Professional Documents
Culture Documents
4, A P R I L 1985 397
TABLE I
A BRIEF D E ~ C R I ~ OOF
N SOME NOTATION
2k 5 sinusoidal
expansionltest on
mode patch k.
3k'Cn& 5 c u r r e n t on patch k due t o _?n ,
feed network removed rr
Cn 6 column vector
containing
coef.
the Cnk
15 c u r r e n t on s e c t i o n n o m
f i c r o s t r i pt r a n s m i s s i o nl i n e
_?n=zJn 4 c u r r ef inl a
t mpeaontrtt n
ii=zJi impressedf eo
ceur drpr oae rtntt 1
[ZF] 2 feed
network impedance m a t rriexf e r e n c et od 2N p o r t s
qg [%I
"
a
1) Lower case lettersandthesubscripts k and I ( k , I =
1, 2, -., N ) will be used t o refer to quantities referenced
to the N patch currents. The symbols i or c w irefer to
l
l
I
thepatchcurrentswithorwithoutthefeednetwork,
Fig. 2. Thevenin equivalent of the array plus its feed network.
respectively.
2) Upper case lettersandthesubscripts rn and n (m,n = Q
[Z,] is the open-circuit (i.e., with the antenna arrayremoved)
1, 2, -., 2N) will be used to refer to quantities referenced
impedancematrixforthe transmission linenetwork.Forthe
to the 2N feed ports.
series fed microstrip array, the transmission line feed matrix i s
Table I lists the notation for most of the important quantities. of the form
Additional notation is introduced in the,Appendix.
Fig. 2 shows the Thitvenin equivalent of the array connected [zF 1 1
.to the load or feed network. According to ThBvenin's theorem, LzF2 1
[ZFl=
the. port currents I , , I , ,
'--: 12N are a solution of the order 2N
matrix equation
[Z, + z F ] I = -Voc (11
1
where [Z,,] is thetwo-portimpedancematrixforthe nth
1
where [ZA1. and [Z,] are the 2N X 2N open-circuit impedance sectionofmicrostrip line connectingpatches n and n + 1. In
matrices for the array and feed network, respectively, and PC particular,
is the 2N open-circuit voltage vector. A typical element of Y O c ,
V:', is the voltage induced at port n by the impressed current,
with the feed network removed.
a
NEWLIAN AND TEHAN: MICROSTRIP A R R A Y A N D F E E D NETWORK 399
'k
--
V
+ 'Fn + hk-Lhk 7 I
-
F,2n
- Zn 1 Yn 5n+l
- -"F,Zn+l c Wk /2
1n+ -9k
I,
11
* 'k
Fig. 3. Geometry for the nth section of transmission line connecting patches b
Wk /2
n and n + 1.
t
where as seen in Fig. 3, L,,, Z F l land y, are the length, character- Fig. 4. Geometry of the kth rectangular microstrip patch.
istic impedance. and propagation constant, respectively, for the
rzth section of microstrip transmission line. minus the voltage induced by the unitamplitudemode gl at
Thenextstep is to compute the antenna array open-circuit port IZ (plus markedterminalontop).Thenchoosing J , = 1,
(i.e.: withthefeednetwork removed) impedancematrix [Z,] rz = 1, 2; -, 2 X , (4) becomes
referenced tothe 2N arrayfeedports.Atypicalelement of
[Z,] is given by
iZFl =["z;.'1
0
In (24) and (25), the second term is the modification caused by
the microstrip line connected to port 2. If the line were removed,
Z i = w, I , = 0, and (24) and (25) given the current and input
where impedance for an isolated patch.
If in Fig. 5 we set ZL = Z A ,,
then we have a simple model
for a two-element array (ignoring radiative coupling between the
two patches). If Z i = 2, ,,
then the current on patch1 and the
input impedance are reduced to one-half the value for isolated
Referring to (6) and (7), thepatchimpedancematrix [Z] has patch 1. Thus, although the array elements look as though they
,
only one element, denoted z1 . Because of the symmetric loca- are in series, they appear tobe in parallel as viewed from port 1.
tions of ports l and 2, the elements of the voltages in (8) obey
,
the symmetry u1 = -u2 ,. Ln this case, the solutions to (6) will C Infinite Periodic Array
be IfinFig. 5 we set 2, = Zin, then we have a model for an
infinite periodic array, one section of which is shown in Fig. 5.
