Computer Physics
Communications
Computer Physics Communications 68 (1991) 1942
NorthHolland
Electromagnetic coupling through apertures by the generalized
admittance approach
Roger F. Harrington and Joseph R. Mautz
Department of Electrical and Computer Engineering, Syracuse University, Syracuse, NY 13244, USA
Received 1 August 1990; in revised form 7 November 1990
This paper is a tutorial presentation of the use of the generalized admittance approach to aperture problems. An integral
equation is written for the unknown equivalent magnetic current over the aperture region. The generalized network
representation is obtained by applying the method of moments to the integral equation. The formulation is applied to the
problem of an aperture in a perfectly conducting plane, an aperture in a waveguide, and an aperture in a cavity. Computed
results for several examples are included.
1. Introduction
tocavity coupling, waveguidetowaveguide
coupling, waveguidetoexterior space coupling,
The problem of transmission of electromagnetic energy from one region of space to another
region through an aperture has received considerable attention over the years. One of the first
papers to use the concept of equivalent magnetic
currents and charges is that of Bethe [1], who
treated a small circular aperture in a perfectly
conducting plane. An error was made in the
determination of the equivalent magnetic current,
or tangential electric field, in the aperture, which
was later corrected by Bouwkamp [21. Since that
time many papers have appeared on the subject
of apertures in perfect electric conductors. Over
100 of these are referenced in ref. [3], and many
since then. The generalized network approach
additional papers on apertures have appeared
was introduced in ref. [4], and used extensively by
researchers at Syracuse University [5271.It has
also been used elsewhere, but a complete list of
those references is not available.
In this paper we consider the general problem
of coupling electromagnetic energy between two
or more regions of space isolated except for one
or more apertures. Some examples are cavity
and so on. The general problem consists of two or
more regions of space coupled by one or more
apertures. There can be sources in one or more
regions, and material bodies in one or more regions. Figure 1 shows a typical problem of two
regions coupled by an aperture. For this example,
it is assumed that sources exist in region a, and a
material body exists in region b. The boundary of
the regions is considered to be a perfect electric
conductor, but other types of impermeable
00104655/91/$03.50 1991
Conductors
,_
region b
region a
~ /
( /
I
.1
sources
/
/
aperture
Fig. 1. A typical aperture coupling problem.
Elsevier Science Publishers B.V. All rights reserved
20
R. F. Harringion, J.R. Mautz
Electromagnetic coupling through apertures
boundaries may be used. Each region may be
closed, as region a is shown in fig. 1, or may be
open (extend to infinity), as region b is shown.
This section is intended to serve as a short
summary of the principal relationships and notation that we use throughout the paper. For some
readers this should be a quick review. Others, less
familiar with electromagnetic theory, should refer
to one of the modern text books [281.
For our purposes, the electromagnetic field is
expressed in terms of two vectors: E the electric field, and H the magnetic field. The source
of the field may be of two types: J the electric
current, and M the magnetic current. We use
these symbols for both volume densities and for
surface densities. It should be clear from the text
Alternatively, the field in a homogeneous region
can be represented as
1
E
v x F, H jwF VIIi,
(5)
=
where the potentials are: F the electric vector
potential, and ~r the magnetic scalar potential.
The potentials F and i~1i are said to be dual to the
potentials A and 4,.
Given an electric current J and its associated
electric charge q in an infinite homogeneous
medium characterized by ~ and e, the potential
integrals are
or equations which density is meant.
In any region of space for which E and H are
differentiable, the field satisfies the Maxwell curl
equations
VXH=jweE+J,
VXEjw1riH+M,
q,
V pH
.
m,
(2)
VM= jwm.
(3)
Here the symbols q and m denote volume densities, but we use the same symbols for surface
densities. It should be clear from the text and
equations which density is meant,
It is often convenient to express the electromagnetic fields in terms of auxiliary functions,
called potentials. In a homogeneous region we
have the representation
1
HVXA,
(4)
p
where the potentials are: A the magnetic vector
potential, and 4,
the electric scalar potential.
E=jwAV4,,
1
_fffq(r)G(r,
r)
dT,
(6)
r) dT,
where the freespace Greens function G is
_________
G(r, r)
where V is the divergence operator and the two
scalars are: q the electric charge density, and m
the magnetic charge density. These are related
to the electric and magnetic currents by the equations of continuity
V~J=jwq,
4,(r)
(1)
where V x is the curl operator. The divergence of
eqs. (1) yields the Maxwell divergence equations
V
A(r) =,ifffJ(r)G(r,
4rrlI
(7)
In
(6) and
the radius
radius vector
vector to
to the
fieldeqs.point
and (7),
r ris isthe
the
source p int. The sources J and q are considered
to be vo me densities in eq. (6), as evidenced by
the triple integral sign with dT representing a
volume element of integration. If the current and
charge are surface densities, it is necessary to
change to a surface integration. The electromagnetic field associated with A and 4, is given by
eq. (4). The charge q is related to the current J
by the equation of continuity (3).
For the dual case of a magnetic current M and
its associated magnetic charge m in an infinite
homogeneous medium characterized by e and p,
the potential integrals are
F(r)
eJJfM(r)G(r,
~ ~
r) dr,
( r ) G ( r, r)
(8)
dT
p
where the Greens function G is again given by
eq. (7). The electromagnetic field associated with
F and i/i is given by eq. (5). The charge m is
R.F. Harrington, J.R. Mautz
Electromagnetic coupling through apertures
related to the current M by the equation of
continuity (3).
If the problem of interest is twodimensional,
that is, if the sources are infinite in extent in one
rectangular direction and independent of that
direction, the Greens function must be changed
to
1
G(p, p)
2~(k
~H~
p p
I).
(9)
Here p is the twodimensional radius vector to
the field point and p is the twodimensional
radius vector to the source point. H~2~
is the
zeroorder Hankel function of the second kind,
equal to J
0 jN0, where J0 is the zeroorder
Bessel function of the first kind and N0 is the
zeroorder Bessel function of the second kind (or
Neumann function). The potential integrals (6)
and (8) must now be surface integrals in the
crosssectional surface transverse to the direction
of invariance of the current.
There may be discontinuities in the electromagnetic field at sheets of current and at surface
discontinuities of the constitutive parameters. In
fig. 2, let S represent a surface between two
regions, a and b, and let n be the unit vector
normal to S pointing into region a. There may be
an electric surface J and/or a magnetic surface
current M on S. (In figures, we use a singleheaded arrow to denote electric current and a
doubleheaded arrow to denote magnetic current.) The tangential components of the field ~
S then obey the conditions
J n X (H~ H b)
M= (E~E) ><
=
Ofl
5,
(10)
where superscript a on a vector denotes that it is
evaluated in region a, and similarly for the super
region b
region a
Fig. 2. A boundary surface S between regions a and b,
possibly supporting surface currents J and M.
21
script b. The field vectors in eq. (10) are evaluated on the surface 5, as implied by on S in eq.
(10), with the superscript denoting on which side
of S.
The normal components of the fields on S
obey the conditions
qn.(e~1Ea_bEE~),
m
.(paHa pbHb)
=
on
( )
where q and m are the surface densities of
electric and magnetic charges on s, ~a, pa are e,
p in region a, and ~ ~b are E, p in region b.
The q and m are related to J and M by the
equation of continuity (3), except that now V
represents the surface divergence operator.
To formulate aperture problems, we make extensive use of the equivalence principle and the
uniqueness theorem. General discussions of these
can be found in chapter 3 of ref. [28]. Our formulations are usually given in terms of integral
equations, which are discussed in refs. [29,30].
For solutions, we use the formalism and nomenclature of the method of moments, which is discussed in refs. [29,31].
2. The generalized network formulation
This section considers a general formulation
for a tworegion aperture problem. First an operator equation is obtained in terms of an unknown
equivalent magnetic current, and this is then reduced to a matrix equation via the method of
moments. The only coupling is through the aperture, whose characteristics are expressed by aperture admittance matrices, one for each region.
These admittance matrices depend only on the
region being considered, being independent of
the other region. The aperture coupling is then
expressible as the sum of the two independent
aperture admittance matrices, with source terms
related to the incident magnetic field. This result
can be interpreted in terms of generalized networks as two Nport networks connected in parallel with current sources. The resultant solution
is equivalent to an Nterm variational solution.
22
R.F. Harrington, J.R. Mautz
Electromagnetic coupling through apertures
Since the problem is divided into two mutually
exclusive parts, one can separately solve a few
canonical problems, such as apertures in conducting screens, in waveguides, and in cavities, and
CONDUCTOR
~GIONaMfl
then combine them in various permutations.
