Progress In Electromagnetics Research, PIER 78, 219253, 2008

TIME-DOMAIN THEORY OF METAL CAVITY

RESONATOR
W. Geyi
Research In Motion
295 Phillip Street, Waterloo, Ontario, Canada N2L 3W8
AbstractThis paper presents a thorough study of the time-domain
theory of metal cavity resonators. The completeness of the vector
modal functions of a perfectly conducting metal cavity is rst proved
by symmetric operator theory, and analytic solution for the eld
distribution inside the cavity excited by an arbitrary source is then
obtained in terms of the vector modal functions. The main focus of
the present paper is the time-domain theory of a waveguide cavity,
for which the excitation problem may be reduced to the solution of a
number of modied Klein-Gordon equations. These modied Klein-
Gordon equation are then solved by the method of retarded Greens
function in order that the causality condition is satised. Numerical
examples are also presented to demonstrate the time-domain theory.
The analysis indicates that the time-domain theory is capable of
providing an exact picture for the physical process inside a closed cavity
and can overcome some serious problems that may arise in traditional
time-harmonic theory due to the lack of causality.
1. INTRODUCTION
The rapid progress in ultra-wideband technologies has prompted the
study of time-domain electromagnetics. Compared to the voluminous
literature on time-harmonic theory of electromagnetics, the time-
domain electromagnetics is still a virgin land to be cultivated. In
the time-domain theory, the elds are assumed to start at a nite
instant of time and Maxwell equations are solved subject to initial
conditions, boundary conditions, excitation conditions and causality.
A metal cavity resonator constitutes a typical eigenvalue problem in
electromagnetic theory and has been investigated by a number of
authors [e.g., 15], and the study of the transient process in a metal
220 Geyi
cavity may be carried out by the eld expansions in terms of the modal
vector functions. When these expansions are introduced into the time-
domain Maxwell equations one may nd that the expansion coecients
satisfy the ordinary dierential equations of second order [3], which
can be easily solved once the initial conditions and the excitations
are known. Recently this approach has been used to investigate the
responses of the metal cavity to digital signals [4].
Although the study of metal cavity resonator has a long history,
there are still some open questions which need to be investigated. This
paper attempts to answer these questions, and presents a thorough
discussion on the time-domain theory of metal cavities. The paper
is organized as follows. Section 2 studies the eigenvalue theory
of a perfectly conducting metal cavity lled with lossy medium.
A fundamental problem in metal cavity theory is to prove the
completeness of its vector modal functions, which has been tried by
Kurokawa [1]. But there is a loophole in Kurokawas approach in
which he fails to show the existence of the vector modal functions.
The main purpose of Section 2 is to provide a rigorous proof of the
completeness of the vector modal functions on the basis of the theory
of symmetric operators. Section 3 summarizes the eld expansions
inside an arbitrary metal cavity lled with lossy medium in terms
of the vector modal functions, and discusses the limitations of the
time-harmonic theory when it is applied to a closed metal cavity.
Section 4 is dedicated to a metal cavity formed by a section of uniform
waveguide, i.e., the waveguide cavity. The time-domain theory of the
waveguide developed previously [6] is applicable to this case, and the
elds inside the waveguide cavity can be expanded in terms of the
transverse vector modal functions of the corresponding waveguide. The
expansion coecients of the elds are shown to satisfy the modied
Klein-Gordon equation subject to homogeneous boundary conditions,
which are then solved by the method of retarded Greens function to
satisfy the causality requirement. In order to validate the theory, some
examples are expounded in Section 5, and the eld responses to typical
excitation waveforms are demonstrated.
An important observation in this paper is that the causality plays
an important role in determining the eld response of a closed cavity.
In other words, one must take the initial conditions into consideration
in order to obtain a correct eld response (transient or steady-state).
The time-domain theory shows that a sinusoidal response can be built
up if and only if the cavity is excited by sinusoidal source whose
frequency coincides with one of the resonant frequencies of the cavity.
The traditional time-harmonic theory, however, predicts that the eld
response is always sinusoidal if the excitation source is sinusoidal.
Progress In Electromagnetics Research, PIER 78, 2008 221
This discrepancy comes from the lack of causality in traditional time-
harmonic theory, which assumes that the source is turned on at t =
instead of a nite instant of time and has ignored the initial conditions.
Another interesting observation is that the eld responses in a
lossless cavity predicted by the time-harmonic theory are singular
everywhere inside the cavity if the frequency of the sinusoidal
excitation source coincides with one of the resonant frequencies of
the cavity, while the time-domain theory always gives nite eld
responses. The singularities in time-harmonic theory can be removed
by introducing losses inside the cavity, which is essentially required by
the uniqueness theorem for a time-harmonic eld in a bounded region.
2. EIGENVALUE THEORY FOR METAL CAVITY
RESONATOR
The metal cavity resonator constitutes a typical eigenvalue problem,
where the eigenvalues correspond to resonant frequencies of the
cavity and eigenfunctions correspond to the natural eld distributions.
Given the eigenvalue problem, one must show the existence and
completeness of the eigenfunctions, and the latter implies that the set
of eigenfunctions can be used to expand an arbitrary function.
2.1. Eigenvalue Problems for a Metal Cavity
Let us consider a metal cavity with a perfectly conducting wall. It
will be assumed that the medium in the cavity is homogeneous and
isotropic with medium parameters , and . The volume occupied
by the cavity is denoted by V and the boundary of V by S (Figure 1).
Since the metallic wall is a perfect conductor, the transient elds in
S
, ,
V
Figure 1. An arbitrary metal cavity.
222 Geyi
the cavity satisfy Maxwell equations
_