This model neglects the radiative mutual coupling between the
elements. Inserting ZL = Zin into (25) produces the following
quadratic equation forZin:
The array open-circuit impedance matrix, given by (9), becomes
( z A I , t a n h y l L 1 +z,)zi2,+z:tadylL1Zin
Fig. 6(a) shows the top view of an edge fed patch, and Fig. 6(b)
shows the same patchbutwiththe edge feedmodeled by a
AIICROSTRIP
currentfilament.Notethatthecurrentfilament is not placed
at the edge, but rather is moved in a distance d from the edge.
The value of d affects the soiution through its effect on the
modal voltages of (8). To illustratethiseffect, Fig. 7 shows
h
the imaginary part(the real part is negligible) of themodal
voltage versus d for a 16.1 cm square patch on a slab with E,. = FEED
CURRENT
2.56, tan 6 = 0.0001, and at f = 300 MHz. Curves are shown for
slabs of thickness T = 0.216 cm, 0.648 cm, and 1.935 cm. Only
(b)
values of d near the edge are shown, and the following discus- Fig. 6 . Top view of an edge fed microstrip antenna showing (a) actual
sion only applies to values of d near the edge. On eachcurve, microstrip transmission line feed and (b) equivalent feed current.
the point d = T i s indicated with a heavy dot. Note that values
o f d < 0 represent feed locations off the patch.
The general shape ofthesecurvescan be understood as fol- d (X)
lows. The i polarizedelectric field intensityunderthepatch, -0.02 -001 0 0.01 0.02 0.03
and therefore the modal voltage, tends to be proportional to the
divergence of the patch current in the immediate vicinity of the T = 0.00216A
feed currentfilament [ l l ] . Near the edge, themodalcurrent
density of ( 5 ) varies as d. Thus, near the edge the divergence of
themodalcurrent is constant, which suggests thatthemodal
voltage should be constant. For d > T , this behavior is illustrated
in Fig. 7. For d < T the modal voltage drops, and finally reaches
a value at the edge (d = 0) of about one-half of the constantvalue
for d > T. The reason for this drop is that the amount of current
in the immediate vicinity of d = 0 is about one-half of the amount
of current in the immediate vicinity of a d > T. This has the ef-
fect of reducing the field intensity and the modal voltage at d = 0
byafactor of abouttwo, as compared to a value of d > T.
According to(21),theinputimpedance
square of the modal
selectedcan have uptoafactor
is proportionaltothe
voltage. Thus,for d < T. the value of d
of foureffectontheinput
:::[< L0.161X4
c y = 2.56
DIMENSIONS nsA
t aIN
T = 0.01935X
OR METERS
= 0.0001
We feel that thisincreasein the divergence as d + 0 will tend Previous solutions to the problem of the junction of a wire
to make the actual voltage and input impedance reasonably con- and.aplate,whether near [ 141 or away from an edge [15]
stant as d + 0. have required. an attachmentmode.Thisattachmentmode
There are at least three ways to include this edge effect in our enforcedcontinuityofcurrentatthe wire to plate .junction,
solution. First, we could replace the sinusoidal expansion mode and also theproper singularity ofthe platesurface current
of (5) with a new mode which explicitly enforcedtheproper densityinthe vicinityofthe attachmentpoint. Here we do
dependence. of thecurrent density as d + 0. Second. we not employ an attachment mode. The reason is that for high Q
could place many of the sinusoidal modes near the edge so that microstrip antennas,with electrically thinsubstrates,thecur-
402 IEEE TRAXSACTIONS
ANTENNAS
PROPAGATION,
ON
AND VOL. AP-33, NO. 4, APRIL 1985
APPENDIX
(b)
MODIFICATION FOR GENERAL PATCHES
Fig. 9. Input impedance of an N = 1, 2, and 3 element series fed microstrip
array. (a) Measured. (b) Calculated.