Computer programs can be developed to calculate the aperture admittance matrices for classes
of canonical problems, such as apertures of arbitrary shape in conducting planes, in square waveguides, and in rectangular cavities. Such profor designing electromagnetic networks with
aperture coupling.
Figure 3 represents the general problem of
aperture
grams
cancoupling
then serve
between
as broadtwo
andregions,
versatilecalled
tools
region a and region b. In region a there are
impressed sources Ji, Mi, and region b is assumed source free. The more general case of
treated as the superposition of two problems, one
sources in both region a and region b can be
with sources in region a only, plus one with
sources in region b only. Each region of fig. 3 is
shown to be bounded by an electric conductor,
although other types of electromagnetic isolation
may be used. Region a is shown closed and
region b is shown open to infinity, although each
region may be open or closed. The equivalence
principle, discussed in sections 35 of ref. [28] is
used to divide the problem into two equivalent
parts as shown in fig. 4. In region a, the field is
produced by the sources J, M, plus the equivalent magnetic current
M
(12)
XE
CONDUCTOR
RTURE
REG
REGION b
c~APE
Fig. 3. The general problem of two regions coupled by an
aperture.
(ai EOUIVALENCE FOR REGION
C.
CONDU
RE G ION b
~ EQUIVALENCE FOR REGION b.
Fig. 4. The original problem divided into two equivalent
problems.
over the aperture region, with the aperture coyered by an electric conductor. In region b, the
field is produced by the equivalent magnetic current M over the aperture region, with the aperture covered by an electric conductor. The fact
that the equivalent current in region b is the
negative of that in region a ensures that the
tangential
tinuous
across
component
thetoaperture.
of
electric
The
field
remaining
is conthe
boundary
aperture.
condition
be the
applied
is continuity
of
theThe
tangential
tangential
component
component
of magnetic
of magnetic
fieldfield
across
in
region a over the aperture, denoted H~,is the
sum of that
due to the impressed sources, denoted H~1,plus that due to the equivalent source
M, denoted H~(M),that is
H~ H
=
H~(M),
(13)
R.F. Harrington, J.R. Mautz
Electromagnetic coupling through apertures
Note that H and H~(M)are both computed
with a conductor covering the aperture. A similar
equation holds for region b, except that the
equivalent source M is the only source. Hence,
the tangential component of magnetic field in
region b over the aperture is
Hb
b(
Lf\
t .
fl
bf ~
t .~
23
m 1, 2,..., N. Solution of this set of linear
equations determines the coefficients J/, and the
magnetic current M according to eq. (16). Once
M is known, the fields and fieldrelated parameters may be computed by standard methods.
The above solution can be put into matrix
notation as follows: Define an aperture admit=
f14\
~~)
where H~(M)is computed with a conductor coyering the aperture. The last equality in eq. (14) is
a consequence of the linearity of the H~operator. The true solution is obtained when H~aof eq.
(13) equals H~of eq. (14), or
_Ha(M\
t~
I _Hb(M~=H
t \
~1
t~
(15~
This is the basic operator equation for determining the equivalent magnetic current M.
If eq. (15) were satisfied exactly, we would
have the true solution. We use the method of
moments to obtain an approximate solution. Define a set of expansion functions {M~,n 1,
2,..., N), and let
=
tance matrix for region a as
[ya]
[(
(20)
Wm, HIa(Mfl))]NXN
and an aperture admittance for region b as
rybl
w HbIM ~\1
I
J
I.\
in
t
n)/J NxN~
21
The
minuspositive
signs in
eqs.JV~,
(20) M~.
andD(21)
the
real parts
when
efinemake
a source
=
vector
H~
~\ m t /]Nxi
and a coefficient vector
(2
i~~=
[~~c~]NX1.
(23)
Now the matrix equation equivalent to eqs. (19) is
(16)
[ya+yb]V....J
where the coefficients J~, are to be determined.
Substitute eq. (16) into eq. (15) and use the
linearity of the H~operators to obtain
~fr~H~(M0)
Li~H~~M~)
=H.
(24)
I.
This can be interpreted in terms of generalized
networks as two networks [ya] and [yb] in parallel with the current source I , as shown in fig. 5.
The resultant voltage vector
(17)
(25)
~[ya+yb]~Fi
Next, define a symmetric product
(A, B)
ff A ~B ds
(18)
apert.
and a set of testing functions (W~, m 1,
2,..., N), which may or may not be equal to the
expansion functions. We take the symmetric
product of eq. (17) with each testing function Wm,
and use the linearity of the symmetric product to
obtain the set of equations
=
is then the vector of coefficients which gives M
according to eq. (16).
It is important to note that computation of
[~~] involves only region a, and computation of
[yb]
involves only region b. Hence, we have divided the problem into two parts, each of which
+
V1
[yaj
V2
REGION
_~VnKWm,Hta(Mn))_~Vn(Wm,Htb(Mn))
n
n
=(Wm, He),
[yb]
I 12
REGION b
VN
(19)
~I~l
Fig. 5. The generalized network interpretation of eq. (24).
24
R.F. Harrington, J.R. Mautz
Electromagnetic coupling through apertures
may be formulated independently. Once [Y] is
computed for one region, it may be combined
with {Y] for any other region, making it useful
for a wide range of problems. For example, the
same aperture admittance matrix for radiation
into halfspace would be useful for planewave
excitation of the aperture, waveguide excitation,
and cavity excitation.
3. Numbers dependent on the field (measurements)
A linear measurement is defined as a number
which depends linearly on the field. Examples of
linear measurements are components of the field
at a point, voltage along a given contour, and
current crossing a given surface. Measurements
made in region b will depend linearly only on the
equivalent current M. Measurements made in
region a will depend linearly on the impressed
sources J1, M, as well as on the equivalence
current M. We now illustrate these concepts with
a particular example.
Consider the measurement (computation) of a
component H~of magnetic field at a point rm ~n
region b. It is known that this component can be
obtained by placing a magnetic dipole Ki~at rm,
and applying the reciprocity theorem to its field
and to the original field, sections 38 of ref. [28].
The original field in region b is the field in fig.
4b. The field of the magnetic dipole, called the
adjoint field, is the field in fig. 6. Application of
the reciprocity theorem to these two cases yields
H~K1~
=
ff M~Hmds,
(26)
apert.
Here H~is the magnetic field from Kim in the
presence of a complete conductor, and Hm is the
component in the direction of Kim of the magnetic field at rm due to M in the presence of a
complete conductor. To evaluate eq. (26), substitute for M from eq. (16) and obtain
HmKlm=
(27)
Hm).
This can be written in matrix form as
(28)
HmKlm=ImV,
where
vector
fm
fm
is the transpose of a measurement
[(~i~,
(29)
Hm)]NXI.
Note that th~elements of I are similar in form
to those of I given by eq. (22), except that M~
replaces Wm. Now substitute eq. (25) into eq. (28)
to obtain
Jm[ya + yb]~Fl
(30)
If the magnetic dipole is of unit moment, then eq.
(30) gives H~at r~,directly.
Every linear measurement in region b will be
HmKlm
of the form (30). For example, if a component of
E at r~were desired, we would place an electric
dipole at r~and apply reciprocity. In general, a
to the original problem and to an adjoint probrm
Klm
REG ION b
Fig. 6. The adjoint problem for determining Hm at rm.
lem. A determination of the sources of the adjoint problem is a part of the formulation of the
problem.
linear measurement involves applying reciprocity
If a linear measurement is made in region a, it
will involve a contribution from the impressed
sources J, M added to that from the equivalent
source M. For example, instead of eq. (30) we
would have
HmKlm
H~,K1m+ 1~[Ya + yb]
[1,
(31)
R.F. Harrington, J.R. Mautz
Electromagnetic coupling through apertures
where H~ is the magnetic field from 1, M in
the presence of a complete conductor. Also, in
region a we would define the measurement vector
to be
25
(Galerkins method), then the negative of the
integral in eq. (36) is )~~,1as defined by eq. (21),
and
P~=~VV*Yb*
L.s
m n
(37)
mn
(32)
Tm=[(MHm)]
instead of eq. (29), because the equivalent source
is +M in region a in contrast to M in region b.
Note that it is the difference field H H1 in
region a (due to M) that is directly analogous to
the transmitted field H in region b (due to M).
A quadratic measurement is one which depends quadratically on the field. Examples of
quadratic measurements are components of the
Poynting vector at a point, power crossing a given
surface, and energy within a given region. A
quadratic measurement of considerable interest
is the power transmitted through the aperture,
which we now consider.
