Beruflich Dokumente
Kultur Dokumente
M
ost short-range
Here are design technqiues RF transmitters
for matching to the very use a small loop
low impedance of small antenna for radiation of
loop antennas, including the signal. This article
designs that include addi- provides a review of the
tional low pass filtering impedance characteris-
tics of these antennas
and describes methods of matching the output
of the transmitter to the very low impedance
of the loop antenna, with adequate rejection of
the harmonics.
For the other commonly used (out- At 315 MHz At 433.92 MHz
side the US) frequency of 433.92
MHz, the quantities are C1= 2.82 pF C1 = 1.47 pF
C2= 63 pF C2 = 43 pF
Rrad = .093 ohm L1= 36 nH L1 = 27 nH
Rloss = 0.35 ohm
L = 95 nH Table 1 · Ideal component values
for split capacitor match.
Note that the radiation resistance
is extremely small. In addition, the
resistance arising from the dissipative L1, tune out the reactance of the Figure 2 · Split capacitor matching
losses can be more than 10 times the equivalent parallel circuit. network with bias inductor.
radiation resistance. This means that The combination of C1 and the
the best radiation efficiency possible loop inductance form a positive reac-
with this loop is about 8 percent at 315 tance at the frequency of interest, so The frequency dependence of this
MHz and 27 percent at 433.92 MHz. we can consider the two capacitors match at 315 MHz is shown in the RF
The matching network must minimize and loop inductance as an “L” match- power transfer curves of Figure 3.
mismatch losses and additional dissi- ing network (shunt C, series L) that These were computed by evaluating
pative losses from the matching com- transforms the small loop resistance the expression for the power deliv-
ponents. Typically, a small loop will up to 125 ohms. Looking from left to ered from a source (RS) to a load
radiate only a few percent of the right, it is a low pass high-to-low impedance (RL + XL), where the load
power from the transmitter. impedance matching network. The impedance is the loop antenna
bias inductor L1 is not essential to impedance transformed by the
Basic Matching Network matching, but it is needed as a DC matching network. This is the same
The simplest matching network is path to provide operating current for quantity as the mismatch loss
the “split capacitor,” described in a the PA and can double as a tuning between a real source resistance and
recent article [3]. Connecting it to the reactance for C2 and the small stray a complex load impedance.
PA output with the bias inductor, as capacitance of the PA output.
shown in Figure 2, makes it possible The values that give a perfect Pout/Pin = 4RSRL/((RS + RL)2 + XL2) (4)
to adjust the value of C2 so that it match for the loop antenna described
resonates with the parallel combina- above are given in Table 1. The C2 This expression was multiplied by
tion of L1, the PA-related capacitance, capacitance in the table does not the antenna efficiency and the power
and the residual reactance from C1 include approximately 2 pF that loss arising from the matching com-
and the loop antenna inductance. The comes from the PA output and PC ponents to get the total ratio of power
equivalent series resistance (ESR) of board stray capacitance. This 2 pF radiated to power available.
the capacitor C1 is typically 0.138 was added to the C2 values in all the All plots are shown at 315 MHz,
ohms, so the total resistance of the matching calculations in this paper. and discussion of frequency depen-
small loop with the series capacitor is
0.46 ohms at 315 MHz.
The Appendix provided at the
end of this article explains the
transformations between series and
parallel impedances and shows how
to determine the values of C1 and
C2. At the resonant matching net-
work frequency of 315 MHz, the tiny
loop resistance is transformed by the
series reactance of the loop and C1 to
an equivalent parallel circuit with
the optimum load resistance of 125
ohms (best load impedance for max-
imum MAX7044 efficiency). The par-
allel capacitance, C2, and the paral-
lel inductance of the bias inductor, Figure 3 · Power transfer from RFIC transmitter to loop antenna.
dence applies just to 315 MHz. The At 315 MHz At 433.92 MHz At 315 MHz At 433.92 MHz
behavior at 433.92 MHz is similar
but is not shown. C1= 3.0 pF C1 = 1.5 pF C1= 3.3 pF C1 = 1.65 pF
Assuming that the model for the C2= 33 pF C2 = 27 pF C2= 22 pF C2 = 15 pF
loop antenna is correct and that exact L1= 27 nH L1 = 20 nH L1= 27 nH L2 = 20 nH
values of the matching capacitors are
realized, the mismatch loss is 0 dB, Table 2 · Practical component val- Table 3 · Component values for a
and the antenna loss is just the effi- ues for the split capacitor matching wider bandwidth split capacitor
ciency loss and dissipative loss added network. match.
