214 IEEE TRANSACTIONS ON ANTENNAS

AND PROPAGATION, VOL. AP-26, NO. 2, MARCH 1978

3) The 75 percent BW element design over a quarter-hemi-
sphere maximum VSWR value of 16, and consequently, to be
practical, will require an elaborate matching network.
4) It is apparent that, with a careful aperture design, a dual
ridgerectangularwaveguideoffersapracticalsolutionfora
wide-band phased array element. This conclusion is confirmed
by the results of [ 71 .
5) The presentstudydoesnotconsiderthequestion of
polarization. A minor addition to the computer program will
yield the axial ratios and the tiltangles.
REFERENCES
[ 11 S . B. Cohn, Properties of ridge wave guide, in Proc. IRE, pp.
783-788, Aug. 1947.
[2] S. Hopfer,Thedesignofridgedwaveguides, IRE Trans.
Microwave Theory Tech., pp. 20-29, Oct. 1955.
[3] J. R.Pyle, The cutoffwavelengthof TElo modeinridged
rectangular waveguide of any aspect ratio, ZEEE Duns. Micro-
wave Theory Tech., pp. 175-183, Apr. 1966.
[4] W. J. Getsinger, Ridge waveguide field description and applica-
tion to directionalcouplers, IRE Trans. MicrowaveTheory
Tech., vol. 10, pp. 41-50, Jan. 1962.
[5] J. P.Montgomery, On thecompleteeigenvaluesolutionof
ridgedwaveguide, IEEE Trans. MicrowaveTheoryTech., pp.
547-555, June 1971.
S. S. Wang, WideAngle wide-band elements for phased arrays,
Ph.D. dissertation, Polytechnic Institute of New York, Farming-
dale, NY, June 1975.
[71 J. P. Montgomery, Ridged waveguide-phased array elements,
IEEE Trans. Antennas Propagat., vol.AP-24,pp.46-53,Jan.
1976.
G. V. Borgiotti, Modal analysis of periodic planar phased arrays
of apertures, Proc. ZEEE, vol. 56, pp. 1881-1892, Nov. 1968.
G. N. Tsandoulas, Wide-band Limitations of Waveguide Arrays,
Microwave J., pp. 49-56, Sept. 1972.
H. Altschuler and L. Goldstone, On network representations of
certain obstacles in waveguide regions, IRE Trans. Microwave
Theory Tech., pp. 212-221, Apr. 1959.
N. Marcuvitz, Waveguide Handbook. NewYork:Dover, 1959,
p. 57, pp. 218-220.
J. C. Slater, MicrowaveElectronics. Princeton,NJ: D.Van
Nostrand, 1959, pp. 80-82.
L. R. Lewisand A. Hessel,Propagationcharacteristicsof
periodicarrays of dielectricslabs, ZEEE Trans. Microwave
Theory Tech., pp. 96-104, Feb. 1972.
B. L.Diamond and G. H. Knittel: A new procedureforthe
designof a waveguide element for a phased array antenna, in
Phased Array Antennas, A. A. Olinerand G . H. Knittel,Eds.
Dedham, MA: Artech House, 1972, pp. 149-156.
B. L. Diamond, A generalized approach to analysis of infinite
planararrayantennas, Proc. IEEE, pp. 1837-1851. Nov.
1968.
G. F. Farrell and D. H. Kuhn Mutual coupling in infinite planar
arraysofrectangularwaveguidehorns IEEE Trans. Antenna
Propagat., vol. AP-16, pp. 405-414, July 1968.