The complex power P~transmitted through the
apertureis
This can be written in matrix form as
yLu1I*r~*.
(38)
Note that this is the usual formula for power into
network [Y~]of fig. 5.
4. Apertures in plane conductors
Consider a conducting plane covering the z 0
plane except for an aperture, as shown in fig. 7.
The two regions z> 0 and z <0 are identical
half spaces, and hence their aperture admittance
matrices are the same. Therefore, we let
=
[yayb]
(39)
2[yhs],
P~=
ff EXH*
ds,
(33)
where [yi~s] denotes the aperture admittance for
the aperture opening into half space, say z > 0.
When the aperture is covered by a conductor, the
where the asterisk denotes complex conjugate.
Substituting from eq. (12), we have
z 0 plane is a complete conducting plane, and
image theory applies. The magnetic current expansion functions are on the surface of the z 0
.~
apert.
ff M H
ds.
(34)
This involves only the tangential component of
H, which in region b we denoted by H( M).
For M we use the linear combination (16) and
obtain
=
plane. Their images are equal to them and are
also on the z 0 plane, according to sections 36
of refs. [28]. The result is that {YI~s]is the admit=
apert.
H~b( )W)
k~,H~b(M~~).
(35)
tance matrix obtained using expansion functions
x
PLANE
CONDUCTOR
INCIDENT
WAVE
~~VrnV~* ff M~.H~*(M~)ds.
m
ro
MEASUREMENT
U~
Substituting this for H and eq. (16) for M into eq.
(34), we obtain
1~t
POINT
z
APERTURE
apert.
(36)
If Mm is real, the conjugate operation can be
taken outside the integral. Moreover, if M~ ~,,
=
Fig. 7. Aperture in a plane conductor.
26
R.F. Harrington, JR. Mautz
Electromagnetic coupling through apertures
2M~radiating into free space everywhere. This
problem is dual to that for the impedance matrix
of a plane conductor, a problem considered in
the literature [321.
The original excitation of the aperture is by
the impressed sources J, M in the region z <0.
The impressed field ~ used in the operator
equation (15) is the tangential magnetic field due
to J, M with the aperture covered by a conductor (fig. 4a). In this case the z 0 plane is a
complete conductor, and image theory again applies. The result is that the tangential component
of H over the z 0 plane when it is covered by a
conductor is just twice what it is for the same
sources in free space. Hence,
=
Hti
2H~,
(40)
where H~
is the tangential component of the
magnetic field over the aperture due to the
sources J, M in free space. The components of
the excitation vector F defined by eq. (22) are
now
I~, 2
=
Jf Wm
H~
ds,
(41)
apert.
where Wm is the mth testing function.
A case of special interest is that of planewave
excitation. A unit plane wave is given by
component of magnetic field at a point rm, a
magnetic dipole Ki~ is placed at the measurement point rm. This radiates in the presence of a
complete conductor over the z 0 plane, and
hence, analogous to eq. (40), we have
=
(44)
H1tm 2H~m0.
Here Jf~tmdenotes the tangential component of H
=
over the aperture from 1~7m when the z 0 plane
is
covered
by a conductor,
from
Kl~when
it radiatesand
intoHtmO
freedenotes
space. that
The
=
components of the measurement vector
fined by eq. (29) are now
where M~is the nth expansion function.
A case of special interest is that of farfield
measurement. This is obtained by a procedure
dual to that used for radiation and scattering
from conducting wires [29,311. To obtain a component of H on the radiation sphere, we take a
source Kim perpendicular to r,~and let rm ~.
At the same time we adjust K1~so that it produces a unit plane wave in the vicinity of the
origin. The required dipole moment is given by
*
..
where u1 is a unit vector specifying the direction
of W,k1 is the propagation vector of magnitude
2ir/A
and rpointing
in the vector
direction
of propagation, and
is the radius
to an
arbitrary
field point. These vectors are shown in fig. 7. The
components (41) of the planewave excitation vector are then
2
Jf
apert.
Wm
and the planewave field it produces in the vicinity of the origin is
0 Urn e_j~cm.r.
(47)
=
H~
Here u~is
a unit vector in
the direction of
H~O,
km is the propagation vector, and r is the radius
vector to an arbitrary field point. Again these
vectors are shown in fig. 7. The components (45)
u
1 e~~
ds.
fl
(46)
4irr~
(42)
P,~,=
de
(45)
apert.
Kl~
H u e~~
ff M~.H1tmOds.
I~= 2
n
Fm
(43)
The symbol
has been used for this particular
vector to distinguish it from the more general
excitation vector (41).
Similar simplifications apply to the adjoint
(measurement) problem. For the evaluation of a
of the farfield measurement vector are then
m= 2
M~~Um e_uIcm~T ds.
(48)
P,,
ff
apert.
The symbol J~tm is used for this particular measurement vector to distinguish it from the more
general measurement vector (45) The farzone
R.F. Harrington, J.R. Mautz
Electromagnetic coupling through apertures
magnetic field is now given by eq. (30) with Kl~
given by eq. (46), fm ~m, I
ii, and [Ya +
yb] given by eq. (39). Hence
=
e_j~m1~m[YI~5]_iP~.
811 Tm
characteristics of an aperture is the transmission
crosssection r. It is defined as that area for
which the incident wave contains sufficient power
to produce the radiation field Hm by omnidirectional radiation over half space. For unit incident
magnetic field, this is
I H,.0
by eq. (38), so that
(49)
The usual two radiation components H0 and H4
are obtained by orienting K1~1 in the 0 and 4,
directions, respectively.
A parameter used to express the transmission
where Re(P~)denotes the real part of P~,given
1
T=
jw
(50)
27
Re(V[Yim]*V*).
(55)
r7S cos
Note that T depends on both the direction of
incidence and on the polarization of the incident
wave.
Finally, because of symmetry about the z 0
plane, the difference field H Hi which exists in
the region z <0 is simply related to the transmitted field which exists in the region z> 0. The
difference field in the region z <0 is produced by
an equivalent current M on a plane conductor
over the z 0 plane. By image theory, it is also
the field produced in the region z <0 by the
source 2M in free space. Analogously, the transmitted field in the region z> 0 is produced by
=
the source 2M in free space. Hence, the difference field in the region z <0 and the negative of
the transmitted field in the region z > 0 are both
Substituting from eq. (49), we obtain
~ 2~2
=
3211
pm[yhs]~
(51)
Note that r depends upon the polarization and
direction of the incident wave (via ~ 1), and upon
the polarization measured and direction to the
measurement point (via i~).
Another parameter used to express the transmission characteristics of an aperture is the transmission coefficient T, defined as
T
(52)
~rans.
P10C
1~trans. is the timeaverage power transmitwhere
ted by the aperture, and P
11~~is the freespace
power incident on the aperture. For unit incident
magnetic field, the incident power is
0inc.
where
(53)
produced by the same magnetic current 2M radiating in free space.
If an admittance matrix [Y] is defined by [Y]
~ya + yb] then, according to eqs. (39) and
(21), the mnth element of [Y] is given by
=
_41( Wm, H(M0)),
(56)
termsscalar
netic
of an potential
electric vector
ii,, as potential
[33]
F,, and magH(M,,)
jwF,,
Vfr,,,
(57)
where
C
e~~~
F~=~
apert.
ffM~ rrI
~/~i7~
is the
intrinsic impedance of
is the
angle between k. and n. The transmitted power
is
~j
Yb)mn
where H(M~)is the magnetic field produced by
M~radiating in free space. As shown in section 1,
the magnetic field H(M0) can be expressed in
~1Scos
free space, S is the aperture area, and ~
~ya +
ff m,,
1
~
ds
IrrI
ds,
(58)
(59)
apert.
V M,,
Ptrans.Re(Pt),
(54)
m,,=
(60)
28
RE. Harrington, JR. Mautz
Electromagnetic coupling through apertures
where r and r are the vectors to the field and
source points in the aperture, respectively. Substituting eqs. (18) and (57) into eq. (56), we obtain
APERTURE
fJ W~~(jwF,,+
V~,,)ds.
~mn~
(61)
~ENT
~I~REMENT
apert.
If the component of U~normal to the rim of
L~~x
the aperture vanishes on the rim of the aperture,
then
ff
Z
CONDUCTING
PLANE
(62)
V(IIInWm)dS=0.
apert.
Fig. 8. Rectangular aperture in a conducting plane.
The above equation can be rewritten as
ff W,,~
Vl/1
0 ds +
apert.
ff
~,,V Wm ds = 0.
