the flare angle increases, the slow-wave effect supporting a
narrow beam is reduced and the beam widens.
The net effects of these two dynamic factors are summarized
in Figure 4.5. It is evident that below 210 the beamwidths are
insensitive to flare angle. This contrasts with at least two

separate experiments for the fin line LTSA which showed HPm

unaffected by a, but also proved that HPm is proportional to 1/2a

[Refs. 6 and 7). Another departure from empirical studies is the

absence of beam symmetry in the predicted patterns. It should be

noted that in experiments, the beam symmetry appeared in narrow

ranges of flare angle and only for longer antenna lengths. In

general, the relative contributions of the cited competing

factors seem to offset one another in modeling below 210.

Above 210, HPr increases, possibly due to the phase velocity

effect dominating the illumination mechanism. In the H-plane

there is a concurrent decrease in HP". This may be attributed

49
 -10~

-0

CD

0 40 -40 C

IL

200

0 li .0 2 .0-270 .

12 0

1 CI -

a) -L -20

L 0
00 20w
CM

CDC
400
-4
A-dbLEGEND>
a bBeamwLdth
20 0 - 10db BsaftwLdth -50oC
R- Iet S~.daLobe LaveL

015.0 21.0 27.0

Rlare Flrng~e, 2o [Degrees)

FLgure 4.5b H-PLane Effects of' Chonq~ng FLare flnqLe, 2ck.

50C
to larger current components which move perpendicular to the H-

plane as the flare angle increases. Figures C.1 and C.23-C.30

show the modeled current distributions for respective flare
angles. Using Janaswamy's model, surface impedance does not
change and the current distribution appears to change only
slightly as a function of flare angle. However, to resolve this
question in further detail requires more accurate modeling of the
current distribution.
As a matter for design consideration, the radiation patterns
for different flare angles are plotted in Figures A.32-A.39.

Although these designs have approximately equal directivity, the

high sidelobes are not necessarily acceptable. For this reason
(as Figure 4.5 also indicates) a range of flare angles between

190 and 220 appears to be best, given the same set of other

design parameters.

E. SUBSTRATE FACTORS

As discussed previously, the plates are treated as imperfect

conductors in the dielectric supported antenna. The surface
impedance is a function of dielectric thickness (d) and the

relative permittivity (E). A modest range of dielectric designs

were modeled and the respective predicted radiation patterns are

presented in Figures A.40-A.48.

It is first noted that increasing "d" noticeably reduces the

large E-plane backlobe and slightly increases the relative gain

of the H-plane backlobe. This is a point of interest because

51
backlobes are a minor disagreement between the LTSA experimental

and theoretical models. The predicted results suggest that the

discrepancy is due to the manner in which Janaswamy's model

approximates the effects of the dielectric substrate (Ref. 1].

The LTSA behavior as a function of substrate thickness is

characterized in Figure 4.6. It is observed that the first

sidelobes increase slightly with d. The E-plane first sidelobe

is relatively large.

The primary effect of increasing d is the HP, improvement.

Between d = .01Ao and .05Ao, HPH decreases by more tha' 30%. The

E-plane beam is equally well-formed but it appears unaffected

except where it suddenly narrows by 30% between d = .04Ao and

.05N,. Approximately 2.2 db improvement in directivity is

achieved by increasing d from .021N. to .05Ao.
The noted behavior is supported by theory which allows that a
dielectric material typically slows the currents in an endfire
producing surface wave antenna. As the phase velocity decreases
a corresponding narrower beam width is produced. The concurrent
increase in current is noted by comparisons among Figure C.l and
C.31-C.36. Coincident with the cited HPn decrease, Figures C.35c
and C.36c show that the transmission mode current at the outer
edge along H does not taper off as noted in most other modeled
designs. Because the calculated current distribution is not
accurate, it cannot be ascertained if this is a meaningful
observation or a numerical aberration.

52
 220 0

e LEGEND
Lo
A - 3db FeamwLdth
m s - 10db BeomwLdth 2
19U-let SLdoLcbe LeveL
.5A

~ 40-40 U

i 0
-0 C

0
0.01 0. C 0.03 0.40.05
Substrate Th~cknessC>

FLgure 4.6a E-PLone Effects of ChongLn 9 Substrote Th~ckness.

120 0

IDO -10

L
m 00C -2
-Im

so -30

40 -

CLC

CL A - 3db LGErnDh
2o0 - 10db BoamwLdth -50
M - 1ot SLdaLobs LeveL

0.01 0.02 0.03 0.04 0.05

Substrate Thickness C)X
13

FLgure 4.6b HF'Lone Effects of Chan 9 Lng Substrate Th~ckness.

53
 The other substrate parameter to consider is the relative
dielectric constant (S.). Figure 4.7 depicts a proven
insensitivity of pattern behavior to changes in e, between 2.33

and 10.5. (For detailed corroboration, see Figures A.46-A.48 and

C.37-C.39.) Simply stated, this occurs because the surface

impedance only depends on d when the substrate is electrically

thin, i.e., when (e,-l)d/ Ao S .2. In other words, for small d,

the surface impedance equation simplifies as

170
, 110
Z - tan(kodlsF) - kod / = %kd (4.2)

This finding is commercially relevant because the cost of

substrate materials is a factor in building these relatively

inexpensive antennas.

F. SUMMARY OF RADIATION PATTERN EFFECTS

The general purpose of modeling the LTSA is to gain insight

and apply the necessary theory to understand its behavior.
Equally significant, however, is the specific purpose, which is
to provide useful guidance for the design and application of this

antenna. Based on the credibility of the theoretical model,

highlights of the parametric analysis deserve reiteration. Data

already exists for studies of LTSA's with large lengths. Modeled

herein are design lengths of 3Ao, 4Ao and 5Ao. The results show

that directivity is directly and consistently proportional to the

antenna length. The design process may commence by selecting an

appropriate L for a desired directive gain.
54
 120 a

- ---
-- - - U-m----------- ---- --
1W -10.
3 LEGEND
•
L
A - 3db BeomwLdth
S0 - 10db BeomwLdth -2
E? - lot SLdeLobe LeveL

so -30
E

40 -40

2.0 4.0 6.0 8.0 10.0

Substrate ReLatLve OLeLectrLc Constant

FLgure 4.7a E-PLone Effects of ChangLng Substrate PermLttLvLt .

L
01 0
80-- -200

so -30

0 10..j

A - 3db BeomwLdth
2 - lOdb B0orwLdth
C- let 5LdeLobe LeveL

i.0 4.0 i.o ei.0 -. 0

Substrate RstotLve OLstsctrLc Constant
)1
FLqure 4.7b H-PLone Effects of ChongLn 9 Substrate PermLtt~vLt .

55
 The applied model uniquely provides an aid to investigate,
the height factor. A key finding is that the empirically
observed optimum range of H/Ao is also evident in the predicted
pattern behavior. Results further show that while the range of
an optimum H/A0 does depend on antenna length, the pattern itself

depends on L/Ao and H/A. independently.

For the modeled designs of flare angle between 60 and 300,
two mechanisms appear to cancel the effects of one another.
Consequently, the directivity is virtually unaffected by the
flare angle. Given other design parameters the same, flare
angles chosen between 190 and 220 give the best sidelobe

suppression.
Substrate thickness is the dominant contribution to surface
impedance. For electrically thin substrate, s, has no effect.
In the modeled range of d, the H-plane beam narrows about 50 per

.01 Ao. In the otherwise unaffected E-plane, the backlobe is

suppressed below -10 db for d .03Ao and the main beam suddenly
narrows between d = .04Ao and .05 A0. Although beam sharpening

can be done with thicker substrate, it may be relatively costly

and impractical.

56
 V. CONCLUSIONS OF MODELING STUDY

The scope of any analysis includes a purpose, supporting

theory and a demonstrative application or proof. For the

LTSA, the purpose is to develop a model which accurately

predicts its behavior. Since experiments and applications
began in 1982, such a model has become an important goal in
the antenna community. The supporting theory advanced here
is to use the MM to solve the reaction integral equation.
Thereafter, the far fields are computed once the current

distribution is known. The primary limitation to exploiting

this model is the availability of CPU time required for

rigorous MM solutions. Using this model, the contribution of

this study is a parametric analysis of 39 LTSA designs. The
resulting modeled behavior generally agrees with known
empirical data. Consequently, the analysis not only is
consistent with applied theory but also is useful for the

LrSA design process.