The Design of Small Slot Arrays

ROBERT S. ELLIOTT, FELLOW, IEEE, AND L. A. KURT2

Alstrrrcf-The differences in mutual coupling for a centtal slot and THEORY

a peripheral slot cannot be ignored in small arrays if good patterns and
impedance are to be obtained. A theory has been developed whereby
the length and offset of every slot in the array can be determined, in
the presence of mutual coupling, for a specified aperture distribution
and impedance match. The theory enlarges on Stevensons method, and
uses a modified form of Bookers relation based on Babinets principle
to treat nonresonant longitudinal shuntslots in the broad wall of a rec-
tangular waveguide. A general relation between slot voltage and mode
voltage is developed, and then formulas are derived forthe active, self-,
and mutual admittances among slots. These formulas result in a design
procedure. Analogous treatments of inclined series slots in rectangular
guide and of strip-line-fed slots are possible. Comparisonbetween
various experiments and the theory is presented. Tests of the theory
include the resonant length of a zero offsetslot, resonant conductance
versus offset and resonant conductance versus frequency for a single
slot,andself-andmutualadmittancesfor two staggered slots. The
design and performance of a two-by-four longitudinal shunt slot array
is also described.
Manuscript received June 4, 1976;revisedMay 13, 1977. Thiswork
was supportedbytheRantecDivision,EmersonElectricCompany.
R. S . Elliott iswith theDepartmentofElectricalSciences,Uni-
versity of California, Los Angeles,CA 90024, and Consultant to the
RantecDivision,EmersonElectricCompany,Calabasas,CA91302. module, but it is important to remember that these modules can be
L. A. Kurtz is with the Rantec Division, Emerson Electric Company, arranged to form either a linear array or a planar array. Double sub-
Calabasa, CA, 9 1302. script notation could be used in the latter case.
ONSIDER the module consisting of the solid lines shown
C in Fig. 1. This is a section of rectangular waveguide Ag/2
long containing a longitudinal slot of length 22 and displace-
ment x cut in its upper broad wall. One- and two-dimensional
slotarrayscanbeconstructedbyplacingsuchmodules in
tandem and parallel positions.
The module of Fig. 1 is a two-port device, the ports being
at z = +Ag/4 if the origin is taken in thewaveguide cross section
which bisects the slot. But no loss in generality occurs if the
ports are taken at the positions z = &Ag, shown dotted in Fig.
1 , becauserelationsbetween thetwosets of ports involve
simple known linear transformations. It is convenient t o choose
the ports at z = *Ag; with this convention adopted, theequiva-
lent circuit for the nthmodule1 is as shown in Fig. 2.
This equivalent circuit is subject to thefollowing interpreta-
tion. It is assumed that only the dominant TElo mode can
propagate in the waveguide. This mode is represented by the
voltage/current pair V , , I, at the input port( z = -Az). A load
1For notational simplicity the single indexn is used to identify this

0018-926X/78/0300-0214$00.75O 1978 IEEE

ELLIOTT AND KURTZ: SMALL SLOT ARRAYS 215
,r-__---7

9p;,/*,*
In words, (4)states that the active admittance at the terminals
. ,,
I
,I I
I
' .',' ' I

5 r - - - - - -9
I
of the nth module equals the self-admittance of the nth slot
plus a term which accounts for mutual coupling. This latter
_ _ - -- ,,2'=Ih 9 term is asummationwhich involves not only themutual
,c:
admittances between ports, but also the relative voltages at the
,,' bl ,," different ports. As the analysis develops further, it will be seen
,, I
z=O
z = -4-
that Y n A is decisive in determining the amplitude and phase of
c:------.e
.. a ,,,I .Ag
I

I
., ,
,.- 1
I
I
c
I z=-
4
the electric field in the nth slot. Since the latter is dictated by
pattern,
the
radiation
desired point focalY n A
thebecomes of
Y - - - __ - --I*.
I design. array
z=-A
It is well known that the scattering off a shunt element is
Fig. 1. Waveguide/slot module. symmetrical
and given by

""

L =
cc\y
I -A 2=0
I I
z=\
1 Y,*
B=C=---V
2Go
in which B and C are the amplitudes of the back and forward
T E l o scattered modes. In the manner of Stevenson [ I ] , one
can showthat B is related totheslot voltage Vns bythe
equation
(5)

Fig. 2. Equivalent circuit of nth module.

admittance Y n L is placed at the output port (z = +X,). This

sin -
Q (cos pz,
B=-j
[.'a E:;]
-a-
112 7rx,

admittance
(transformed
through 3hg/4)represent
what
could - cos kl,) vns (6)
the nth module "sees" looking down its branch line at all the
modules beyond, or it could be an appropriate termination, wherein k = 27r/h0,and 7) is the impedance of free space; u and
such as an open circuit. Y,A is the active admittance of the b are the interior dimensions of the rectangular waveguide.
nth slot.It is an important parameter in this analysis andits When (5) and (6) arecombined,oneobtains
meaning can be appreciated by considering the interrelations
among & the
l modules.
To account for mutual coupling, onecan write