_
E(r, t) =

t
H(r, t)
H(r, t) =

t
E(r, t) + E, r V
E(r, t) = 0
H(r, t) = 0
(1)
with boundary conditions u
n
E = 0 and u
n
H = 0, where u
n
is the
unit outward normal to the boundary S and all other notations have
their usual meaning. From (1), one may obtain
_
_
_
E(r, t) +

2
E(r, t)
t
2
+
E(r, t)
t
= 0, r V
u
n
E(r, t) = 0, r S
(2)
_
_
_
H(r, t) +

2
H(r, t)
t
2
+
H(r, t)
t
= 0, r V
u
n
H(r, t) = 0, u
n
H(r, t) = 0, r S
(3)
If the solutions of (2) and (3) can be expressed as a separable function
of space and time
E(r, t) = e(r)u(t), H(r, t) = h(r)v(t)
it follows from (2) and (3) that
_
e k
2
e
e = 0, e = 0, r V
u
n
e = 0, r S
(4)
_
h k
2
h
h = 0, h = 0, r V
u
n
h = 0, u
n
h = 0, r S
(5)
The functions u(t) and v(t) satisfy
1
c
2

2
u
t
2
+

c
u
t
+ k
2
e
u = 0 (6)
1
c
2

2
v
t
2
+

c
v
t
+ k
2
h
v = 0 (7)
where k
2
e
and k
2
h
are separation constants, =
_
/ and c =
1/

. Both (4) and (5) form an eigenvalue problem. However their

Progress In Electromagnetics Research, PIER 78, 2008 223
eigenfunctions do not form a complete set. To overcome this diculty,
(4) and (5) can be modied as [1]
_
e e k
2
e
e = 0, r V
u
n
e = 0, e = 0, r S
(8)
_
h h k
2
h
h = 0, r V
u
n
h = 0, u
n
h = 0, r S
(9)
These are the eigenvalue equations for the metal cavity system.
2.2. Completeness of the Eigenfunctions
The eigenvalue problems (8) and (9) have been discussed by Kurokawa
[1]. Kurokawas approach suers a drawback that has overlooked the
proof of the existence of the eigenfunctions and eigenvalues. In the
following, a rigorous approach will be presented, which is similar to
what is used in studying the waveguide eigenvalue problem [6]. Let us
rewrite the eigenvalue problem (8) as
_
B(e) = e e = k
2
e
e, r V
u
n
e = 0, e = 0, r
(10)
where B = . The domain of denition of operator B is
dened by
D(B) =
_
e

e (C

())
2
, u
n
e = 0, e = 0 on
_
where C

(V ) stands for the set of functions that have continuous

partial derivatives of any order. Let L
2
(V ) stand for the space
of square-integrable functions dened in V and H = (L
2
(V ))
3
=
L
2
(V ) L
2
(V ) L
2
(V ). For two vector elds e
1
and e
2
in (L
2
(V ))
3
,
the inner product is dened by (e
1
, e
2
) =
_
V
e
1
e
2
dV (a bar is used
to designate the complex conjugate) and the corresponding norm is
denoted by = (, )
1/2
. (10) can be modied as an equivalent form
by adding a term e on both sides
_
A(e) = e e + e =
_
k
2
e
+
_
e, r V
u
n
e = e = 0, r
(11)
where is an arbitrary positive constant. It is easy to show that
the operator A is symmetric, strongly monotone. In fact, for all
224 Geyi
e
1
, e
2
D(A) = D(B), the energy product is given by
(e
1
, e
2
)
A
= (A(e
1
), e
2
) =
_
V
(e
1
e
1
+ e
1
) e
2
d
=
_
V
[e
1
e
2
+ ( e
1
)( e
2
) + e
1
e
2
] d(12)
Therefore the new operator A is symmetric. Thus e can be assumed
to be real. A is also strongly monotone since (A(e), e) e
2
.
Therefore one can introduce the energy inner product of the operator A
dened by (u, v)
A
= (Au, v). The energy space H
A
is the completion
of D(A) with respect to the norm
A
= (, )
1/2
A
[7]. Let e H
A
,
and by denition, there exists admissible sequence {e
n
D(A)} for e
such that e
n
e 0 as n and {e
n
} is a Cauchy sequence in
H
A
. It follows from (12) that
e
n
e
m