InSection 11, andinparticular ( 9 , we assumed thatthe
patchcurrentscould be approximated by a single sinusoidal
NETWORK
AND NEWMAN
FEED TEHAN:
AND ARRAY
MICROSTRIP 403
function. This reduced thecomplexity of the remainingequa- R. J. Mailloux, J. F. McIvenna, and N. P. Kernweis, “Microstrip array
technology,” IEEE Trans. Antennas Propagat., vol. AP-29, pp. 25-
tions, but. also limited the results to rectangular patches, where 37, Jan. 1981.
the currentis dominated by the lowest order; polarized resonance. [31 I. E. Rana andN. G . Alexopoulos, “Current distribution and input
Here we will indicate how to modify theseresults for a com- impedance of printed dipoles,” ZEEE Trans. Antennas Propagat., vol.
AP-29, pp. 99-105, Jan. 1981.
pletely general array composed of elements of varying size and N. G . Alexopoulos and I. E. Rana, “Mutual impedance computation,”
r41
shape, and which may be large enough to support hi&er order IEEE Trans. Antennas Propagar., vol. AP-29, pp. 106-111, Jan.
modeswithboth polarizations. Itmust be emphasized that 1981.
r51 D. M. Pozar, “Impedance and mutual coupling of rectangular microstrip
we areonlysettingup a formal solution: andthatcarryingit antennas,” ZEEE Trans. Antennas Propagat., vol. AP-30, pp. 1191-
to the point of numerical results would be much more difficult. 1196, Nov. 1982.
The basic approach will be to obtain a generalization of (5): E. H. Newman, J. H. Richmond, and B. W.Kwan, “Mutual impedance
computation between microstrip antennas,” IEEE Trans. Microwave
since mostofthe remaining equations will apply with minor Theory Tech., vol. MTT-31, pp. 941-945, Nov. 1983.
modification. Thus, we wish t o compute the currents on the array r71 M. D. Deshpande and M. C. Bailey, “Input impedance of microstrip
caused by the current filament J, = J n i at port n, and with the antennas,” ZEEE Trans. Antennas Propagat., vol. AP-30, pp.645-
650, July 1982.
feednetwork removed. Regardless ofthecomplexityofthe M. C. Bailey and M. D. Deshpande, “Integral equation formulation of
array, the current on the N array elements, with the feed net- microstrip antennas,” ZEEE Trans. Antennas Propagat., vol. AP-30,
work removed, canbe written as pp. 651-656, July 1982.
r91 T. Metzler, “Microstrip seriesarrays,” ZEEE Trans. Antennas
N’ Propagat., vol AP-29, pp. 174-178, Jan. 1981.
r 101 B. B. Jones, F. Y. M. Chow, and A. W.Seeto, “The synthesis of shaped
p=1 patterns with series-fed microstrip patch arrays,” ZEEE Trans. Anten-
nas Propagat., vol. AP-30, pp. 1206-1212, Nov. 1982.
where cn is thearraycurrentduetoexcitation at port n, the E. H. Newman and P. Tulyathn, “Analysis of microstrip antennas using
moment methods,” ZEEE Trans. Antennas Propagat., vol. AP-29, pp.
g p are a known set of N’ basis or expansion functions, and the 47-53, Jan. 1981.
cnP are a set of A“ unknown coefficients. Notethatthesub- H. J. Carlin and A. B. Giordano, Network Theory. Englewood Cliffs,
scripts p and q ( p , q = 1, 2, --,N’ > N) are exclusively used t o NJ: Prentice Hall, 1964, pp. 144-146.
J. R. Mautz and R. F. Harrington, “Modal analysis of n-port
reference t h e p or qth mode in thegeneral expansion of (28). scatterers,” ZEEE Trans. Antennas Propagat., vol. AP-21, pp. 188-
At this point, most of the results of Section 11-A apply if we 199, Mar. 1973.
make the substitutions k + p , I + 4, and N -+ N’. In particular, r141 D. M. Pozar and E. H.Newman, “Analysis of a monopole mounted near
or at the edge of half-plane,’’ IEEE Trans. Antennas Propagat., vol.
(6)-(8) represent a system of N ’ simultaneous equations which AP-29, pp. 488-495, May 1981.
could be solved for the cnP(n = 1, 2, -., 2N and p = 1, 2, .-, E.H. Newmanand D. M. Pozar, “Electromagnetic modeling of
A”) coefficients in (28). Equations (9) and (10) then give the composite wire and surface geometries,” IEEE Trans. Antennas
arrayopen-circuitimpedancematrixand voltage vector. As an Propagat., vol. AP-26, pp. 784-789, Nov. 1978.
illustration of the substitutionsdescribed above, (9) becomes
N’