(63)
directed magnetic currents and a set M,,~of y
directed magnetic currents defined by
apert.
t~x ,~
ft
Hence, eq. (61) becomes
Y~,,=4jw
(FnWm+~nmm) ds
(64)
Mx
p+(q~iXL~~i)
~
M~~(q_
TX(
\~YI
(66)
1)~~=UyTqY(y)P;(X),
apert.
(67)
where
m~=
where
V~Wm
.
Jw
(65)
We must now consider a specification problem in
order to choose appropriate expansion and testing functions.
u~
and u~are unit vectors. 7~(x)and
T~(y) are triangle functions defined by
x(p1) i~x
TX(x)
Z~x
(pi) z~xx p ~x,
(p t 1) ~x x
5. The rectangular aperture
0,
The geometry and coordinate system for the
rectangular aperture in a conducting plane is
shown in fig. 8. For this problem, we choose the
set of testing functions Wm equal to the set of
expansion functions M,,. The rectangular aperture 0 x L5 ~x, 0 y L~z~y,where L~and
L are integers, is divided into rectangular subar
eas of length ~ x in x and ~ y in y. The set M~
of expansion functions is split into a set M,,~of x
pz~xx(p+1)
~x,
Ix p z~xI ~x,
(q 1) LIy
~
(q
T,~(y)=
(q
1) ~ y y
1) t~y y
q z~
y,
q ~y
0,
(68)
y (q + 1) ~ y,
q z~yI
ky,
(69)
R.F. Harrington, J.R. Mautz
Electromagnetic coupling through apertures
and P(x) and P,~(y)are pulse functions defined
by
(pi) ~xx<p
P(x)={l0,
(q1)
0,
py~{1,
z~x,
all other x,
(prnu)
4~
dx 4dy,
ff
29
apert. M,
u,~,eu/~~~~05
(70)
z~yy<q~y,
(71)
all other y.
(77)
2
(prnu)
ff M:.uxe
cos4
dx dy,
apert.
(78)
The magnetic charge sheets, say m,,x and m,,~
associated with M~and M,~,are obtained from
eq. (60) as
(P(x) P~i(x))P~(y)
where u is either x or y. The superscript u is
necessary because M~has been split up into M,~
and M~of eqs. (66) and (67). In eqs. (75)(78),
u
9, u~,u4 and u~are unit vectors in the 0, y, 4,,
mp(q_I~L~_i)=
jw ~x
(72)
and x directions, respectively. As shown in fig. 8,
0
y
mp(q_i)L~ =
(P~~ Pi(y))P(x)
ito L~y
is measured from the positive x axis in the
0 plane and 4, is measured from the positive
m1~)
ytors,
axis00180,
in the x 0 plane.
For measurement
04,180.
(P,, vec0~is
for a u0 polarized measurement in the y 0
plane, ~p mu) is for a u,, polarized measure=
Introduction of the two types of expansion functions M and M,,~and the two types of testing
functions
M,~and M,~
into by
eq. (64) gives rise to
four
Y submatrices
defined
2
T
4jw
U =X,
ff (F,
~, apert.
V
X,
1X2
M~+ fr,1
1m~,)d
y.
(74)
.O~.O~4
The mathematical details and approximations for
numerically
evaluating
eq.
(74)form
can be
found
inpm
research
(43)
The
and(48)
plane
report
the
plane
wave
[34].
excitation
measurement
vector
P
vector
offor
eq.aa
of
eq.
are
of wave
the
same
except
minus sign. We therefore need to evaluate only
one of them, say the measurement vector Ppm. We
specialize it to four principal plane patterns as
1
.34
Hi
i
IS
.05
~pmu~
=
=
ffff M~
2 apert. M,~ u6 eu/tx
(P,~)~~2
u,,
cos
dx dy,
~dx dy,
~II
apert.
(76)
.1
.15
.2
.2
(C)
Fig. 9~Transmission cross section for slots of length L in the
x direction and width A /20 in the y direction. (a) L = ~A, (b)
L = ~A, (c) L ~A, (d) L = A. Excitation is by a plane wave
normally incident on the conducting plane with magnetic field
in the x direction.
30
R.F. Harrington, J.R. Mautz
ment in the y
Electromagnetic coupling through apertures
0 plane, (pmu)4 is for a u4
polarized measurement in the x 0 plane, and
(p~~) is for a u~polarized measurement in the
nent of magnetic field being considered. We use
the notation:
r0
0 plane. Because our set of testing functions
W,,, is the same as the set of expansion functions
Mm, the planewave excitation vector P of eq.
(44) is obtained by putting
1800
2rrr,~,I H0 2
in the y
0 plane,
in the x
0 plane.
(79)
2rrr,~I H~I
3600
eqs. (75)(78).
The mathematical evaluation of eqs. (75)(78)
can be found in the report [341. A computer
program, complete with operating instructions,
for computing the transmission through a rectan
We choose ~y A/20 so that q 1 in eqs. (66)
and (67). In this case, there are no y directed
magnetic current expansion functions (67). Thus,
the matrix solution (16) for the magnetic current
M is an x directed vector. As a farfield result,
H~ 0 in the y 0 plane and H4 0 in the x 0
gular aperture in a conducting plane is given in
plane. In other words, for the case being consid
part two of ref. [34].
A number of representative computations using the above matrix solution is given in the
report [34]. We summarize some of these results
here.
The first computations were made for a narrow slot, of width A /20 in the y direction and of
variable length L in the x direction. The farzone
quantity plotted was the transmission cross section, given by eq. (50), where H~ is the compo
ered, the components of H orthogonal to those
in eq. (79) are zero.
Figure 9 shows plots of T0~ and r.,.~ for x
directed slots of width A /20 and length (a) L
~A, (b) L
~A, (c) L ~A, and (d) L =A. In
all cases the excitation was due to a plane wave
normally incident on the conducting plane with
the magnetic field in the x direction. Note the
large transmission crosssection for L ~A, case
(b), due to the slot being near resonance. The
1800
3600
in the negative of one of the
90
b6O8
(a)
w6~
L.
/D~0%...\
OLIBO
I80
9O~
I
,I
tLa
l6aoo6oa6dI
I0.
~I6O 0
I80
wsJ
w~
Z
I80
.
(b)
~
180
8
/000\
~
Cc)
Cd)
Fig. 10. Magnitude and phase of M/E I, where M is the x directed magnetic current and E is the incident electric field, for the
same slots as for fig. 9. (a) L = IA, (b) L ~A,(c) L = ~A, (d) L = A. Circles denote magnitude, triangles denote phase.
R.F. Harrington, JR. Mautz
Electromagnetic coupling through apertures
plots of T are of the same form as those of the
scattering crosssection from the complementary
conducting strips, as known from Babinets principle.
Figure 10 shows plots of the equivalent mag
31
AZ
t/A2
_____
netic current in the aperture region for the same
slots. Since M u~,X E, they are also plots of the
tangential component of E in the slots. Again
note the large value of M for the case L 1A
2
which is near resonance. Note also that, for short
slots (L ~A), the M is almost equiphasal and
closely approximated by a half sine wave.
Next, computations were made to test the rate
of convergence of the solution as the number of
I~O
1.0
(a)
=..~T~,x2
2
subsections was increased. A slot of width ~A
and
length 2A iswas
chosen
for normally
the study.incident
Again
the excitation
a plane
wave
on the conducting plane with the magnetic field
in the x direction. Figure 11 shows plots of T
1.0
0~
and ri,. for the cases with (a) 39, (b) 19, (c) 9, and
(d) 4 triangular expansion functions. Note that
the patterns (a) and (b) are essentially the same,
and pattern (c) is only slightly different. They
differ appreciably from (d), which results from
only 4 expansion functions. The difference in the
solutions as the number of expansion functions is
6~
Fig. 11. Transmission crosssection when the number of expansion functions is (a) 39, (b) 19, (c) 9, and (d) 4. Computations are for a slot of length 2A in the x direction and width
A /10 in the y direction. Excitation is by a plane wave normally incident on the conducting plane with magnetic field in
the x direction.
90.
I~J
6.
Ui
___________
I
______________
Ui
(I,
4
~
6
0.4
0.8 1.2
a/A
(a)
1.6
180
2.0
90
6
Ui
0
0.4
0.8 1.2
a/A
(b)
1.6
90
~:/~~\
U,
_90e0.
0.4
0.8
I.2
a/A
(a)
1.6
180
2.0
~0.4
0.8
1.2
a/A
Cd)
180
1.6
20
Fig. 12. Magnitude and phase of I M/E I, where M is the x directed magnetic current and E is the incident electric field, when
the number of expansion functions is (a) 39, (b) 19, (c) 9, and (d) 4. Circles denote magnitude, triangles denote phase.