The presentation of ideas for review begins with the MM
technique. As detailed in Chapter 2, MM makes it possible

and practical to solve for an unknown function, f(x), in an

equation which does not have a closed form solution. The
original equation is essentially transformed to an equivalent

matrix equation which is then solved by conventional methods.

There are two major considerations in MM. Foremost is the
choice of basis functions, which ideally are selected to

57
faithfully represent the true f(x). However, since the form

of f(x) is often unknown, selecting basis functions in

electromagnetic applications is usually a result of trial and

error. This alludes to the second key aspect of MM. CPU

time may be scarce or expensive, the two factors affecting

the economy of those resources are rate of convergence and
desired precision in the solution. A good (or fortuitous)
choice of basis functions improves the rate of convergence.
The matrix size (hence CPU cost) grows geometrically with the
extent of required accuracy.

The present model employs the popular Piecewise

Sinusoidal-Galerkin technique. The main reason for this

choice is that it helps to reduce the CPU time involved in

computing the elements of the impedance matrix. Furthermore,

it has the potential for reducing the number of unknowns for

a given problem. A convergent solution for the radiation

pattern (far-field) is obtained with four to five segments
per wavelength. An accurate solution for current
distribution (near-field) entails much more CPU time because

more segments are required for convergence. The model is

rigorous for the air dielectric case and approximate if the
LTSA has an electrically thin dielectric substrate. To

precisely include the dielectric in the algorithm contributes

little more to the solution than increased CPU time.

In summation of the parametric analysis, there are

several conclusions which are valid for the ranges of modeled

5F
design parameters. Moreover, it is reasonable to extrapolate
behavior based on these significant modeling results:
* An appropriate antenna length is best suggested by the
initial determination of required directivity.
* For a given length there is a range of optimum height
which will contribute to a well formed narrow beam.
* Although the flare angle has no effect on directivity in
a given design, there is a flare angle bracket within
which side-lobes are ameliorated.
* Substrate thickness has a significant and inversely
proportional relationship to H-plane beamwidth.
* For electrically thin substrate, the relative dielectric
constant has no effect on the radiation pattern.
Future design and development of the LTSA will be
enhanced by this research. Having successfully modeled the
pattern behavior, the next step is to obtain a convergent
solution for the LTSA current. Such a goal will probably
await lower cost microprocessing when improved accuracy of MM
techniques is more cost effective. In the interim, it
appears that the concept, the ideas and the algorithm used in
this study may be equally effective in the analysis of
similar antennas.

59
 APPENDIX A LTSA RADIATION PATTERNS

The parametric analysis of radiation patterns begins with

the known empirical design which was accurately modeled

(Figure A.1) using the following specifications:

* Feed Gap = .02Ao

* Length = 3Ao

* Height = .9\o

* Flare Angle = 120

* Substrate Thickness = .021Ao

* Substrate Relative Dielectric Constant = 2.33

The study of these parameters and their effects is based on

Figures A.2-A.48 which show the computed radiation pattern

for each case. Each computed pattern is templated with the

reference pattern to illustrate the immediate effects of the

design change.
Since the theoretical model uses a strip dipole source,

the feed dipole length is varied as a design parameter but

the feed gap is fixed at .021A. The comparative effects of

this design parameter (Figures A.2-A.4) serve to help

understand the source model rather than the antenna itself.

Figures A.5-A.16 show calculated patterns for three

different antenna lengths of 3A,, 4A. and 5A.. In each case,

the height of the antenna was changed in the range .25Ao-

1.5Ao. These same designs are given a different perspective

60
in Figures A.17-A.31 which show the effects of changing

height for given antenna lengths.

The range of flare angles which were modeled is 60 to

300. Each modeled design is compared with the original

design (120) in Figures A.32-A.39.

The surface impedance is determined by both the substrate

thickness and permittivity. Varying these parameters results

in radiation patterns shown in Figures A.40-A.48.

61
 RADIATION PATTERNS
~H 0.9x
L 3X.

25H-PLone E-PLano

-1e0 -150 -120 -90 -60 -3 0 33 so 90 120 150 150

AngLe Off Bor'esLgh. (Degrees)

FLgure A.1 PredLcted Patterns for a Known EmpLrcoaL OesLgn.

0 RHI~ORTION PRIIERNS
OLpoLe Lsngth - .0X

L L
CL. Is5
v'

-2"

lea10 -2 -90 - - a so -0 10 -o IO

FLgure F1.2 Effects of DecreasLng Feed OLpoLe Length to .05X.

62
 OLpoLe Length -.20X.

-10 -15 -12 - -0 a so- -3 20 -0 IO

o -

I'

H-PLane E-PLane

180 -W -120 -g -80a -30 a 30 so gO 120 150 ISO

fingLe Off BoreeLght (Degrees)

FLgure IA3 Effects of Irncres.ng Feed OLpoLe Lengh to .30X,.

RA~iTIONPATTRN63
 70

ISO-15-2 -0 - a 0 50 s 1U IS S
*n~ Of Boa~h Dgea

Efecs
F~greA. f ecrasngLenthto3X fo H.2X,

-L - - -,; fl5a.
L 4 H_-.25x I

a--1 -15 -12 - - - a-0 s o 12 S S

*ig~ Of Baa~h Dges

F~~~~~qure~~~~~
fetIfIcesn ~ A. Lnt oSfrH.5,

64I I
 0

it

E-PIan

-1 -150 -120 -90 -.60 -33 0 a 0 s BsC 2 IO N

FingLe Off Sor-eatght (Oegrese)

flur
.7 Effects of DecreasLng Length to 3X, for H-.50O>.

L -5,; H- . 50 ,

_.A I I

-10

L AI

It ti Ii Pan EP~

-'- 0 tm-120 3 a 3 o g 10 Itt 10 S

Itg~ Of Boe*h fDg-9

RgIr 1. EfettfIcesn ent o5,frH.0,

~ HP~os ii -P~65
 L 3, .75 - -

10

IV II
0- is

-180 -150 -120 -90 sGo -30 0 30 5 so 120 150 180

AngLe Off Bor'eaight (Degrees)

F~qure R.9 Effects of Decr-easLng Length to 3X. for H-.75X,

L -5),. H - .75 .

.0

tL -153~

I Aj

-5H-PLane E-PLone

-180 -5 -120 -DC -60 -32 0 32 so go 120 150 152

AngLe OffC BoreaLght (Degrees)

FLgure A9.10 Effects of IncroosLng Length to 5X, for H-. 75Xy

66
 C RADIRHON PHTIERNS

.0
a- is ---

11 u

H-*as 1 I -~as

IN. -Is- 2 - 0 I -3030 w s 2 5 o

LIg~ Of Brs'h (eres

F Iur A.1 EfetIf0cesn ent o3,frHIO

RAIA IOPfi I
L SxI -H IIO

a- -1 i ia -o -o - ~ a ~ s o io 10 1

L10-4 H-.O -12 0 s o -9 10 -6 IO -3 N

- Rn- Of-ae~h Dges

F~~~~~gure~~~~~~
Efcso ~ A.2 o5,frHIO,
Ira~gLnt

L 67
 L - 3.\; H - 1.25),,
I; H-- 125 . .

-_
0 -

J ",, 1,,
.j I J
0
~20 ---- -

H-PLane E-Lne

-100 -150 -120 -30 -60 -30 a 30 50 90 120 150 1 @0

AngLe Off BoresLght (Degrees)

FLgure R.13 Effects of OecreosLng Length to 3X, for H-1.25x O .