by the capacitors (radiation resis-
tance divided by total resistance) of
–14.1 dB. This match is a vast nents are 5 percent high in value, for Notice how narrow the perfectly
improvement over the 36.2 dB loss example, the power transfer drops to tuned networks are and how the
with no match at all (25 dB mismatch –26 dB. “detuned” networks have more loss
loss plus 11.2 dB efficiency loss) and The power transfer characteristic but are wider in bandwidth.
the 34.7 dB loss from a single shunt can be widened in frequency—and How well do these simple split
capacitance that tunes out the anten- made less sensitive to component tol- capacitor networks reject harmonics?
na reactance (19 dB mismatch loss erances—by “detuning” the matching Figure 3, which extends to 1,000
plus 15.7 dB efficiency and capacitor network. This can be done by the MHz, shows that the theoretical
dissipative loss). The power transfer “brute force” method of simply adding match frequency response is down by
for the single shunt capacitor “match” resistance to the loop antenna, or by 56 dB at the second harmonic and 58
is included in the plots as a reference. transforming the impedance to a dB at the third harmonic. Because it
In practice, the small loop anten- value that is not perfectly matched to is down by 14 dB at the fundamental
na has a much lower Q than theory the transmitter. With either frequency, its second and third har-
predicts. Calculations made from lab- approach, the matching bandwidth is monic rejection is 42 dB and 44 dB,
oratory measurements of the printed broadened at the expense of either respectively. Practical and “detuned”
circuit board loop of Figure 1 yield a higher power dissipation in the matches are much more typical, so
total equivalent series resistance of added resistor or higher mismatch they are better indicators of actual
2.2 ohms at 315 MHz instead of the loss in the detuned matching net- harmonic rejection. The practical
theoretical 0.46 ohms. The big work. It may be better to take some match is down 20 dB at the funda-
increase in resistance (all of it loss power loss in return for having a pre- mental and 50 dB at the second har-
resistance) can be attributed to lower dictable power transfer, because the monic, so the second harmonic rejec-
Q capacitors than predicted, higher penalty for drifting off frequency in a tion is 30 dB. The “detuned” match is
skin effect losses from solder as narrowband match is so high. down 22 dB at the fundamental and
opposed to pure copper and proximity The broadening approach used 46 dB at the second harmonic, so the
of the loop antenna to metal traces on here is to match the loop antenna to second harmonic rejection is 24 dB.
the board and to the board material. a higher impedance than the This is not enough rejection if the
Using this resistance, the standard MAX7044 likes to see (500 to 1,000 transmitter radiates the maximum
capacitor and inductor values that ohms instead of 125 ohms) and accept average power allowable in the US at
match the loop are given in Table 2. the mismatch loss (along with the 315 MHz. The allowable radiated
The power transfer for the practi- unavoidable dissipative losses). This field strength of about 6,000 µV/m
cal loop antenna is also shown in approach has an added benefit of corresponds to radiated power of
Figure 3. Because the loss resistance reducing the operating current. –19.6 dBm. The second harmonic
of the practical loop is roughly four Table 3 shows a set of L and C val- cannot exceed 200 µV/m (–49 dBm),
times higher than the theoretical ues that transform the loop so almost 30 dB of harmonic rejection
loop, the best power transfer is about impedance to about 500 ohms. They is needed at the maximum allowable
–20 dB instead of –14 dB. Although are rounded to the nearest standard average power level. Because FCC
the power transfer curve is wider in component values. regulations for the 260-470 MHz
frequency than for the theoretical This circuit decreases the power unlicensed band permit low duty
loop, it is still narrow enough for com- transfer at 315 MHz to –22 dB, but cycle peak power radiation at levels
ponent tolerances to move the peak reduces the variation in loss with a 5 as much as 20 dB above the average
to another frequency and reduce the percent component tolerance to 3 dB. power, there are situations where
power transfer at the intended fre- Figure 3 shows the losses in the even more than 30 dB of second har-
quency. If all three matching compo- tuning networks discussed above. monic rejection is needed.
Figure 5 · Power transfer from RFIC transmitter to loop antenna with a low
pass filter added to matching section.