N
XXn
sin -(cos 02, - cos kl,)
Q

in which Y m , is themutualadmittancebetweentheinput It will be seen shortlythat (7) is one of the twoprincipal design
ports m and n. Y,, is the self-admittance of port n ; that is, equations whichemerge fromthe analysis. A study of (7)
reveals thatthemode voltage andslot voltageare in phase
quadrature if Y E A / G ois pure real. In most slot array design
problems, Vns is governedby the pattern requirements and
in which Y , is the value that Y n A would assume if all other V , is a common voltage in any given branch line. Thus if all
input ports wereshort-circuited. Y , is commonly called the the V n s slot voltages are to be in phase with each other, and
self-admittance of the nthslot. all the mode voltages V , are t o be in phase with each other, it
is
Generally, the input admittance at the nth port followsthat all the active admittances Y n A should havea
common phase. A simple choice is to require that all Y n A be
pure real. But a study of (4) indicates that, if Y n A is to be
pure real, in general Y , , the self-admittance of the nth slot,
will not be pure real. In other words, when mutual coupling
is taken:intoaccount,onecannotexpecttheresonant self-

= Y,, +
m=l
x N, Vm
-Ym, = Y,
Vn
+ YnL +
m=l Vn
conductance data w l be pertinent in the
i design. Indeed, in
many practical applications, the requisite value of Y , can be
quite far off resonance.
The other principal design equation arises from linking the
performance of the slot array to that of an equivalent dipole
array via Babinet's principle. Clearly, if the usual assumption
in which the prime on the summationsign means that the term
of an infinite perfectly conducting ground plane is made, and
m = n has been excluded. It follows that
if the feeding currents of the center-fed strip dipoles match

Y,A = Y , + x v,
m=l
N I Vm
Ymn.
the slot voltages of the slots, the patterns will be essentially
the same. To get the impedance characteristics to match also,
one needs t o place a load impedance Z n L in series with the nth
21 6 IEEE TRANSACTIONS
ANTENNAS
ON AND PROPAGATION, VOL. AP-26, NO. 2, MARCH 1978

equivalentdipole to accountforthefactthattheresonant Equation (1 1) leadsto the interesting conclusion that

length of the slot is affected by its offset, whereas no corre-
sponding effect exists for the dipole. When this is done and
complex powers are equated for corresponding elements inthe
two arrays, it can be shown [ 21 that
When use is made of (lo)-( 12), it is important to remember
that Y , , Y , , are admittances associated with the mode volt-
ages in the slot array, and that, whereas Z,, is the conven-
tional mutual impedance between dipoles, Z,, is the loaded
self-impedance of the nthdipole since it containsZnL.