PIRD TI ON P AT teRN
/
L - x;
H - 1.25x.
L-4;H-125, I
-5 . . ...
I I f

.0- o
---15 I I . ~I \ 'I
I Il"

0- o.Il'- o
-~n
~n
0.-25 --
I I I III

-10 -350 -120-9O

-0 -30-O a 30
- 15 O

RngLe Off BoreeLght (Degrees)

FLgure R.14 Effects of IncreasLng Length to 5x, for H-1.25X..

68
 3.;
x H 11.50 ,I
L I, H- 1.5Ox
-f F

0 - ' -

0- -- '35

-2S

-IBO -Im2 -120 -90 -60 -30 0 m so 90 120 150 152

RingLe Off BoreaLgh. (Degrees)

FLgure 19.15 Effects of Oecreast~ng Length to 3X,~ for Hi1. 5Ox.

-RifiDIRTION PATlRN
L - Sx;H - 1.50

0 -15 - . - ___A", - -

-20

-2 on ,-P 2~n

v-10 I -12 oA-- -M s -6 0 -3 10 10 -0

An2 Off Boe-h (Dgres

Rg. r A.1 Efet ofIcesn eghtIX o -. 0,

*6
 H - 0.25),; L -3x
H_- j.OX,; L_-3X._

IFI

-~n
-2 H-In

0 HIRT 5i~ IAT

o 5 H

-1 0 II

o"

WP~ane PLane

L I 1

-160 -1W -120 -60 -60 -0 Q 20 s0 so 130 150 16

AngLe Off BareaLght (Degr-ees)

FLgure 19.18 Effects Of OecreosLng HeLght to . 50X, for L-3 0 .

o
70
 0 A

L
vi

* ~ II

ILLI

It iE-Pae
HP~ae

-180 _ 10 -90 -80 -30 0 30 s0 so 120 15 10

fingLe Off BoreeLght (Degrees]

F~gure R.19 Effects of DecreosLng HeLght to .75N. for L-3X,.

L - 3x,
H .25X ;,
H_- 1._!O ,,; L_ 3x

a fL

H-Pan E-P I'

-1010 -S -0 6 3 0 s go 120 ,IS

~
fing Bae~h ~ ~ ~
Dges
Of a

F~guGffets
820 f Icres~n Haghtto .25, fr L3a'

a 71
 H - 1.50O, ; L -

L %

H-Iae ii E-Pa
-25

-010--10 0 -6 - - o g 10 IO

Ilg~ Of SoeIh Dges

o~ur A.2 Efet fIcesigH~Itt .0k o ),

- H IF O N'a_
- - -

~
PL neEP~n

IIt

20

-255

L I

o 18 _35 _12 -0 _ I o o 10 IO
-r--~h fl..~ Off (De-g--ee&-)

*~ur 8.2 Efet ofDceen egt o.5,frI,

I 2
 0 RPFWAIION PRIE NS- - - -

H - 0.50),,; L - I,
H_- I.OX L X

fh,

CL -105
ItI

I H-in I -

HH0.75x,; L 4),,
-5 - - - -- ~

11 II It

-180-150 -120 -90 -60 -0 a 30 S0 so 120 ISO 190

RngLa Off Bor'eaLght (Degree@)

FLgure R.24 Effects of UecreasLng HeLght to .75x, for L-4X 0 .

0 AAIRIIN
PAIE73

..
. ... H.. S~
 RAOIRTION PRIIEiRNS
H -1.25x ; L - 4
H- I.POX.; L- - X_

0 11

I I -n
o 2 11

-J 120 -i - 1 20 -9 - - - m 0 g_ 2_ IO I
AnU Off BoI~h (Dges

FIur A.25 Effct ofIcesn Igtt .5,frL4

RRIIN P N5~
II

-2
1. P0x
a;4xPo
H -H RX.
- -. L--
- -i- --
-

At ~ I

-10-10 -31) -90 -50 -30 30 50

s 0g 120 150 15]
flngLe Off Bor-eaLght (Degrees)

FLgui-e A.26 Effects of Increos~ng HeLght to 1.50'A for L-4>,,.

0 74 - - -
 RHU)IAhIDN PATIERHS- - -

H - 0.25),,; L - 5
H_- i._OX; L_- 5x0_

.0

I I~ H-IIeEP~
-2SV

-18 -i -10 -0 -0 _ D o s S O

§X,- - -_

0l i \, 11 I - -
IN 111

-1-150 -12 -6O0 _ -31 0 a I0 so g0 12150

AngLe Off BoreaLght (Degree]10 15

FLgure A.27 Effects of Oecr'eosLng HeLght to .50X, for L-5% 0 .

H S7.1
LO.5~,
 H 0.75),A;L5 j Ix
-5 -1.~L5 -

L
I e.
0
CL. -35 - - - - A
0 1 [A I

Ii WPKane IEne
-25---

-180t -190 -120 -90 -60u -M a 30 so so 120 150 180

flngLe Off Bor-saLgh. (Degrees)

FLgure R.29 Effects of OecreasLng HeLght to .75X. for L-5),,

RACJJRTIONPATIERNS I k

20

H-P~ono -PLons

-I80-9M -120 -10 -60 -3D 03 0 5o so 120 150 190

fingLe OffC BoreeLght~ (Degrees)

FLgure 19.30 Effects of IncreosLng IleLght to 1. 25X, for L-5X,.

76
 H- .OfJL
SOX;, I

CL-15 -

d) I
>

_j J
-2 I ~ Ai t- l, 11,'
1
i i It
HP-Porna
- r '
-25 -

-ISO -150 -120 -90 -60 -30 0 30 so so 120 150 18

AngLe Off' Bar-eLgh. (Degrees)

FLgure F.31 Effects of IncreosLng HeLght to 1. 50X, for L-!5X,.

_RR0fl91lON PRIIER~
Fre anL - 4 I - - -

.0qr _pEL?-

CL -1'

0 1

IV -IS
-2 - --

-25

-180 -5 -120 -go -60 -30 0 30 50 so 120 ISO 10

AngLe Off Bar-eeLghi. (Degrees)

FLgure 19.32 Effects of DecreasLng PLare Fln 9e to 60.

77
 -5 ar - - - -*

CL -1s --

-j -20-
Man
- -H-PLane PLane

-1S0 -15u -120 -90 -560 -30 a 30 so so 120 ISO 180

AngLo Off Bar'esLght (Degrees)

FLgure A.33 Effects of DacreasLng FLore flngLe to 90 .

FLors ong~
La 1
Rar 2'

L I

a. is

0Z 1. 0 5 -120 9 -0 - - 0 s o 12 S S
Flg* Off Boo~ Dges

FIur A.4Efcso n~
Iri~qRr o10
 RAUIA11ON MITEhRNS- - - - -

Racre anqLa - 180

-s ar pnq_- - 1_ - - -

All

I0
L

a---
~I

FigeOfBorsaLght (Degres)

F'Lgure A.35 Effects of Incr-easLng FLore AngLe to 18'.

aRHLIATIOWPAITTERN
PLare angLe - 210
FLare onaLe- -12'

1
L j

* 1-'- -IS
go 3 -12 1 g 2 -t 5
an~ Of b.ragt Dges

o~ ur A.6Efcso Incesn RaeAIi o20

a7
 iROIfilION PATTERN S-

Cl-1 is --

0 N

20 v

-2S

-160 -1l3 -120 -90 -60 -30 a 3 60 90 120 150 1Wo

AngLe Off' Bov-aaLght (Degrees)J

FLgure R.37 Effects of IncreosLng FLare AngLe to 240.

01RRDIFIIION PATTERNS- - - - -

.0 1; V1j

~-10- - - -0 - - o 9 2 5 8
I
Rn Off BoeIh Dges

FI ur F13 Effct ofIceen ar n~ o20

a I ~80
II I
 Fo r an- -

FLare anqLe--- 12'

160o
- -5 -12 -9 3 D s -6 o 10 IO I

o 0 0
. -15 * -- - -- -

S -12 -1 I- 3 0 s o - Ia -0 1D -6

AngLe Off BoreeLght (Degrees)

Lure . Effects of ncres~..g Sustare ngLckes to 0 100

 ThLeknss (d) .015)k
ThLcknese (d) -. 021X Q_

CL -1

.0
-- 20 +- -

25H-PLane E-PLane

-180 -150 -120 -M -60 -30 a 30 s0 90 120 150 150

AngLe Off fforeeLght (Degr-eeal

Figure Al.41 Effects of Decreas~ng Substrate ThLckness to .015X,

.0c nse ( ) .0 1 q

.0

-25 H-Plane

-180 -5 -120 -W0 .6 -3D a 3D so 90 120 150 150

AngI, Of f Bor-.uight (Degres)

FLgure Fi.42 Effects of OecreosLng Substrate ThLckness to .025X,.