Computations are for the same slot as for fig. 11.
32
RE. Harrington, JR. Mautz
Electromagnetic coupling through apertures
decreased is better illustrated by plots of M, as
shown in fig. 12. These are for the same cases as
for fig. 11. It can be seen clearly how the computed equivalent current in the slot region
changes as the number of subsections is reduced.
As a rule of thumb, for nearfield quantities (such
as M) one should use subareas of length ~ or
less and for farfield quantities (such as r) length
or less.
.~PLANE CONDUCTOR
~
WAVEGUIDE
~ASUREMENT
~OINT
\......
INCfflE~~~ REGION
E
~~~ApERTURE
REGION
Fig. 13. Waveguidefed aperture in a conducting plane.
6. Waveguidefed apertures
Consider now a uniform waveguide feeding an
aperture in a conducting plane, as shown in fig.
13. In general, the aperture may be of different
size and shape than the waveguide crosssection.
The halfspace region z >0 is the same as in the
previous problem, fig. 7, and the analysis of the
preceding chapter applies. An analysis of the
waveguide region is given below,
Let the excitation of the waveguide be a source
which produces a single mode of unit amplitude,
incident on the aperture. This mode (usually the
dominant mode) is denoted by the index 0. The
field transverse to the z direction can then be
expressed in modal forms as (section 81 of ref.
[28])
F0 is the reflection coefficient for the 0th mode,
and F, is the complex amplitude of the z trayeling component of the ith mode. The e are
normalized modal vectors, so that the modal orthogonality relationships are
~0
*
e1 ~e1ds
(83)
d
1, i 3,
~UI
where the integration is over the waveguide
crosssection.
To evaluate the aperture admittance (20) for
the waveguide region, we consider a single expansion function M,, on the z 0 plane in the waveguide region. The tangential field produced by
M,, will be of the form (80), except that there is
no incident wave. Hence, this field is
E1 = e~y0ze0+ ~F1 &~ze~,
E~(M,,)
V0
Xe0
~F~Y
___________
2, A > A
a1
k1~/it (A~/A)
1,
where A is the ith mode cutoff wavelength, and
k, 2 IT/A, is the ithmode cutoff wavenumber.
The V, are the modal characteristic admittances
H~(M,,)
~A,,1
1
e~~zu~
Xe,.
It is assumed that all modes, TE and TM, are
included in the summation. The y, are modal
propagation constants
2, A <A,,
jf31 =jk~!1 (A/A1)
(81)
=
(80)
=
ff
~
(84)
>A,,1Y. e~u Xe.,
I
Z
where the A,,, are modal amplitudes. At z
have
0 we
M,, u~X E~I ~=o ~A,,1U~ X e,.
(85)
Multiply each side of this equation scalarly by
u~X e
3 and integrate over the waveguide crosssection, obtaining
=
(1 M
U X
e ds
Jy1/jw~t,
TEmodes,
TM modes.
82
( )
guide
=
LA,,1
ff
guide
(u~Xe,) .(uxe)
ds.
(86)
R.F. Harrington, JR. Mautz
Electromagnetic coupling through apertures
By orthogonality (83), all terms of the summation
are zero except the i =j term. Hence,
33
This is the H11 used in eq. (22) to evaluate the
excitation vector J I, Hence, the components of
the excitation vector are
A,,
1= JfM,~u~Xe,ds.
(87)
I,~,=2Y0
apert.
We have replaced the integral over the waveguide crosssection by one over the aperture,
since M,, exists only in the aperture region. The
elements of the aperture admittance matrix (20)
are now given by
~zwg
mn =
I
3)
iii
~
i
\ ..i
~lVI,,)
US,
/oO\
~~OO)
apert.
where the superscript wg denotes waveguide. The
HI~ of eq. (88) is given by the second equation of
eq. (84) evaluated at z 0, so that
ff WmUzXeo ds=2Y0B~0.
(93)
apert.
The total aperture admittance matrix is
rtya
[1
+1
tzbl
rvwg
~hs
I
where [y gI is the waveguide aperture admittance and [V~] is the halfspace aperture admittance. The coefficient matrix V is given by eq.
(25) with the admittance matrix given by eq. (94),
or
~7 [yw~+
(95)
yhs]~fi
LAOIYI
ff
Finally, the equivalent magnetic current M is
Wm u~Xe, ds.
(89)
given by eq. (16) where the coefficients 1, are the
components of V.
Once M is found, the modal amplitudes F,. in
eq. (80) can be evaluated from eq. (12) and the
orthogonality properties of the modes. From eqs.
(90)
(12) and (80), we have
apert.
Now define the constants
B,,,,
Jf Wm
u2 X e, ds,
M=u~XE~I~..ti=u~Xe0+
LFiu~Xe1.
apert.
which are similar in form to the A~1of eq. (87).
The elements (89) then are given by
=
(91)
LA,,IB~IYI.
Hence, each element of the waveguide aperture
admittance matrix [yw~] is a linear combination
of the modal characteristic admittances V,. For
Galerkins method, H~, M,, and A,,, and B,,1 are
equal.
We next evaluate the equivalent magnetic current M, given by eq. (16). The incident field is
given by the first term on the righthand side of
eq. (80). When the aperture is covered by a
conductor, the waveguide is terminated by a conducting plane. According to image theory, the
tangential magnetic field at z = 0 is then just
twice that of the incident wave or
=
H,=2Y0U~Xe0.
(92)
(96)
Now multiply each side scalarly by u~X ej and
integrate over the waveguide crosssection. By the
orthogonality relationships (83), all terms of the
summation vanish except the term i =j. The resuit is
Jf M u~Xe1 ds
(1 + F
F
(97)
guide
Here the integration over the guide can be
changed to that over the aperture because M 0
except in the aperture. Substituting for M from
eq. (16) into eq. (97), and using the definitions
(87), we have
=
E V,,A,,0
+ [a,
L~,,A,,1=F1, i*0.
98
34
R.F. Harrington, JR. Mautz
Electromagnetic coupling through apertures
Finally, by defining modal measurement vectors
as
where P~is given by eq. (38). Note that this gain
is a function of the H component measured, as
well as the direction to the field point.
(99)
and using eq. (95), we can write eq. (98) as
1 +F~,_A[Y~
and, for
(100)
yhs]fI
0,
(101)
+ Y~]~f.
F, ~j[yw~
7. Application to a rectangular waveguide
The parameter of most interest is F0, the reflection coefficient of the incident mode. This is
often expressed in terms of an admittance
1 To
1+F
(102)
(I
which is the equivalent aperture admittance seen
by the incident mode.
The region z> 0 for the waveguidefed aperture is the same halfspace region as existed in
the previous problem of an aperture in a conducting plane. Hence, evaluation of the fields in terms
of M in this region is done in the same way as in
section 4. For example, the u~component of the
farzone magnetic field at a point r,~is given by
1fI
(103)
4rrrm ei~fim[Y%~+yhs]~
Hm
We now apply the general theory to a rectangular waveguide feeding a rectangular aperture in
a conducting screen. Figure 14 shows the problem to be considered and defines the coordinates
and parameters to be used. The perfectly conducting plate covers the entire z = 0 plane except
for the aperture which is rectangular in shape
with side lengths L~~x and L~z~yin the x and
which is eq. (49) with the term
Y directions, respectively. L~and L~are positive
integers and L~ 2. The aperture is fed by a
rectangular waveguide. The excitation of the
waveguide is a source which produces one mode,
of unit amplitude, which travels toward the aperture.
The general method of solution discussed in
section 2 is to cover the aperture with a perfect
electric conductor, to place magnetic current
sheets +M and M respectively on the lefthand
and righthand sides of this conductor, to obtain
an integral equation for M by equating the tangential magnetic fields on both sides of this conductor, and to solve this integral equation using
the method of moments. The testing functions
2yhs]~J~~
replaced by eq. (95). The excitation vector F has
elements given by (93), and the farfield measurement vector Ii has elements given by eq. (48).
The powerintensity
gain pattern
G is direction
the ratio to
of the
radiation
in a given
radiation intensity which would exist if the total
power Re(P~)were radiated uniformly over halfspace, or
G= 2~r~IH~I2
(104)
15i~,
ii~P
_______
Re(P
1)
Substituting for H~ from eq. (103), we have
G=
2e217
811wRe( P,)
hs]1fi
2,
pm[yw~~y
Fig. 14. A rectangular waveguide radiating through a rectan
(105)
gular aperture into a halfspace bounded by an electric conductor.