EXPERIMENT
If the foregoing theory is valid, the proper design of a one-
In (8), Z,* is the active impedance of the nth strip dipole, or two-dimensional longitudinal shunt slot array involves the
defined by choice of offsets and lengths for the various slots such that (7)
and (8) aresimultaneously satisfied for all values of n. One
begins by knowing thedesired aperture distribution ( Vms/Vns
Z , A = z, + z n L + 2N , Irm ,z, for the slots, or I , / I , for the equivalent dipoles) and the rela-
m=l n tive mode voltages V m / V n (these would all be the same in a
standing wave linear array, but would depend on the selection
wherein 2, is the self-impedance of the dipole, ZnL is the load of main-line/branch-line coupling coefficients in a planar array).
impedance placed in series with it, Z , , is the conventional Then knowledge of the function ZnA(xl, .-, X N , 1 1 , -a*, IN)
mutual impedance between dipoles calculable from formulas permits determination of all the lengths and offsets such that
suchasthose of BakerandLaGrone [ 3 ] , and I m / I , is the the desired aperture distribution is achieved, and such that the
aperture distribution. Thusif the pattern requirementis known individual values of Y n A / G o cause the branch line admittances
(so that Im/In is known), and if ( Z , -I- ZnL)is known as a and main line admittances to add up to give the desired match.
function of x, and I , (this relation will be deduced shortly), A keyingredientinthisprocess is t o find thefunction
then ZnA can be calculated, placed in (8), and Y n A / G o can be Z n A (x1, --,xN, 11, -*, ZN). As mentioned earlier, the mutual
determined. part of Z n A can be calculated from conventional formulas if
Equation (8) permits the interesting interpretation that the the aperture distribution is specified. Now we turn our atten-
normalizedactiveadmittance of alongitudinalshuntslot is tion tothedetermination of the self-part of Z n A , namely
equal t o Stevensons expression for the resonant normalized ( Z , -t- Z n L ) . Ifwe assume that (Zn + Z n L ) is essentially the
conductance (the factor in curly brackets) divided by the active same whether the other dipoles are present and open circuited,
impedance of the corresponding loaded dipole normalized to or absent, then (2, + ZnL)= (ZSELF+ Z L O A D ) ;that is, it
73 a. equals the loaded self-impedance of an isolated dipole (corre-
Equation (8) also applies for the case of an isolated slot, sponding t o an isolated slot). But for this case(8) becomes
in whichcase Z n A reduces t o Z D +Z L , with ZD the self-
impedance of the isolated strip dipole and Z L the load imped-
ance in series with it whose presence modelsthe reactive effects
of internal higher order mode scattering off the slot due to
its offset. This serves to point up some of the limitations of
Stevensonsoriginalexpression.Not only does it apply only
for resonant length slots, but strictly it becomes a less accurate
approximation as the slot width and/or its offset is increased.
- cos kZ,)2 sin2 %
a } .
This is because Z D is affected by the width of the strip dipole,
and Z L is affected by the offset of the slot. Regardless of the shape of the slot (rectangular, rounded ends,
Equation (8) can be partitioned [ 2 ] t o yield the first-order dumbbell, etc.), if one meaxurex YSELFIGO as a function of
results offset x and length 1, (13) can be used t o express (Z, i-ZnL)
as a function of x, and I,. This can then be used in (8) for all
aperturedistributionsandfeedingarrangements.Forrectan-
gular slots, the theoretical values of Y S E L F / G Oobtained by
the method of Khac [ 4 ] can be used in lieu of experimentally
obtained information.
It is desirable to accumulate the data on Y S E L F / G Oin the
universal form discovered by Stegen [ 51 and illustrated in Fig.
9-5 of Jasik [6].This figure shows plots of the real and imagi-
nary parts of Y S E L F / G +
O G R E S / G Oversus 1 1 1 ~The
~ ~range
.
of greatest use in the design of slot arrays is 0.95 < l / l R ~ s<
1.05 and the theoretical workof Khac [ 41 supports theassump-
\--, tion of universality in this range. Fig. 9-5 of Jasik requires his
Yn zmm companion Figs. 9-6 and 9-7, in which G R E S / G Oand 2lRES/hO
(cos 01, - cos kl,) sin -
a are plotted as functions of slot offset. When polyfits are made
ELLIOTT AND KURTZ: SMALL SLOT ARRAYS 217

0.70
to the four curves in Fig. 9-5, 6, and 7 of Jasik, (2, -I- ZnL) 0.60
can be expressed in a form easily handled by a computer. 0.50
Fig. 9-7 of Jasik leads t o a first test of the theory. Stegen 0.00
dealtwithroundendedslotsinawall0.050inthick.The
question arises as to the length of the equivalent strip dipole
2W0
0.30
W'
of rectangular contour in a wall of "zero" thickness. This can 0.20
be determined by the following argument. As the offset x 0, --f +0
the amplitudes of all the modes scattered off the slot tend to n
2

zero. With respect t o higher order mode scattering, this has the z
8 0.10
implicationfor thecomplementarydipolethatitsloading I-

impedance tends t o zero also. But in this circumstance, (13)

5 0.08
z
indicates that Z ~ E L Fshould be pure real for the dipole when 8 0.06
a
Y ~ E L Fis pure real for the slot. Tai has shown [ 71 that a strip
dipole of width w and negligible thickness is equivalent t o a - 0.05
cylindrical dipole of diameter d= w/2. Tai also provides a con- B
< 0.04
a
venient formula [ 71 for the impedance of a cylindrical dipole 0
0.02
as a functionof its length22 and its radiusa = d/2. Since Stegen
used slots 0.0625 in wide, if one places a = 0.0156 in in Tai's
formula, one can deduce that 2l,/&, = 0.464, wherein 21, is
0.01
the resonant length of the unloaded strip dipole. On the other 0 OX60 0.100 0.1500.250 0.200
hand, a study of Fig. 9-7 of Jasik reveals that Stegen's a s y m p SLOTOFFSET I N INCHES
totic value is 21,"lho = 0.483, in which 21,' is the resonant Fig. 3. Gr/G, for resonant longitudinal slot versus offset 9.375 GHz,
length of his round ended slot at zero offset. From this it fol- a = 0.900 in, I , = 0.400 in, slot width = 0.0625 in, wall thickness =
lows that f = 2Zro/2Z, = 1.04. This length adjustment factor is 0.050 in. Points are Stegen'smeasured values; curve is theoretical.
in agreement with the findings of Oliner [ 8 ] , who attributes a
2 percent correction for round ends and a2 percent correction 0.8
for walI thickness in this situation. -I THEORYI
Q
When the foregoing theory is used t o design slot arrays, the Q EXPERIMENT