82
 ThLckr as s (d)
IhLcknessC)- .00-
O1~ ,,

C
0.. -IS-

H-PLane - - 2 Ine

-150 -1m -120 -90 -6 30 0 3 60 90 120 ISOI0

flngLe Of oresLght (Degrees)

F~..ure A.43 Effects of DecreasLng Substrate ThLckness to .030X,.

ThLckness (d) - . 0 3k
ThLcknee (d) - .02110

.0

L I2I

I. N -25-2 - G - - D G 2 -0 S S
2n~ Of2osLh Dges

FIguI A.4 Efet fDcesn ustaeT2 kest 0O

0 2 :2
 ThLckns (d) - .050),
ThLckneee (d)] 02 Q

0 t

IL HI

H-PLane -an

-100 -1m -120 -90 -60 -30 a m 60 g0 1213 150 ISO

FingLe Otff BoresLght (Degrees)

FLqgure HAS Effects of Oecreoe~.ng Substrate Th.,ckness to,.OSO>X,.

RH0TO - - -RN

Cr of' Substroie -3.50

L
m

on E-PLone
25H-PL

-go0-80 -120 -90 -60 -30 a 32 so so 120 ISO lea

AngLe Of f DoreLght (Degrees)

FLgure R1.46 Effects of Increosi~ng Substrate PermLtt~vLt : cr-3 .S.

84
 RAIION PATTERNS]- - - - - -

c- of Substrate - B.00
- eraf Substrote- 2.33 - - - -

-2-- - A.. I

*2
.v
-15
&/ - - - - -
v
H-P

O-25 one-

-180 -350 -120 -90 -60 -30 0 30 so so 120 ISO ISO

AngLe Off BoreaLght (Degrees]
02 . . - -.. - -

FL"gue R.47 Effects of Incr:, sLng Substrate PermLtLvLt : crB.0.

RROIHRI Wr'['HI

Cr of Substrate 10.5
2 3
_
-Cof Substrote - . 4 -

a
L

.-0 J -

H-PLone E-PLano

-110 -150 -120 -90 -G 3 0 6 so 10 IO m

fingLo Of'f BoreeLght (Degrees)

FLgure R.48 Effects of IncreosLng Substrate PermLttLvLtS: er-0.5 .

85
 APPENDIX B LTSA RADIATION CHARACTERISTICS AS FUNCTIONS
OF THE LENGTH AND HEIGHT

The summary of pattern behavior is made concise, but

useful, by examining the beamwidth and sidelobe
characteristics. Figures B.1-B.9 present this information
(in both the E-Plane and H-Plane) versus the antenna length
and height. These figures demonstrate the results of
changing length for different modeled heights (Figures B.l-
B.6). Similarly, Figures B.l-B.9 complement the analysis by
illustrating the effects of changing height for each modeled
length.

86
 LEEN 0

0-1 -2D

0 - 6bBem~t

00 -M

0 -

0-

C 0t

3.0 4.0 i.0

Nor'maLLzed Length (~k,J

FLgure B.lo E-PLane Effects of ChonqLng L; H - .25X..

233 0

13 - 10 n
ID0

0 -
A - 3db BoamLdth -2D-
S 1db BeomwLdth
lt,
13 SLdsLobe LavaL____________________-3

-40O(

NOTE: 10db BW not aLwa~s observed.

NormaLtzed Length (>.*I

FLgure B.lb H-PLone Effects of ChanqLng L; H - O.25XO.

 0m 0

LEGEND 0
O -
D a - 3db Beomwidth -

0- 10db Beomwdth En
0 lent
Sideiobe LevqL
so-30 .

~40 -40

u 0

2a -W S

a .0 4.0 5.0
Normalized Len~gth, (*J

FLgurs B.2o E-PLane Effects of ChangLnq L; H ~

0.50X,.

120 0

1W -20.

0 .

so -20-

o 0 -C

CLC
I A- dbBomJt
10(j Bsow~da
19L ~de~be LveLCK
oO~ ~bS o awy bevd

U Sd
- dengbh
let. Lee)

FLaure 8.2b H-PLone Effects of ChanqLn9 L; H - O.5OXO.

88
 B---m

L, LEG~END 2
00 - 1db BeomwLdt.h
mU let, SLdeLobe LeveL -

aa
0>
a--
20 . -so--

3 .0

NormaLized Length (,

FLgure B.3a E-PLane Effects of ChongLng L; H - .5O

Im ' -10 -

L 0a
m0 -2

.J so -30w

0 -C

o LEGE 0 40
I A - ?db Soomwidth
2o 0- 10db BsovnwLdth-s*
* let, 5tdeLoba LeveL

NorrnaLLzed Length (X,

FLqure e.3b Il-PLana Effects of ChonqLnq L; H - .75XO.

89
 120 0

* A - 3db BeamwLdth
go0 lOd BDSmwfthWOf -20 -e

m le1t SLdeLobe Level,

in

60 -30.
E
0
00
C* 40 -40 C
a M
0>

0
3.0 4.0 5.0
NormaLLzed Length (X,)

FLgure B.4a E-PLone Effects of ChangLng L; H - .OOXO.

120 0

so -30

0, _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
0 -0-
CO-
-0

CD

w -EG0ND

* 0 i.0 0.
Normo~zedLengt
IN,

F~que B4b
Efect HP~oe ofChon3Ln L; 1.OXO

a.. 90
 120 0

-10.a

-2

.C.
SO -20-
A --d Bawut

0 - 3db BeomwLdth-0
a - let SLdeaoba LeveLu
C 40 -40
a >

Li

30 450 5.0

NormoL~zed Length *

FLgure B.5a EfPLone Effects of ChangLrng L; H - .25XO..

M 0
m IK)-20-

0
6 0 -40

m~-

a - 3db BeamwLdth W
0
0 -10db Beomnwidt,
a* let S~dsLobe LeveL
03.0 0 i5.0
NormaLLzed Length (X,

FLqure B.5b H-PLai Effects of ChangLnq L; H -1.25),0.

91
 120 0

too -10.

La
83 -20

so -30 .

00

r 40 -40
0

2DA - 3db Beomwtdth w

9 - 10db Beamw.Ldt~h
m-tt SLdoLobe LeveL
al
i.
Nor-maLLzed Length

FLgure B.6o E-PLone Effects of ChongLnq L; H I.50X,

1

18 ~I .----------........----
1-

U) 0
00-20 -

-30

83 -30 CS

0 -- e.Saeoe ee

*omL~e Legt

Rqur B.b
Efect HP~aeofChaq~nqL; 1.0XO

a92
 0-0

oo

sa - 3db Seorn~dt~h -a
-rbBefw~t in
M- let, SLdeLobe Level.-

CC

C 40 . 4
a

0 c

0.5 .0 1.
NormnaLzed He~ght IX,

FLgure 8.7a E-PLone Effects of ChcngLrig H; L -XO

120

L- 0
so 1.
20 ..

so -

C -4

0 o- S~bBom~t

0. 1.00.
LomL~e - eh IX,

F0ur ffcsofCag~gH L-3O -20-'~n

S93
 1200

00
L A- db BeamwLdth -W

0
40 -4 10bJemtdV

20

12- 0

-0

..j s 3

00

0.5 1.0 1.5

NormaLized Height ),,)

FLgure B.8b H-PLane Effects of ChangLng H; L - 4O

12094
 120 0

100SC-10.D

a) 0
a 80 -20 -
LEGEND -D0
A - 3db BgamwLdt, (n
10db
6J800- Beamwidth -o.
£U- lot SLdeLobe LeveL

CD
a>

20

0.5 .01.5
NormaLLzed HeLgh. (X,)

FLgure B.9o E-PLone Effects of Chong~ng H; L -5NO.