R.F. Harrington, JR. Mautz
Electromagnetic coupling through apertures
are the same as the expansion functions for M
and are denoted by Mm. Each Mm is a triangle in
the direction of current flow and pulse in the
direction perpendicular to current flow.
.003
UXT;(X
1, 2,..., L~ 1, q
xi)P,,~(y
1, 2,
...~
UI
0
.00
I
I
15
.00 I
I
0.4
I
0.5
I
0.6
I
0.7
I
0.6
I
0.9
1,0
a/x
L~,
_COMPUTED}COHEN CROWLEY LEVIS
NEAS
0
COMPUTER PROGRAM
.002
Expression (87) for A,,, requires a knowledge
of the expansion functions M,, and waveguide
modes e,. The set M,, of expansion functions is
split into a set M,~of x directed magnetic currents and a set M,,~of y directed magnetic currents defined by
iXL~i)
35
Fig. 16. The equivalent aperture admittance seen by the
dominant mode for an openended square waveguide of width
a radiating into halfspace. Our computed results are
(106)
M,~+(q._i)Lo= u~T,~(y
y1)P(x x1),
pared to those calculated and measured by Cohen, Crowley
and Levis [36].
p=1,2,...,L~, q=1,2,...,L~1,
(107)
1( y) are triangle functions
where
T,1 and (69), and P,~(x)and
definedT,~(x)and
by eqs. (68)
P,~(y)are the pulse functions defined by eqs. (70)
and (71). An evaluation of the matrix [Y~] is
given in detail in ref. [35]. The elements of [yhs]
are given by the righthand side of eq. (74) with
the factor 4 replaced by 2. The listing and documentation of a computer program for the problem of fig. 14 is given in part II of the report [351.
We now give some representative computations for the aperture of fig. 14. Figure 15 shows
computed results for a rectangular waveguide of
dimensions A by ~A radiating into halfspace
through a narrow centered rectangular slot 2
of
1A
dimensions
A by A/b,
is, a = Aofand
b =
Figure
15a shows
the x that
component
equivalent
magnetic current, which is also equal to the negative of the y component of the tangential E field
in the slot. No y component of magnetic current
was obtained because only one pulse in y was
used. M is normalized with respect to
ff
V
e
0
dx dy,
(108)
guide
where the integral is over the waveguide crosssection. In other words, the normalization factor
is the rootmeansquare value of the E field of
the
phaseat of
is with All
respectincident
to that wave.
of thisThe
E field
theMaperture.
computations are for dominant TE~
0mode exci
80
/z1iIIi~
90
IET1 w441 187 m447 187 lSBT
90
~o
.~a
.~a
.~a
.~a
I80
a
b
Fig. 15. Equivalent magnetic current M~(a) and radiation gain patterns (b) for a rectangular waveguide of dimensions A by ~A
radiating through a centered rectangular slot ofdimensions A by A /10. Squares denote magnitude and triangles denote phase.
36
RE. Harrington, JR. Mautz
Electromagnetic coupling through apertures
tation. Figure 15b shows the radiation gain patterns in the two planes x = 0 and y = 0. The
notation G0~denotes the gain pattern due to H0
in the y = 0 plane. The notation ~ denotes the
gain pattern due to H~in the x 0 plane. The
horizontal axis in fig. b5b is the z axis.
Figure 16 shows a plot of the equivalent aperture admittance ~ seen by the dominant mode
of an openended square waveguide of width a
radiating into halfspace. It is defined by eq.
(102), where F11 is the reflection coefficient and
Y() is the characteristic wave admittance, both for
the dominant mode. Our computations are compared to some previously obtained by Cohen,
Crowley and Levis [36]. Also shown are measured
values reported in ref. [36]. Additional numerical
computations are given in the report [35].
=
8. Cavitybacked aperture, cylindrical cavity
An important special case of the cavitybacked
aperture is that where the cavity is a finite cylinder of arbitrary crosssection. The aperture exists
at one end of the cylinder, and the other end is
completely covered by a conductor. The cavity
can then be viewed as a shortcircuited waveguide
and waveguide theory applied. Figure 17 represents a typical problem, where the excitation is by
an incident plane wave from the halfspace region.
The halfspace part of the problem is identical
to that treated in section 4. For a unit incident
region b
LA e sin k,(
,
H =jEA.Y.(u
I
z)
(109)
sin k,d
Xe.)
Z
cos k,(dz)
sin k,d
(110)
Here e, are the normalized modal electric field
vectors (discussed in section 6), V, are the modal
characteristic admittances, and k, are the modal
wave numbers. The term sin k1d is placed in the
denominator for later convenience. The summations in eqs. (109) and (110) are assumed to be
over all modes, both TE and TM if necessary. In
particular, the modal wave numbers are
k~1(A/A.
k,
A<AIC,
)2
(111)
~C
_jk,~~/iii(A1~/A)2,
A >A,~,
l.....__ plane conductor
u1
eq. (43). For a given set of expansion and testing
functions, the aperture admittance matrix of region a remains the same as that for section 4,
denoted by [yITh]~ The new aspects of the probtern are those of determining the aperture admittance matrix for the cavity region, which we denote by [~c~V]
and the measurement vectors jm
for various desired field quantities.
The field in the cavity region can be expanded
in terms of shortcircuit waveguide modes. For a
cavity depth d, the transverse to z components of
an arbitrary field can be written as
where A.
211/k.. is the ithmode cutoff wave.
length. The modal characteristic admittances are
incident
wave
plane wave of polarization u~and propagation
vector k,, the excitation vector remains that of
conductor
aperture
Y{kI/W~,
TEmod~,
The
intrinsic wave
number k may be complex if
and/or e are complex to account for a lossy
medium in the cavity. The e, are normalized
modal vectors, so the modal orthogonality relationships are
p.
____________________
cylindrical cavity
region a

Fig. 17. A cav,ty backed aperture where the cavity is formed
by a shorted waveguide.
ffe..e.ds={
1,
S
(113)
RE. Harrington, JR. Mautz
Electromagnetic coupling through apertures
Here the surface of integration S is the cavity
(waveguide) crosssection.
To evaluate the aperture admittance (21) in
the cavity region, we consider a single expansion
function M,, on the z 0 plane inside the cavity.
The only source in the cavity region is M,,, hence
the tangential field produced by M,, is of the
form of eqs. (109) and (110). Using the boundary
condition M,, U~XE and specializing eq.
(109) to z 0, we have
37
We now define the constants
B~,
fJ W,,,
is~Xe,
ds,
(119)
apert.
which are similar in form to A,,, given by eq.
(116). Then the admittances (118) are given by
ycav. = .jLABY
cot(k,d).
(120)
u~~
LA,,1u~Xe,.
(114)
Here the additional subscript n is placed on A,,,
to denote that it is that due to M,,. Multiplying
each side of eq. (114) by u~Xe1 and integrating
over the crosssection S, we obtain
JJM,, u~X e~ds
S
LA01ff (u~Xe1) (ui Xe1) ds,
(115)
problem. This result is
By orthogonality (113), all terms of the summation are zero except the i =j term. Hence,
ff M,, u~Xe, ds.
A,,1=
(116)
apert.
We have replaced the integral over S by one over
the aperture because M,, exists only in the aperture region.
The elements of the aperture admittance matrix (21) for the cavity region are now given by
inn
ycav
Jf Wm H~(M,,)ds.
Hence, the elements of [YC~~~]
are linear combinations of the input waveguide admittances for
each shortcircuit waveguide mode.
Some specific quantities of interest in the solution are (a) the equivalent magnetic current M,
or tangential E in the aperture, (b) the amplitude
of some specific mode, and (c) the electric field
intensity at some point in the cavity. Once M is
obtained, then quantities (b) and (c) are easily
obtained.
We have the equivalent magnetic current given
by eq. (16) where J, are the elements of i~,
obtained from eq. (25) specialized to the present
(117)
apert.
~7
~C~V.]1~
(121)
where the elements of fi are given by eq. (43).
The modal amplitudes A1 are obtained from eq.
(114) as it applies to the total magnetic current in
the cavity, which is M. The result is
M
LA,u~
<i,.
(122)
Again we multiply each side by U~X e~,and integrate over the crosssection S, as in eq. (115), to
obtain
A~=
ff (M) u~Xe1 ds.
(123)
apert.
The H,b in eq. (117) is given by eq. (110) evaluated at z 0, so eq. (117) becomes
=
vcav.
mn
LA,,~Y,cot(k1d)
Substituting for M from eq. (16), and using eq.
(121), we have

A,= A~=...A[yhs+ycav.]~fli,
X
Jf
apert.
Wm u~Xe, ds.