The combination split-C and low pass 2. W.A. Stutzman and G.A. Thiele,
filter networks chosen in this article Antenna Theory and Design, Wiley,
can improve harmonic rejection by 1981.
roughly 20 dB over that of the simple 3. F. Dacus, J. Van Niekerk and S.
split-C matching network. Bible, “Introducing Loop Antennas
The values in the matching net- for Short-Range Radios,” Microwaves
works presented here may need to be & RF, July 2002, pp. 80-88.
adjusted by the user slightly to
accommodate stray reactances and Acknowledgments
losses in the circuit board or the The author wishes to thank
matching components themselves. Messrs. Andy Zocher and Farron
Care should also be taken to be sure Dacus of Maxim Integrated Products
that all of the matching components for their suggestions and guidance
are operating well below (two octaves while reviewing this article.
preferred) their self-resonant fre-
quencies (SRF). Appendix
What is more important than the An Appendix, “Using Reactances
specific value of each matching com- to Create Resistive Impedance
ponent is the basic structure of these Matches,” follows on page 28.
matching networks. The purpose of
the split-C section is to transform the Author Information
loop resistance value to a more rea- Larry Burgess works in Corporate
sonable range. The purpose of the pi Applications at Maxim Integrated
network low pass filter is to reject Products, where he initiates the
higher frequencies, perform addition- development of RF products in the
al matching if needed, and establish 300 to 1000 MHz frequency range. He
the bandwidth of the match. As long received a BSEE and MSEE from
as the user approaches the network MIT and a PhD in EE from the
with this in mind, the right compo- University of Pennsylvania. Dr.
nent values can be found. Burgess has worked for over 30 years
in applications for cellular, unli-
References censed wireless, military advanced
1. C. Balanis, Antenna Theory, communications, and radar. He can
Analysis and Design, Harper and be reached by e-mail at:
Row, 1982. larry_burgess@maximhq.com.
Parallel and Series Impedances A4 leads to the following expressions for RS and XS.
A resistance value can be increased or decreased over
a limited frequency range by adding a series and paral-
(A5)
lel reactance to the resistor. The key to understanding
the way this works is the equivalence between a series
combination and parallel combination of a resistance, R, Which can be expressed in terms of the quantity QP =
and a reactance, ±jX. The series and parallel circuit rep- RP/XP as
resentations are shown in Figure A1.
The parallel resistance and reactance as a function of
(A-6)
the series resistance and reactance can be determined by
equating the admittances of the two circuits.
In each equation, the Q has the same meaning as the
Q of a tuned circuit. It is a ratio of stored to dissipated
(A1)
energy (reactance to resistance in series, resistance to
reactance in parallel). It is also a measure of how narrow
Equating the real and imaginary parts leads to these the bandwidth is of an impedance transformation made
expressions: with reactances. The higher the Q, the narrower the
transformation bandwidth.
(A2)
Calculating Matching Network Reactances
The purpose of the split capacitor matching network
Which can be expressed in terms of QS = XS/RS, as shown in simplified form in Figure A2 is to transform
the small series resistance of the loop antenna, RA, to a
larger resistance, RL, that can be driven by the power
(A3)
amplifier. Because RA and RL are known, the desired QS
can be found from Equation A3 by letting RS = RA and
Likewise, the series resistance and reactance as a func- RP = RL.
tion of the parallel resistance and reactance can be deter-
mined by equating the impedances of the two circuits. QS = (RL/RA – 1)1/2 (A7)
XA – X1 = QS RA (A8)
Equating the real and imaginary parts of Equation
From which X1 can be determined because XA is known.
Now the parallel impedance seen by the power amplifier
has a real part equal to RL and an imaginary part that
can be calculated from equation A3.
(A9)
Figure A1 · Series and parallel representations of an
impedance. The parallel reactance X2 needed to tune this out is
equal and opposite to XP. The split capacitor matching
network is based on the assumption that the negative
series reactance from the series capacitor is smaller in
magnitude than the antenna reactance XA. Therefore,
LOOP the XP reactance that results from adding a capacitor in
ANTENNA series to the antenna is positive. This means that a neg-
ative reactance is needed to tune out XP. Given that the
power amplifier needs a parallel inductor for its bias, a
parallel capacitance must be chosen that tunes out the
Figure A2 · Simplified representation of the matching positive reactance formed by the parallel combination of
network. XP and the bias inductor.