LEGEN 0

00 2

D - A - 3db BSemwLdth -0-

S 1db BeaiuvLdth
MaU let, SLdeLobe LevaL
.j

0 C

aa
a-
00
r

20

NOTEI l
1db BW not aLwagjs observed.

Nor-maLLzed HeLghtC*

FLgure B.9b H-PLone Effects of ChangLng H; L - 5O

95
 APPENDIX C LTSA CURRENT DISTRIBUTION

A key facet of this unique planar antenna is the slow

wave structure which is achieved by the use of the dielectric

substrate. However, the travelling wave nature of the plate
currents is responsible for obtaining an end-fire main beam.
This effect is illustrated in the modeled current
distribution for each design used in this thesis.
The current distribution plots in Figures C.l-C.39
provide a means to correlate observed pattern changes to the
respective changes in current on the antenna. These
computations are based on fair approximations sufficient to
accurately predict the far-field pattern.

96
 ('6BO) SSDou4d ('60O} e9 4d

1. 3 0

o ,
.,,.4 <o C o C"
-
• I - --

0I .0 0

ML C
L
, - , - .,
0 - - -

o C - L

0 0 c

SI C-,

JIC I cc~ LZ 0

.0
3"
0
C ~CL
~c
-n n0
0

CC L
t 0

r- ---- C "
0C - - - - -- - - - 0)
C
c c
0o , 0 0
- 2-I14, - ' 2
- c
- - 4
A0 "' II A 0 'U

- 0 0 LO.
-- - - --- - -
.N .- N
..
• .
.-
4 2 ,, -*j J
,

- )I4
LL . .J 0
- 00 c-
0" I 0 C
(E fC-
-- -:2, I 2
L -
- C,0 "; 2
C.)

> -'
- < : c I u

(oi p In) I.-&, o e n ")u o I

0 a

97r
ii-VUow
uowj(F) to-]) n pniiu6ow

97
 (-Bea) eOLq d (-sea) asoyd
2 a

-, F', - ,I

-- - -

A 0 A 0

--
a. 0. I -, 0 C

0 • . E .0 -C
L L 0
o 0 -
.. .. ".0.0 I . C
L) ..
.N ' N b

C;~~ C;C;2
c

8 aDrr 0 .0 0 L)
'. LC
L L
• _ (r

--- -- - - . . . -. .... - ..I

-~I 0- CD
M --- I---- - - -

a 1 a

__0
0,,

9 aa I3o U
(-oo Gq(-r e ) sq
9c
c
a: 02C
0:0
L O
(OW) (r) epni'!uB°, (OW) (rI *Pnli'!uEOw

U tj

2C

0j .0 0 C; C; .0

(owOpSua (r o) r p~u

- L I.98
 (-sea) esoLid (-Boa) 0O8Od
C
r---r 0

- I
II

I 0 -( 0
00 . 0 .

K. 0 .

" .. ' . 4 I. 3 4
I_0 oZ0 L I 0 DL

LL
-- I - -. . - . ---. - - - ---- N L
Cr0 I 0 L
C:: c
-o 1-
C.
C3 CD

aa)

2 0 C0
(ow) (r) Ppn' !UBoW lOw) (r) aPn'%'!u~oW

('BeO) esm48 rsBeaj eeoqi o

- - -- LM.. c
U
i 0 I L

/ 00 CD
0 ~0
S. c .

o 9I Zr

-*(0 " .-- 0 ,-

. 0

C; CD
. -- N I "
. . . .. ""' 0 0N C
.. . . - C.
, 00 C.
- 7 "- 4
.0 o a4.4-

LO L

L L

-
--- - - -

ent o Uo (w llp nuo c; c

I
G! 0

(Dwi (r) epn~lu~oW (ow) (r) opnil-Bow

99
 (.SaO) eGS-J (-OB ) eqDj

UU

- -- - - - - a 0

a C* 0 _0 C )
c ' ,
o c' 0 -c

.0 Ii .0 C)
-3 -! -
H C.Q3
) D ) el° ~ uD N C3
Cp.g) D

]00

. _ . . .. . . . V! 0 W P

(o-QpjgSocm (o)()opiu I C; CD
=
-a.-
.' .

-, to

4 0 U

0 0
----- - - - - - - - W ?

inA 03 n n i a

N L

U 0 ~C3
U,

-a -
L C .

00

CC C V

0 0! i m L 0

.~ 00 C 00
 0 0

o C ICc
'K .0 A 0

0 0 j

-C 0 .3I .

-
-~~ - v - 0 c4 r
* L L a)

LD L
- - - - - - - - - - -L ) L
0 0 C3 C

Ln

D (D

0R 0

m c
cc a CD

a C L

C,
m 04

L L
0 C C

E CD
3 3

U L)

- - - - - - -- - cc a

U!U

(ow] (r)opIluSoW (a-) (r)apnl~Sl.Jo

101
 (-Sa) 6SL4d (.6801 *uoqj

0 IR

a: C

a~0- 0

o 0K- 0-
.0
.0 I 0

- - - - - - - - -- - - L .4

L 'L
o CC
L L CD
-- L L

-
------ - o

0 I 0

03 m c
_____CC__ co a)

- L

0 3U
0 0

0 -3 0

0 I 0 c CD
-L
L:

1020
 ('600) eso-d ('6eo) 0904d

0 0 I

~
-~
--- ~
. . . . --..
-
.. 1. . -c
. ,-. /-
. . . . . -.. N" U I .

o 0

4..A .0 A .0

, , 0
,=, .. .... 0 C1
_L L U)
a
LO - - - --- l
L. L

,.2

-o
0 0 - r.
('690) esoqj (-sea) eaoqd c

m--
_
98 1
(6Q) d 'IO eoq
a:o aa)
L
s-. L
c C
0 0

i .0 - - - - -- - - - - - - --- CD

L ID
0 0 C U 0 C

- -- N - - - ccCC
- - E

('I'. o fl. (= IF s n L6
0 0 0 .
- -o -' I-~C -

-~~~
CO - N.
... . *, ~ ~ ~ C .4C; .
4 .
(r 9pCD~o CDw (o; r anu

103
 ('6e0J
0 s)o4d ['6o(] esoqd
a. 2

-- -4- --
. --
. . .0I c
a:----
9
l---
I 2 m

- -
_------ U -
----- --0

.- 0 01

N in N -' -

0 L3 0 -
L m:
Un c
.. ..
.. '..
.. .. 00A
3 I3I ifl
00
.j I~
0

r L m

- - , ---
- -

r -90 I-Se L0
L
a: .-
a ...J)
-- _. I-,
3-"_

- c

-3
- - - -0 - I - 0 :
5
U
9DD4 ~QL 2L L~D

0 0
-C' L
CD
I f 0
0 A
- L L an)-0
I0
(oW) [f) oPn ,u6°o co
L-
1 Itj oPnv u6°ij L

0
-
-
-- -- - -
L'6 - ..... -4
0(
Q
tow (r tow anlu 0piuo -r

- .3 104
|0
 (-6eoo) ao4d (-6 oo eeoq d

00

- 0 I,
U? I " . '

C-

WN
C0 C
, - - -- - - - - -. - ." I --
4 a<
{
0 CD
9 a
L L

, S.J ,
0L r

0 o

. OZ 0;
.,
. ..
.....
.J. -
U

I
-

-0 a0-
0U€ a i I I, a
- -0 -6 0
N~~~~ 0 N0 0 0)

." \ , , "
C 0 _u
o 0

a 0

0L L - -C

0 c c
(DW) (r) oPnl.u6°lW UU
(ow,,(rl aPnil-Bu6D

-- -4 -

e - 0

tow Cr.)lI~Wtw)()apiu

Ia105
 (-sea) *B04J (-Bea 68l04d
B5 . a

0 0a
4 3
S Ij

- - -. - -

0 C L. cC

-i 0.3
0 C

LI!