(118)
(124)
where A is the row vector with elements given by
eq. (116). Hence, the measurement vector for
obtaining mode amplitudes is the vector A.
38
R.F. Harringlon, JR. Mautz
Electromagnetic coupling through apertures
Finally, to obtain the E1 field at any point in
the cavity, we use eq. (109) with the A, given by
eq. (124). At a frequency near resonance, that is,
where sin k~d 0 for some j, a good approxirnation to the field (109) is given by only the jth
term of eq. (109), or
sin k.~dz)
A.e.
E
1
(125)
srnk3d
If the cavity is truly loss free, then eq. (125) seems
to predict an infinite field at resonance. This,
however, is not correct, since A3 0 as sin k,,d
0 giving a finite value for E~at resonance. The
excitation of the resonant mode must then be
given by
~
A3
sin k .d
,
=
sin k .d
sin ~~O
3)
M u~X e3ds
apert.
(126)
However, the theory for this special case will not
be considered here, since it requires considerable
modification of the formulas of this section.
now of arbitrary shape. It is desired to calculate
the field E and H in the cavity region R, cornpletely enclosed by a conducting surface S, and
excited by the equivalent current M over the
aperture surface. In general, the magnetic field in
R can be represented as the sum of a curlfree
(irrotational) part H~ plus a divergencefree
df
(solenoidal) part H , that is
H=H~+Hdf,
(127)
The division (127) is not necessarily unique, since
part of the field may be both curlfree and divergencefree. For simplicity, we restrict consideration to simply bounded and simply connected
cavities.
The modal representation of the magnetic field
in a cavity bounded by a perfect electric conductor is available in textbooks [38,39]. The use of
this representation for aperture problems is summarized well in ref. [40]. The divergencefree part
of the field can be represented in terms of resonant cavity modes H1 as
ff (M) H1 ds.
H.(r)
2k2
H~jwe~ k
(128)
apert.
9. Cavitybacked aperture, arbitrary cavity
The H~are solutions to the eigenvalue problem
When the cavity backing the aperture is of
arbitrary shape, there are several ways of calculating the field in the cavity. One way is cornpletely modal. This is the method used in this
chapter.
wayinis ref.
a nonmodal
similar toAnother
that used
[37]. Thatapproach,
method,
however, will not be considered here.
Figure 18 represents the cavity part of a problem of the type of fig. 17, except that the cavity is
VX
, X
V X H1
k~H,, in R,
v X H.
(129)
(130)
1
2 are real, and only k~> 0 are
The
used eigenvalues
in eq. (128).k, The eigenvectors H,. are also
real, and orthonormal according to
jjj H
H. ds
1
on
~.
fO,
~1,
(131)
ij
The H
Conductor
1 defined above are the usual resonant
modes of timeharmonic fields in the cavity.
The curlfree part of the field can be expressed in terms of static modes as
R
region b
n
Fig. 18. Equivalence for cavities of arbitrary shape.
HCf
Jf (M) G~ds,
EG~(r)
apert.
(132)
R.F. Harrington, JR. Mautz
Electromagnetic coupling through apertures
where
G1
(133)
(134)
VtJi1,
2i4i,
in R,
=0,
on 5.
(135)
39
To obtain the aperture admittance elements
for the cavity region we use eq. (117) where Hb j~
now given by the sum of eqs. (128) and (132) with
(M) replaced by M,,. The result is
b b
~a,,
2
(141)
ycav
mn
1a,,~+jwe~ , k,~k
m,
jwp.
where
=
The modal vectors G. satisfy eqs. (129) and (130)
for k~ 0, that is, for zero eigenvalue. Hence, we
can think of k~=0 as being an eigenvalue of
infinite degeneracy. The G
1 are real and orthonormal according to
=
FM,,1
~ni~
[a~1jff[~ j G1 ds,
[ 1= If
apert.
FM,,]
apert. [wJ.~~ld5.
b,,,1
(136)
JfJG1G1 ds=
(142)
bni]
(143)
Applying the divergence theorem to ~i1H~,using
eqs. (129) and (130) to deduce that H1 has no
normal component on 5, and recalling that H1 ~
solenoidal, we have
fff 6, H1 ds
(137)
For a lossfree cavity, a term of eq. (141) can
become infinite at resonance, that is, when k k..
In the case of lossy cavities this does not occur,
although numerical problems may arise if the
cavity is only slightly lossy.
The magnetic
produced
by the nth
pansion
function field
is given
by substituting
M,, exfor
=
(M) into the sum of eqs. (128) and (132), or
for all i and j. The total field in R is simply the
sum of eqs. (128) and (132) according to eq. (127).
There
is an alternative
form
foris eqs. (132)
which
is sometimes
convenient.
This
1
b,,,H1
H(M,,)
a,,1G1 + jwe
2 k~ (144)
where a,,, and b,,~ are given byk eqs. (142) and
H~
(143). The electric field is obtained from eq. (144)
according to jwE
V X H, resulting in
~,
(138)
where
b,,,V x If~
1L~(r)ft
~=
(m)~
ds.
(139)
apert.
(145)
Note that the first summation of eq. (144) drops
out when eq. (145) is derived, since V X G
1 0.
Remembering that the source in region b of the
original problem is M, we can use the superpoE(M,,)
k2k~
Here m is the magnetic charge associated with M
according to the equation of continuity
1
VM.
Jw
(140)
The ~ are still solutions to the eigenvalue problem (134) and (135), and orthonormal according
to eq. (136) where G. is given by eq. (133). The
derivation of eq. (139) from (132) involves substitution for G, from eq. (133) and application of
the divergence theorem to i/i~M.
sition (16) to obtain the total magnetic and electric fields in the cavity as
H=LV,,H(M,,),
(146)
LI,E(M,,).
(147)
E=
Here the elements of t~are obtained from eq.
(25) in general. For the particular case of a cay
40
R.F. Harrington, JR. Mautz
Electromagnetic coupling through apertures
itybacked aperture in a conducting plane, l~is
given by eq. (121) where [Y~] now has the
elements (141). In eq. (146), the coefficient of the
jth static mode
is
Again we will not consider this special case here,
since considerable modification of the formulation is required.
a~=~LV,,a,,1
10, Discussion
3
1 V,
(148)
jwp.
where
is the row vector with eleients a,,j
given by eq. (142). Using eq. (121) for V, we have
for the amplitude of the ith static mode
=
jwp.,,
[Vhs +
(149)
jcop.
Similarly, the coefficient of the jth resonator
mode H~in eq. (146) ~
=
jw
jwe
k2 k~
k~ k~~1
(150)
The generalized network formulation of aperture problems is a very general one, being applicable to a wide variety of special cases. The
examples of this paper are some of the simpler
ones. The references [527]give a number of
additional examples. A recent application of the
theory to waveguide coupling problems is given
by ref. [41].
By specializing the theory to small apertures,
all of Bethehole theory can be described in terms
of generalized admittances [161.While we do not
have the space to go into the theory here, we do
where
is the row vector with elements b,,
1
given by eq. (143). Again using eq. (121) for ~2,we
have for the amplitude of the jth resonator mode
jwe
k~ k~~
] i
(151)
2 k~)are the
Hence, 1/jwp.vectors
and jweb~/(k
measurement
for determining a mode
amplitude.
In the vicinity of a resonant frequency in a
relatively lossfree cavity, only one term of eqs.
(146) and (147) may suffice to approximate the
field. For example, if k is near k
3, then
H~rb
3H1,
(152)
=
hs + ycav
b.VXH.
E~ .
~.
(153)
JwE
If the cavity is truly lossfree, then eq. (153)
appears to predict an infinite field at resonance.
Again this is not correct, since ~,,V,,b,,1 0 as
k k~.In this case the excitation of the resonant
mode must be determined from
,
b3
k_,
i .j~
2
(,k2
ki
(154)
wish to point out the simple relationship between
the polarizabilities of a small aperture and the
generalized admittances. Given a small aperture
in a perfectly conducting plane, Bethehole theory expresses the equivalent electric and magnetic dipoles in terms of magnetic polarizabilities
ami and a~2ae.
(2 The
components)
an electric
polarizability
halfspace and
admittance
matrix, which appears in eq. (39), is related to these
polarizabilities by
[
I
1
2j~p.a,,,~
[yhs]
01
I.I
2jcop.a~2
jwE
(155)
This result neglects radiation which contributes
to the real part of [y~]. Details of the development of eq. (155), and application to several
examples, can be found in in ref. [16].
The aperture admittance matrix is also of use
in the problem of coupling through an aperture
to objects behind the aperture. Application to the
R.F. Harrington, JR. Mautz
Electromagnetic coupling through apertures
problem of coupling to long wires can be found in
refs. [171and [24].