procedure just describedcan be utilizedt o determine the length 0.7

adjustment factorf. Experience showsthat fis quite sensitive t o
a
the b dimension of the waveguide, as well as t o wall thickness. c)
' -0.6
A second test involves a predictionof resonant conductance c)
W'
0
versus offset for an isolated slot. Since the higher order mode z
a
scattering off this slot is nonpropagating and thus contributes 6 0.5
2
primarily to the storage of reactive energy, it seems reasonable n
z
t o assume that theload impedanceZL possesses a small resistive 8
component RL. Inpracticalcircumstances, the dipole self- e 0.4
IL!
impedance ZD has a resistive component in the neighborhood a
of 73 a, and thus one should expect that RL < RD. For a I
0.3
resonant slot X, = -XD,and in this case (8) can be approxi- z
mated by
0.2

(COS /31r - COS k1,)2 sin2 - . (14)

a 0.1
8.4 8.8 9.2 9.6 10.0
10.8 10.4
RESONANT FREQUENCY, GHz
For standard X-band guide, a frequency of 9.375 GHz, and a Fig. 4. Resonant conductance versus frequency.
length adjustment factor f = 1.04, (14) yields the solid curve
found in Fig. 3. Stegen's experimental points are shown for
comparison. the implication that if k2, remains constant in (14), thatis, the
It should be recognized that the agreementseenbetween slot length is continually adjusted as the frequency is changed
theory and experiment in Fig. 3 is not a case of adjusting a so as to maintain resonance, then for a slot of a given offset,
parameter in the theoretical formula to get curve fitting. All G J G , is a function of .frequency only because /3/k varies with
that has been done in (14) is to ignore RL and assume that the frequency. For a slot of offset 0.183 in, (14) yields the solid
equivalent dipole is resonant. A plot of the original Stevenson curve shown in Fig. 4. Stegen's experimental data points are
formula would lie 20 percent below thesolid curve of Fig. 3 at shown for comparison.
the low end, and 10 percentbelow it at the high end. Now let us consider situations involving more than just one
A third test involving an isolatedslot concerns the frequency isolated slot. As a first step, an array of two slots, one each in
dependence of resonant conductance. Stegen [ 51 found exper- two parallel waveguides, with the slots staggered longitudinally
imentally that his curve of resonant length versus offset for a aquarter ofaguidewave-length, was constructedwiththe
longitudinal shunt slot (Fig. 9-7 of Jasik) is universal in the dimensions shown in Fig. 5 and imbedded in an 8-in by 10-in
sense that if the offset remains constant, 21,"/ho also remains groundplane.Thisarray was used to testthevalidity of
essentially constant even though the frequency varies. This has (10)-(12) in the followingway.Withoneslotcoveredover
218 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. AP-26,NO. 2, MARCH 1978

When the values listed in Table I for Z l l , Z,,, and Z12 are
used in (16), computed values of Y1/Go and Y2/GO can be
INPUT OUTPUT
PORT PORT
ascertained. These values have been entered in the sixth and
NO. 2 NO. 2 seventh rows of Table I. The measured values are listed along-
rc2y-I side for comparison.
INPUT
PORT
__
L.J. OUTPUT
PORT
Finally, (12) of the theory can be put in the form
t--
1
NO. 1 NO. 1
- I