-. - S I ;
-

0 51 0

r
-----
---- ~ ~~ ~ I. -
L co L

C-)

C2 .c CD
r; A 0

o 1o
L
.0 L
--------- -

-1 M 0

CL L
IDU~

0 CJ

0 0

100
 69
(3 ~ ~0 0I

ai

N NI

0 C3Ma
0 C 0

A C3

80 -- .3 d .j

- N C

-r . J 4
I- L3
UU C

C. CD
c. C:

m..

o3 0
La L
0 -u0
o) 8
Oa a L

(O(f) 0pnl'juSDW (ow) inl epnllu~oW -

107,
 0 0
(-Bea 9 Hd (-Sao) 6804.4

=r -

w 0 -r -
- p.,

0 -, 0
- c o
--- - - - - - -- - 0 0 a

L L~
NN

-~ --- -J -2 -

-- UL L .j
0 0C 0 c J
L L.
U L

-~
0

- - CD
0 --

.00

(ow) cr) opnl1uBCW (ow) (r)apnijuSoW

-C
a

m CD
00

CD
0. (D0
0n M

0C

-. 0 .0

£~ C C 0
aL L

L L :

c I
-- . V)

(ow)~~~ -r ap~uo o) r p~uo

- a0-
 ('600) oOODd (-6ea) OO8Cd

---- c" to
UU

I ' .c ' "

a A0 *A
c
.
0

....
...
..
, .. .4
- 40.'
, 0
*.'. "

0 L ' ~N .)
. .
-__. " I N.
a0 cf 0 c.
E. I C
. n L . .' --
LL
L L -

0 E.

a a:
00

0 0 - . N 0
- N0
fa
•
4 -
0
(ow) (r) Qpnj u6DW (OW) (r) oPnvju6OW
(D
U, .) -0- c
c
(.600] *aoqd [&6001 0Dq a)

0 T~

--- - - - : a -- -

C.3 0
I- L

nH - L
.- - - -n ,-6
.- C.;
0 0
J3 A

.0.'N . N
0 0D

- ----- - - --
- 0 C) --
C.)
- ., I
,, I0 I I
- - -- --- --- - .

L L'
,,z• *

in
(ow -Cij~c (ow in snluo
i I0)

/ 90
a 109
 (-00) O0 O4d (' o4 d
9001

r- (-- I--

- .
o
0
c0
S1,

A 0
- - 0 0
0
- N . ~ .3

C; -C
° a) - I .d
C m
E E
n L 01
L"o
" ----- _: a ."-
4 c 0 - *c 0
- - - - - -- 0 C
C-
C3 C

. . . .. ... -L :3 a -C ,: .
.0 9 1-

P1~~- 1 L

Q C-)
.... nn 04 I L~ D

a: - a: -.
- - -- - - Ic

0 l0
L2

N 0 0 I 01 0 0

(o~.
----- (r - - opnvi!uSW 0 0 (OW) (rl
- sPn1 !uBa C 3
* 0 " 0 D

L- L

m1C.0 C-
- - -. -1 -. -l- - - - ! I-

- 4 .4.;.
0 009

(a- (r
jSW1wi(iapiuD -.

110I.,
 (-Boo) e9D~d .6001 ega'4d

- ----- ---- - -- q
0 c

a Ma
L

Ur
m
L
L

0
L

L
i
c

(D

(D

00
N 0 Nto to

- - - - - - - - - - - )

9a

0 :
0 :
L L

6 0 L,

0J C 0 0
L0 L L

- ---u--- -- - - - a
9. N' W N

0 w
~
(owl
pjuc ~ ~ ~(r , ow r pjuD

- *o a CI
 (-Bea) oesmd (-sea) 6ooed

IR 0
rr-4I -

CC
r 1 I- -

C.3 L.3
- .- I :

.. I.
- - - - - - _-
o c

- - -"-m- - -

.0 .l0
L L
N - I 0 c -:

I-I - * ' I - .
E C)
S - N..)• .- L n
-,-

cr
a -C

- ,. ,U, e.d .0,o I o -" IC

-

- - 0 - - 10 - - -- 0 C)

N '_" L)

-- '.

""1Oo e0 ,
" a
s I-'I- C
0 Na LOa
----------------------------------- C

D 0
0 C
- E

-m -i

4,, C; L L
LI
0

C.) F { ()e~' .6 9 C.)n!l~l

I', - I -

0 a -

(o-) (r opnii.u6ow tow) (r)opnv;1u6ow

112
(-Bo) 094 [.SeQ) 680Y 4
0 IR

a: 2
C I3

0 )

-0 C~, 0 _jIf
00 0

LN L

L L -V

i C.)

D.
a:a -
0 7oI

-- .0 1

0 0)
-c
-~r ~ . r co
L

CD

0, 0

J Ii
rZ L L

==3 cr
o 0 -

m-

IQ 09 !2 0
r- AJ .J -
spluo -0- 00) (ow (r 0pju

113~-J 2
 (-Boo) esoqd (.68o O9OLqj

CCo

- 0 C.3 U
- 4. c
0

- 0A 0

I 0

-~ --- - -,-N -~ -LI

-~~~C
. - .
-) C 0 -
oE
L L
Z0 U

- CD
- - - L g C
-- 0

00

op-uo .ow (w m apluo Ir

- 0 I
- -
.90 Gq- -sa0 - - q- M

0 . 10 N 0 1 4.2

C-2 0

- a
- CD
00

0a V

-. L0L
- ---
--- -; ; a,
0 .4 I E 4 *Ct
-L 0L .3a
0 I 0 c (1

E a0L 0
L0
L

U- C
V .)
cc cr E"..
- I-
-, -

C;

EmuI)
(I-)
spn-pu6oW (ow) in spnl1)uGW

114
 ('E00] e d (-Boo) eqo d

a
13 a
-- -.-....
- -- - -- - - - - in - (
.- - - - IJe

C3 u

C II C

a A. 0
.. . . . . " . . ,
o 0 .-.j
- -- -- - - - - - . ,---
.-_....- C m. ".

0. 0 . 0C

- _. - - .-

." . rr .0
--- v *-" I -v *'
... - ..- -- -- - " Ln

:) t

. 00
I TI
In
,C.,, C -

--- 1

BO- I C

N~~ I N V AI 0
-- a a - - 0 0 -
0)
0
.: i L
t.- I-. L
o o
--- - - m - -
-U-
. . . . - - ,- . -
0

o o -
L C C
o 0 c (
CC3
.0 .0
-- .- | .- -
L L)
.4 . .

- C 3

0 0

(OW) I"n opnlusaw (o-] (r) opnI u6oW

115
 (-sea) 990dL (-Bo) esoq. a
0 IR

- - - - vI I-

0
U! a

-C
0 C3 7
C) C0

Ifl -

I, * W) - ---

(Dw( 0r a-,i-ow --Iu~o

00 -~ .

4
esoq- 0..)

0 r L

L0 C

- - -~- - -0-
-u A .3 -u
0

OD 0
S 0 a:

L I-
- 0 C e
0 ~~ -- L

IIfl
CC
0 I I'

(6(0O) S60L41 Lul

5 a 5116
 e(-Sa)
es00
4j .69o1 eeo j

tn 0

.........--- - -

-- -- - -- - - - -
,..- -
--- - -
-. . -o I , --.
o A6Occ
0 -,

N .j
a .2 C

0 O I -0 ..

L- L
I " I "E

0
at <
-. u C,. © -C.. ,, Ln
Ln

a- - 0 I
- - (r w.-] In")- -*,.u r n('3cW"

-C
6
(aw] (r) oprluSDW (ow) ir) apfl1ju eW

['6eo ) e880cl 1 6 0 ) eeo4o 0

.C

20 0 -

. CD
.. .....
.
-1 CC 03

"' - --

. .. -' " .- I .-

- . . . . . . --,-
z 0 0
M
--
S 0 00 0
------ - LM L
N-
C.
N

a C: 0

0
C M

------- -LL a a :

00 - a-

t-) ir) spn'iuoow p.] (r) epnIIu6o,,

117
 6
('60Q) oeoL-d V' eoI 6S0 Ld

""I cI I-
U L)

'_ I Co "
0 cc
-,o 0 . C u)

*0 m

. L

- 0 * I 0 n
- - - - - ------ l
• I - € I- I o -

a:Z W A.
(o- (r pSu0 , o ( t' ~ 1uo,-'
0 m
"

I)

0r
(D.
pnISWPInIri plu -r

m . . 0 . . . _m0 _ -- I *- C

0 T
-

-- - I-.

cn c

. ... ....
------ -
1-. -"
- - - -
(.90 een".q - c

0 I
C
-, 0 I OD I
- 0 - -- '

4 - L- L L

-- C) .
U

C CC

0 CCo C

8118
C- em0

|L

(m) opIU0 oW) r pi-

 6 0
(-Bo0) eao4d (60) 4d

.
--- - - -

(C m
- I r

I - - c
A 0 x 0
0-_.0--
0' 0. .