Specialization of the theory to twodimensional problems.results in considerable simplification of the mathematical details. Application to
transmission through slots in thick conducting
planes is given in refs. [7,9,10]. The case of a
narrow slot in a thick conductor is treated in ref.
[11]. In this reference an interesting resonance
phenomenon occurs when the conductor thickness is slightly less than a half wavelength, or this
thickness plus an integer number of half wavelengths.
The definition of characteristic modes for
aperture problems provides a way of extending
small aperture theory to higher frequencies [23].
For very small apertures, the first three characteristic modes can be identified with the three
dipole fields of Bethehole theory [27]. For larger
apertures, additional modes become impOrtant.
The specialization of characteristic modes to
twodimensional slot problems is given in refs.
[25,26].
References
[1] HA. Bethe, Theory of diffraction by small holes, Phys.
Rev. 66 (1944) 163.
[2] C.J. Bouwkamp, On Bethes theory of diffraction by
small holes, Philips Res. Rep. 5 (1950) 321.
[3]C.M. Butler, Y. RahmatSamii and R. Mittra, Electromagnetic penetration through apertures in conducting
surfaces, IEEE Trans. Antennas Propag. 26 (1978) 82.
[4] R.F. Harrington and J.R. Mautz, A generalized network
formulation for aperture problems, IEEE Trans. Antennas Propag. 24 (1976) 870.
[5] R.F. Harrington and JR. Mautz, Electromagnetic transmission through an aperture in a conducting plane, Arch.
Elektron. Ubertragungstech. 31(1977) 81.
[6] J.R. Mautz and R.F. Harrington, Transmission from a
rectangular waveguide into half space through a rectangular aperture (computer program description) IEEE
Trans. Microwave Theory Tech. 26 (1978) 44.
[7]D.T. Auckland and R.F. Harrington, Electromagnetic
transmission through a filled slit in a conducting plane of
finite thickness, TE case, IEEE Trans. Microwave Theory Tech. 26 (1978) 499.
[8] R.F. Harrington and J.R. Mautz, Computational methods for transmission of waves through apertures, in Elec
41
tromagnetic Scattering, ed. P.L.E. Usienghi (Academic,
New York, 1978) ch. 12.
[9] D.T. Auckland and R.F. Harrington, Electromagnetic
transmission through cascaded rectangular regions in a
thick conducting screen, Arch. Elektron. Ubertragungstech. 34 (1980) 19.
[10]D.T. Auckland and R.F. Harrington, A nonmodal formulation for electromagnetic transmission through a filled
slot
of Trans.
arbitrary
cross section
in a thick
screen,
IEEE
Microwave
Theory
Tech.conducting
28 (1980) 548.
[11] R.F. Harrington and D.T. Auckland, Electromagnetic
transmission through narrow slots in thick conducting
screens, IEEE Trans. Antennas Propag. 28 (1980) 616.
[12] CC. Cha and R.F. Harrington, Electromagnetic trans
mission through an annular aperture in an infinite conducting screen, Arch. Elektron. Ubertragungstech. 35
(1981) 167.
[13]E.E. Okon and R.F. Harrington, The polarizabilities of
electrically small apertures of arbitrary shape, IEEE
Trans. Electromagn. Compat. 23 (1981) 359.
[14] R.F. Harrington, Resonant behavior of a small aperture
backed by a conducting body, IEEE Trans. Antennas
Propag. 30 (1982) 205.
[15] Y. Leviatan, R.F. Harrington and J.R. Mautz, Electromagnetic transmission through apertures in a cavity in a
thick conductor, IEEE Trans. Antennas Propag. 30(1982)
1153.
[16]J.R. Mautz and R.F. Harrington, An admittance solution
for electromagnetic coupling through a small aperture,
AppI. Sci. Res. 40 (1983) 39.
[171 Yang Naiheng and R.F. Harrington, Electromagnetic
coupling to an infinite wire through a slot in a conducting
plane, IEEE Trans. Antennas Propag. 31(1983) 310.
[18] H. Nikano and R.F. Harrington, Integral equations on
electromagnetic coupling to a wire through an aperture,
Trans. IECE Japan E66 (1983) 383.
[19] H. Auda and R.F. Harrington, A moment solution for
waveguide junction problems, IEEE Trans. Microwave
Theory Tech. 31(1983) 515.
[20]E. Arvas and R.F. Harrington, Computation of the magnetic polarizability of conducting disks and the electric
polarizability of apertures, IEEE Trans. Antennas
Propag. 31(1983) 719.
[21] Y. Leviatan and R.F. Harrington, A low frequency moment solution for electromagnetic coupling through an
aperture of arbitrary shape, Arch. Elektron. Ubertragungstech. 38 (1984) 231.
[22] T.K. Sarkar, M.F. Costa, ChiLin I and R.F. Harrington,
Electromagnetic transmission through mesh covered
apertures and arrays of apertures in a conducting screen,
IEEE Trans. Antennas Propag. 32 (1984) 908.
[23]R.F. Harrington and J.R. Mautz, Characteristic modes
for aperture problems, IEEE Trans. Microwave Theory
Tech. 33 (1985) 500.
[24] S.W. Hsi, R.F. Harrington and JR. Mautz, Electromagnetic coupling to a conducting wire behind an aperture of
42
RE. IIarrington, JR. Mautz
Electromagnetic coupling through apertures
arbitrary size and shape, IEEE Trans. Antennas Propag.
33 (1985) 581.
[25] K.Y. Kabalan, R.F. Harrington, HA. Auda and JR.
Mautz, Characteristic modes for slots in a conducting
plane, TE case, IEEE Trans. Antennas Propag. 35 (1987)
162.
[26] KY. Kabalan, R.F. Harrington, JR. Mautz and HA.
Auda, Characteristic modes for a slot in a conducting
plane, TM case, IEEE Trans. Antennas Propag. 35 (1987)
331.
[271T. Wang, JR. Mautz and R.F. Harrington, Characteristic
modes and dipole representations of small apertures,
Radio Sci. 22 (1987) 1289.
[28] R.F. Harrington, TimeHarmonic Electromagnetic Fields
(McGrawHill, New York, 1961).
129] R.F. Harrington, Field Computation by Moment Methods (Macmillan, New York, 1968), reprinted (Krieger,
Malabar, FL, 1982).
[30]A.J. Poggio and E.K. Miller, Integral equation solution of
threedimensional scattering problems, in: Computer
Techniques for Electromagnetics, R. Mittra, ed. (Pergamon Press, Oxford, 1973) ch. 4.
[31] R.F. Harrington, Matrix methods for field problems,
Proc. IEEE 55 (1967) 136.
[32] N.N. Wang, J.H. Richmond and MC. Gilreath, Sinusoidal reaction formulation for radiation and scattering
from conducting surfaces, IEEE Trans. Antennas Propag.
23 (1975) 376.
[33] C.H. Papas, Theory of Electromagnetic Wave Propagalion (McGrawHill, New York, 1965) p. 23.
[34] J.R. Mautz and R.F. Harrington, Electromagnetic transmission through a rectangular aperture in a perfectly
conducting plane, Report AFCRLTR760056, Air Force
Cambridge Research Laboratories, Hanscom AFB (1976).
[35] JR. Mautz and R.F. Harrington, Transmission from a
rectangular waveguide into half space through a rectangular aperture, Report RADCTR76264, Rome Air Development Center, Griffiss AFB, NY 13441, DDC No.
ADA 030 779 (1976).
[36] M. Cohen, T. Crowley and K. Levis, The aperture admitlance of a rectangular waveguide radiating into halfspace, Antenna Lab. Rept. ac 21114 SR. No. 22, Ohio
State University, 1953.
[371D.T. Auckland and R.F. Harrington, Electromagnetic
transmission through a filled slit of arbitrary cross section
in a conducting plane of finite thickness, Report TR797,
Department of Electrical and Computer Engineering,
Syracuse University (1979).
[38] J. Van Bladel, Electromagnetic Fields (McGrawHill,
New York, 1964).
[39] K. Kurokawa, An Introduction to the Theory of Microwave Circuits, (Academic, New York, 1969).
[40] J. Van Biadel and C.M. Butler, Aperture problems, Proc.
NATO Adv. Study Inst. on Theoretical Methods for
Determining the Interaction of Electromagnetic Waves
with Structures, ed. J. Skwirzynski (Sijthoff & Noordhoff,
Aiphen aan den Run, 1981).
[41] SN. Sinha, A generalized network formulation for a class
of waveguide coupling problems, Proc. lEE pt. H 134
(1987) 502.