DIMENSIONS(lnches1
--
YI2-[ -.-.-
y1 y2 -G22
(17)
2 i y = 0.668 a = 0.913 Go Go Go 211222
x, = 0.167 b = 0.380
2f 0 = 0.682 t = 0.020
x: = 0.187 w = 0.094
When the various computedvalues found in Table I are inserted
in (17), the predictionis that Y12/Go = 0.153 L-3.1.
f = 8.815 GHr
Theaccuratemeasurement of Y12/Go is difficult.After
Fig. 5 . Two-slot array. some experimentation, the following procedure was adopted.
Short circuits were placed 3Ag/4beyond each slot. Slottedlines
TABLE I were put in tandem with both input ports. A variable attenu-
MEASURED AND COMPUTED DATA FOR TWO-SLOT ARRAY ator was placed before one slotted line, and a variable phase
shifter before the other. The two brancheswere fed through a
Measured Value Computed Value
Qurntity conventional T junction. But from (4),
YIIG, (Isolated) 0 . 5 2 t j 0.065

V2/Go (Isolared) 0.62 + j 0.065

211
63.93 - j7.99
7-22 69.79 - j 7.32
-16.32It follows that one can select the attenuator setting such that
+ j 4.40
212
0.58 + j 0 . 0 5 0.56.+ j 0.06
I V l / V 2 I = 1. If then the phase of Y1/Y2 is varied, YIA/Go
YlGO
and Y2A/Go will have loci which are circles of the same size,
0 . 6 4 + j 0.05 0.66 + j 0.06
YZIG, centered around Yl/Go and Y2/Go, respectively. The radius
Y12/GO 0.146 & 0.153 of thesetwoequal circles is I Y12/Go 1. The phase of the
mutualadmittancecanbedeterminedfromcorresponding
points on the twoloci.
with conducting tape, a short circuit was placed 3hg/4 beyond ThisexperimentalprocedureresultedintheSmithchart
the other slot, and a measurement was taken of its input admit- shown in Fig. 6 . From this data, theaverage measured valueof
tance. This resulted in the data shown in the first two rowsof mutual admittance was deduced to be Y12/Go =0.146L-4.2O.
Table I. A length contraction factor f = 1.03was found t o Lastly, let us consider the application of (7) and (8) to the
apply for this configuration and used to determine I , and I p . design of a two-dimensional array. The procedure can be out-
Equation (8) then gave lined as follows.

33.76 43.72 a) Select thefrequency of operationandthe waveguide

211 = z22 = (15) dimensions.
YJG, (isolated) Y2/Go (isolated) b) Deduce the length contraction factorf. This can be done
as in the earlier discussion of resonant length for zero
from which the entries in the third and fourth rows of Table I offset.
were obtained. c) Specify the slot voItage distribution needed to get the
Z l l and Z 2 2 as they appear in (15) are the loaded self- desired pattern and the sum of the active admittances
impedances of the strip dipoles equivalentto each isolated slot. desired in each branch linewaveguide.
Strictly speaking, they are not the same as the quantities one d) Solve (7) and (8) simultaneously t o give those values of
should use when other dipoles are present but open-circuited; x , and I, which satisfy the required aperture distribu-
however, at this slot spacing the approximation is a good one, tion and admittance level.
and therefore the entries for Z l l and Z 2 2 in Table I will be
When the above procedure was applied to the design of a
used in ( lo)-( 12).
two-by-fourarray, the resultswereasshownin Fig. 7. The
The calculationofmutualdipoleimpedance was made specified admittance level was
using the formulas of Baker and LaGrone [ 3 ] , and provides
the fifth row entryin Table I. X
Equation (10) predicts that if the conducting tapecovering
the second slot is removed and replaced by a short circuit Ag
2A

n= 1
Y,A/G, = 2
n=5
Y,A/G, = 2 +io.

from the center of the second slot, and then the input admit-
tance of the first slot is measured, the result should satisfy
The measured values were

_
Yl - 33.76 _
y2 -
- 43.72 8

Go Z l l -(z122/z22) Go z22 -(Z122/z11)

YnA/Go = 1.90 + j0 YnA/G0 = 1.94 +io.
n=l n=5
(16)
ELLIOTT AND KURTZ: SMALL SLOT ARRAYS 219

Fig. 6. Active admittance loci for two-slot array.