N . '
-.. . a M4
, o

- " Eb
I oL 0a L -J
c
-, --- C. , - C. 0
L- L
C C

D o (D
. . . . .I [.0

-
I
L
. . . .. -- .03 -u
. L.

U, i 0i U, - U, 0

-"-0

(o.' (r epnTu~oW (oW) (r) apn Tu6aw

('o ) eod( •60 SqdL

9'od0
(C-,! (r O irso I ow) (r)onju CD
(-Sao0 984 -6 a

51 0 T

.. . I 0
~0

C.. --
-------
--, - - -T
.4 c
-- I . r_
- / at 0 U, .a L
0C C 0
C
C

I<A - 0"
0 M zA
0.
00
-
L c C.
-
, , .L. **,

N U In0 0 cU

I() cn' epny!'!u~ow (ow) ri' epnyw'.uo H

119
 (-BO) eood I'600l osod
.51

..... . 1i 0 a /

- ----
-
--
-
--- /i
-
co
N Cc
-

0 0
0".'
C)"
41 0.' .0
-- -- - -.-.-.-.-.-.-.-.--
-. .
-~~~ -- - - ---

L.. .j L. j 0
0 0 0 0
0Z C 0
C. C.
u 3 L _
U ) ca 0
L L
C;C
@t u1 tn 0tn c . 01
LiU 3 !L 0 CD

aa

o.-o

enlsoq
(OW)(rO opnvu~j
(-600(] troiaplBL
'6e0
( (-600 Gaq Lp~uo
- . a- , a C
L
0J 09L~ 91 OC SUU

-,. . - 0

N CC a )

*i
0
c0 n
o ,
0 *'
0
0
U
-I

- 0 n .0 L-
a~ L 0 L. 0
N j N j

- a0 03 I"0 0 03
0 a
IL
< L) C

cc:a

- - - - - - - -

0Cc
! F 1? 0n-I
L

V! 0 IV Li0

• I I
(ow) cni opnyIu-oW cow, fin pnP.USCW

120
 6 oeoq
('600) 090 4d

CC

in I--.
U C

-- 0 a

I0 CC
C
.0
N

0 0 0w .
---
----------- - - - - -0

'N .LN

---------
L .L
43

-~
0 C-
L EL L' D

- L L. I
IO :) I ,

a: a
I CD

- - -- - - - -f - L

-c
0 - 0

(Dw) (r)apniVUS0o (OW) (r) apvil~w

c
a)
ego" L
E-6)008(.600) egoq4j L.
:7
0 0

(D
a:- -'
a: m a

- - -.- - - - - - -C..)-

0 C
-0 0 In
------ - S
X. L

0 04 .0

- 0 00
(ow -r- p~ uo (ow in spnl uBI
 ('60) OOD4d (-SOO) 68044

- - - -
- -

- - - - - - - - - -t:L

c C

I A 0 A 0
0 j 0 .j

----- -. -- .0Lu Ln
Nn-) N Co
E E
I-L L
oL L CD

0a C3 mT

0 -C
a:K
aa)
-~ I L
('60) G8~jd(-Se) esqC

.0 - 0

- - - ---------- - - - - - a

in
i N U . i Nn-

JJ
J
-) E
C3 C)
9 C.
r*Le dV~3 8~ L
9 0 0a

L: L

0 C

Cc m
- -' cr

v- nn DIlUD (c- n opnuo

- C. ~ L22
 6
('690) o04,d (- e) eso4d
Io 0 9) 2
' 9'
------------ a m
F_ H
-I
IC

0 0
0
-A - -6 0

-- 0-p 0. ..
. .00
L - -- --- -- 0 L 0
JE 'N .3 N -J 0
..j a.-.
Ca 0
r L. L~C

L L 0
, C:1 I:CI
I r I 0
C3 m

00

-- - - ----- ----- --

0-- a:
(DW] (r apn-u6oW (ow) (r) apnluc 0
C:
0)
L
-SEQ) nuo-cl 1-Q) )S9O4 L.
v! ID

-3
- _ €- r.

.0 m .0

-F aDa
a ,

- A ~ I i a. C.
co C
-in IE In-lI - -

-0 C.,
E _3a
- - - - - • - - - a . - - - .- - - - -
0
'"
c- -
din ,07

L2 .

1239
 (-BSo) eso4j 1.6S0) eeO48

13 R
-- -- - -.- F-
- co
# I - -

N C) ci

-- 4 0

N 0 - - -\ 0

SL
o 00 0 0

L L
-- L L.
IE E
; .. - U ' . . . ,.,- U,

m: C

--------- - - -

(o)(r)o Diiso (ow) irn o - CaC.

(-ea 60d(-o) 68044

Ci
-- - - - - ,
U- 3
- - - :)
L) - ------
a: 0 c

> C C~
.. 0 I ~ C -C.

- .13 U, .1
.24
N ~.

0 U 0
cc crS
cc3I. .
IR

(o- (r) apl4S

- a: 124
 ('6O) eo d (-6eO] eeood
a 51

- - - i - - -

- -' ----- - - - - ...

-
0 0

A 0 A 0
0 -- - - -

- - - --- - - --
, ,," C, , C r
/D . 3L L
c. o
-D 0 I'

0, oI u;<
,< o L

tow) (r) 0G
G&0 -- tow)(-ea
(r) 9Pnju6DI 0vq CPn.u)
L

N N
mC-ID

C
0- a ' L
- c ------------- - c

-3

025

.<0 C 0 C .

L0 L 0

L) U

0 0Ci

L-- -
L- C . L C .

0 0

(0w) (r) apnliiusow (o (flj spnmu~oW

125
 -- --
-

L02
- - - ----- -- L

c c
.
- - - - -
-

c '
a
0 .003 .
.-,-
--t
-- -. ----E --.
" e." 0,
-
N r -
-- - - -- - -- - - - - - -~ -Q ~
L ~ 0
L
- 0 C
1 .J *L I
L
- L
C3L 0M

: ::
W Wl
L

o = (D
SV!
- C
U.I- I LI-,0
(Ow) Cr vpn'juloW (oW) (F) opn'"usow

L
(.600)( L
a L)

I Ia ° °
- a
--- . "-.

a: 0)

0 a

0 C~
0 c
0 1 z

C. C0.4

-- 2
I 2
- 0 0
(ow (r C.-0 ow r piuo

- - C 126
 (-Bea) eeoqd (6o0) eeoqd

- ,-" 8 -- 0 -C-
-- - C. C3
- -. o 0
'c 0 I
*0 A~ o 0
,4 ..
Sn 0 C)
. 0 "'

3 N -
- - o 0'

CE (Ci I -,
L 'L c.
aZ 0 0
L L 0
- L L .- )

.. ,0 - , .4 *.'
J .4 Ci

o. C ooq oo C"60

L)
C-D

Eg L

cc I D

-------------- - - - - -- - -u

CDC
0 C 0
U

-~J . M
-- :a - - -- , 0
- - - - - - - --- -0 - . -

L!C L,
0 L C
NN L
- 0 S0 Ia .4

0
- CO0

- L - L
.~C

- - - -- - -0 ~

- z

- 127

- C.) C.
 (-BO]o 0
d t'6o ) 690 4d
, ,9
a a I 0

-- - -, -, ,
I C I.-

CI C
C 4C
o 0
- ----
- - - -o-- - -

. .N 'N ..

L L

1 0L- -n-.-:-. - -- -- 0 ,
Ca C3 1--
LnI-
0 0

- U) C

(ow) (r) aPnluBDW' (DW] (r) QPn' TU6DW

('N9O) oo D4 (' 801} l

f4 0

CD
- L

I F ~.-,..
0 0 7
03

- - - - - - - -
-
- ~ -
:S
1
DC C 0

-A 0 'A 0
. .j . . 4 0
-
0 (. (D

L0 u, L.

-a 3 c C
.4
.j 0 4 .
-" .- 1 . C. I '.4 0 L
L. I

,44., .,-.
-o1o)en,,~
t, 00 Icr 00n C)D~
Li C2

9pjuCCc) (r) Uow r piuo

I- 0 ~ 128
 000
(*6-0) ~4d (.60) 0ODLd

0
--9' ----- -

-- ---
- - ---
"! c

-- 0 0

L 0o' L'

0 0.
0 C.

L0 00 L C

M C3
- -- - -- -

_- L L. 3

N Li ! ! 0 0!0f

c -, 0

- - L

0 r -u CD
- D - - - - - - -) -

a 0

0 a

C) C3 U
C3,
0 c a C
L -u
U 0

I-u
- 4 C -
CC
V! 0! fq V !

tc),j (r 9pnj-6D (ow triepn~uCo

- - 129
 - - --- - -- --

A 0 Co

0 C)

... -
-

L L

.,
iL
"-L-.:
, .. "
..
0

.. / s
, °L
-? o I

" '' ,,4 N U

130)
(cw)(r)opmluco (DW (r apnluCW -
 (.60f) GG0 Hd v-Beai 69
oyd
8 o 8

ac:

0 0

0 A o 0
-A0 IA0

- - - - -. - - 0
E E

(D-) ID

L L

0 C 0 D

L L L

CE a!

aoluo In 0D. D) r plU

9 6 66 0 0 M,
(66Q ~4d (660) L
L
P40

F I-. -- )

0 0

-c- - - - - - 0 -
*'K 0 1 A 0 IF)

N j

0 c 0 -C

9a L
NL

L L
:3 L .
Lr m

- 0 -

(ow (r oi r p~u Ip~uo

131
 E-600) eooqu 1-600) SD

- n)
a:

a 0 :

-N - 3

0 t. L

.LL

L) 0 -
oZ 0 CL ()

CC I U

M: C

- L

- -0 - - - 03

tO)t~ pn~GO4(00) ((") om4u

0
0
a 0 L
8 0 ~ 0 9 7D

a: 3

o.j 1 0L
Ln 0 L

0 c 0 L

c F
- ,
-- I
-- IF -~ L
9--
0 C 0VC

32
 (-Boa) 09o441 ('590) .eo4,:

a a aI
Ir _ R 0 9I

r- - a ! ,,.
- 0 - -
---- - - - -0 0

-. 9 I ..
a c

---------------------------------
-i ri) -)~

N c,
.0 cCn I(t)
0
Oz -)0
- L

-£- 0 M -C3

oo D oo c m
f0
- C€- _U

L - - - L

" )
(-Boo) Sa~qc ( 690 aoBo4

- L L
-- b II z .O
L L

ar a .)
ccl'

0 -n II 0
-a

A 0

II
. . . . ---

(o ) (r) epnluVoW (ow Irt epn'vju~(

133
 (.600) BOqdq ('6O01 9804d

-.- I/ r-- 1 -'-C D

.

00 0

cc c
0
0 f..j c0D
..
. C2 . L2 I
- --- - - - - --

0 L 0 L
N 0, 0
A 0 - - - - 0
0~
0
L -L
,, C ' 0 C L
N . _)

-C3 L: 0D

CC r .-
U) cr co

--------- a)-.0
N 0 N tf.. . 0 (D

,0 ( I ", 0-
(D-) (r) aPriYIUSOW (ow) (r)apnIjuSoW C
-" r I . r*
-. ',-, 0 L
0-

. . . 0 . . -,

C.34
a a

c 02

- - - - Ii, - - - - -
0
0
a I 0
A 0

L 0 L
-N .- , , - Na 4J
M (D

-0 'C 0 -c
,2 U. "(-, ,Ca * ~ .4
-~ 4 I C*-

L- *LN~N
L U130 0 L

-o en)u l .r
0 C
(wl r enl U m
--

N M. i

-DW (r) gpl-D (r) ap~u Iow

- Ca C134

- I
 5
('6eo)
eoya (' 0) OSo4d

0 -R

.. .. .. .."- 5RJ
tn[
,- -7
- l o
'D C
c gL
a~ 0 L

- C 'c
4A0A0

N 0
-- o '' a
- - -- -

: : j
o 0.4.
M .)c
0 C
, E
L L
C0 C L
L L
-- L L 0

0. C3
m mD

N N I l! •
.. % . (4
n I 4 --4

(r)
r nW) (ow) (r) epnliu6rw

fl I - -
-6 0] OSoul (-sea) eeoqj O
- I I r
2 a 7 C;O
- - -- . -,-

- ---- -- -- --------- U

0 C

in a)
-- - -

N N

0 .c 0 c-

9Et 000
L -L
L L

a: a:

Cr ~ 0
ly U!
-.
r ~ iL
Q I

(ow] (r) opniiu~ow (ow) (r opnhiu~oW

135
 LIST OF REFERENCES

1. Janaswamy, R., "An Accurate Moment Method Model for the

Linear Tapered Slot Antenna," paper submitted for
publication, July 1988.

2. Stutzman, W. L., and Thiele, G. A., Antenna Theory and

Design, Chapter 7, John Wiley & Sons, Inc., 1981.
3. Harrington, R. F., Field Computation by Moment Methods,
Chapter 1, Krieger Inc., 1968.
4. Kollberg, E. L., and others, "New Results on Tapered Slot
Endfire Antennas or Dielectric Substrates," paper
presented at IEEE International Conference on Infrared
and Millimeter Waves, 8th, Miami, Florida, December 1983.

5. Janaswamy, R., and Schaubert, D. H., "Analysis of the

Tapered Slot Antenna," IEEE Transactions on Antennas and
Propagation, V. AP-35, No. 9, pp. 1058-1065, September
1987.

6. Sharma, A. K., Wilson, R. M., and Rosen, A., "An

Experimental Investigation of Millimeter-wave Fin
Antennas," RCA Laboratories, Princeton, New Jersey, 1985.

7. Thungren, T., Kollberg, E. L., and Yngvesson, K. S.,

"Vivaldi Antennas for Single Beam Integrated Receivers,"
presented at 12th European Microwave Conference,
Helsinki, Finland, 1982.

136
 INITIAL DISTRIBUTION LIST

1. Defense Technical Information Center 2

Cameron Station
Alexandria, Virginia 22304-6145

2. Library, Code 0142 2

Naval Postgraduate School
Monterey, California 93943-5002
3. Professor R. Janaswamy, Code 62RJ 4
Department of Electrical and Computer Engineering
Naval Postgraduate School
Monterey, California 93943

4. Professor M. A. Morgan, Code 62MW

Department of Electrical and Computer Engineering
Naval Postgraduate School
Monterey, California 93943
5. Chairman, Code 62
Department of Electrical and Computer Engineering
Naval Postgraduate School
Monterey, California 93943

6. Thomas P. Kelly 2
11 Duval Court
Wilmington, Delaware 19808
7. Professor Daniel H. Schaubert
Department of Electrical and Computer Engineering
University of Massachusetts
Amhurst, Massachusetts 01003
8. Professor K. S. Yngvesson
Department of Electrical and Computer Engineering
University of Massachusetts
Amhurst, Massachusetts 01003

9. J. B. L. Rao
Building #47, Room 202
Naval Research Laboratory
Washington, D.C. 20375

137