-.
.180 .144 -108 -72 36 0 36 72 108 144 180
ANGLE FROM BROADSIDE, DEGREES
-
Slot Offset x Length 21: Fig. 8. H-planepattern of two-by-fourarraydescribed inFig. 7,
No. n (1nches)n (Inches) f = 8.933 GHz.
-
1 -0.122 0.708
2 M.Oa, 0.667
3 -0.099 0.693 one side of the main beam at 30 dB) and a 52element two-
4 +0.060 0.699 dimensionalslotarraywithauniformaperturedistribution.
5 -0.060 0.699
6 x1.099 0.693 CONCLUSIONS
7 -0.060 0.667
a x1.122 0.708 A theory has been presentedwhich canaccount for the array
-
behavior of longitudinal shunt slots in terms of the character-
OFFSETS (Inches) istics of complementary dipoles. Formulas for active,self-, and
a = 0.924 mutual admittances of longitudinalslots havebeenderived.
b = 0.123 Slot arrays can be designed by choosingthe lengths and offsets
t = 0.025
w = 0.064 of individual slots suchthat (7) yields aslot voltage distribution
f = 8.930 GHz
consistent with the desired pattern, and such that(8) yields an
active admittancedistributionconsistentwiththefeedand
8 by 10 inch Grwnd Plane
match requirementsof the array.
The analysis can be repeated, in a step-by-step analog, for
the case of inclined series slots in the broadwall of rectangular
Fig. 7. Two-by-fourslotarray. waveguides. It can also be extended t o arrays of strip-line-fed
slots,
The theory has been tested experimentally in a variety of
The specified aperture distribution was uniform amplitude/
situations involving a single slot, a pair of slots, and a small
uniform phase, so the predicted pattern has a broadside beam,
two-dimensionalarray.Ingeneral,theagreementhasbeen
symmetrical sidelobes, and a13.5-dB sidelobelevel. The exper-
found tobe quite satisfactory.
imental H-plane pattern is shown in Fig. 8.
A study of the table of slotlengthsandoffsets (Fig. 7)
reveals several interesting and surprising things. First, there is a REFERENCES
2: 1 range in slot offsets. (Were one to ignore mutual coupling, [ l ] A. F. Stevenson, Theory of slotsinrectangularwaveguides, J.
or include it but ignore its variability from slot to slot, a l l off- Appl. P h y ~ .V, O ~ .19, pp. 24-38; .Ian. 1948.
sets would be the same.) Second, no slot in this array is self- (21 R. S. Elliott, Longitudinal shunt slots inrectangularwaveguide:
resonant; each slot is detuned appropriately t o make the indi- Part I, theory, Rantec, Calabasas,CA, Rantec Report No. 72022-
TN- 1.
vidual active admittance resonant. Third, there is a quadrant I/ [3] H. C. Bakerand A. H. LaGrone,Digitalcomputation of the
quadrant I11 andquadrantII/quadrant IV symmetry to the mutual impedance between thin dipoles, ZEEE Trans. Antennas
lengths and offsets, but no symmetry around the X axis nor Propagat.. vol. AP-10, pp. 172-178, Mar. 1962.
aroundthe Y axis. Thiscanbetraced to nonsymmetrical [4] T. Vu Khac, A study ofsome slot discontinuities in rectangular
effects caused by staggering the offsets. waveguides,Ph.D.dissertation,MonashUniversity,Australia,
Nov. 1974.
The range of lengths and offsets found for this two-by-four [5] R. J.Stegen,Longitudinal Shunt SlotCharacteristics, Hughes
array illustrates the general observation that small arrays pre- TechnicalMemorandumNo. 261, Culver City, CA.Nov., 1951.
sent a more difficult design problem than do large arrays. In [6] H. Jasik, Antenna Engineering Handbook. New York:McGraw-
the latter, o n l y elements near an edge see a different mutual Hill, 1961, Chapter 9.
[ 71 C. T. Tai, Tharacteristics of linear antenna elements,in Antenna
coupling environment, so achieving the proper activeadmit- Engineering Handbook, H. Jasik,Ed. New York:McCraw-Hill,
tancebecomessimpler.Further,mechanicalandelectrical 1961, Chapter 3.
tolerances ease off as the array getslarger [ 91. [8] A. A. Oliner, The impedance properties of narrow radiating dots
Though the details are not being reported here, the above in
the
broad face of rectangular
waveguide, IEEE Trans.
procedure has been used successfully t o design a 12-slot linear AntennasPropagat., vol. AP-5, pp. 4-20, Jan., 1957.
[9] R. S. Elliott, Mechanicalandelectricaltolerances for two-
arrayfora 3 0 dB side lobe level, a19-slotlineararray for dimensionalscanningantennaarrays, IRE Trans. Antennas
asymmetric side lobes (all at 20 dB except the inner three on Propagat., voL AP-6, pp. 114-120; 1958.