2
A
= e
n
e
m

2
+ e
n
e
m

2
+ e
n
e
m

2
Consequently {e
n
} and { e
n
} are Cauchy sequences in H. As
a result, there exist h H, and L
2
(V ) such that e
n
h and
e
n
as n . From integration by parts, one may write
_
V
e
n
dV =
_
V
e
n
dV, (C

0
(V ))
3
_
V
( e
n
)dV =
_
V
e
n
dV, C

0
(V )
Letting n yields
_
V
h dV =
_
V
e dV, (C

0
(V ))
3
_
V
dV =
_
V
e dV, C

0
(V ).
In the above C

0
(V ) is the set of all functions in C

(V ) that vanish
outside a compact subset of V . Therefore e = h and e =
hold in the generalized sense. For arbitrary e
1
, e
2
H
A
, there are two
admissible functions {e
1n
} and {e
2n
} such that e
1n
e
1
0 and
e
2n
e
2
0 as n . Dene
(e
1
, e
2
)
A
= lim
n
(e
1n
, e
2n
)
A
=
_
V
[e
1
e
2
+ ( e
1
)( e
2
) + e
1
e
2
] dV
Progress In Electromagnetics Research, PIER 78, 2008 225
where the derivatives must be understood in the generalized sense.
Now one can show that the embedding H
A
H is compact. Let
J(e) = e, e H
A
. Then the linear operator J : H
A
H is
continuous since
J(e)
2
= e
2

1
_
e
2
+e
2
+ e
2
_
=
1
e
2
H
A
.
A bounded sequence {e
n
} H
A
implies
e
n

2
H
A
= e
n

2
+e
n

2
+ e
n

2
=
_
V
_
(e
nx
)
2
+ (e
ny
)
2
+ (e
nx
)
2
+ (e
ny
)
2
_
d c

where c

is a constant. So the compactness of the operator J follows

from the above inequality and the Rellichs theorem [1, 7]. Thus the
following theorem may apply to the operator A [7, pp. 284]
Theorem: Let H be a real separable Hilbert space with dimH =
and A : D(A) H H be a linear, symmetric operator. Assume
that A is strongly monotone, i.e., there exists a constant c
1
such that
(Au, u) > c
1
u
2
for all u D(A). Let A
F
be the Friedrichs extension
of A and H
A
be the energy space of operator A. If the embedding
H
A
H is compact, then the following eigenvalue problem
A
F
u = u, u D(A
F
)
has a countable eigenfunctions {u
n
}, which form a complete
orthonormal system in the Hilbert space H, with u
n
H
A
. Each
eigenvalue corresponding to u
n
has nite multiplicity. Furthermore

1

2
, and
n

n
.
It follows from this theorem that there exists a complete set of
real eigenfunctions {e
n
}, called electric eld modal functions, and
the corresponding eigenvalues, denoted by k
2
e,n
, which approach to
innity as n . The set of modal functions will be assumed to
be orthonormal, i.e.,
_
V
e
m
e
n
dv =
mn
. It can be shown that each
modal function can be chosen from one of the following three categories
[1]:
I. e
n
= 0, e
n
= 0
II. e
n
= 0, e
n
= 0
III. e
n
= 0, e
n
= 0.
Similarly one can show that the eigenvalues k
2
h,n
of (9) are real and
positive, and the corresponding magnetic modal functions h
n
are
226 Geyi
real and constitute a complete set, which will be assumed to be
orthonormal, i.e.,
_
V
h
m
h
n
dv =
mn
. Also each modal function can
be chosen from one of the following three categories:
I. h
n
= 0, h
n
= 0
II. h
n
= 0, h
n
= 0
III. h
n
= 0, h
n
= 0.
The modal functions belonging to category II in the two sets of modal
functions {e
n
} and {h
n
} are related to each other. Actually let e
n
belong to category II. Then k
e,n
= 0 and one can dene a function h
n
through
e
n
= k
e,n
h
n
(13)
Therefore h
n
belongs to category II. Furthermore
h k
2
e,n
h = k
1
e,n

_
e k
2
e,n
e
_
= 0, r V
and
u
n
h
n
= k
1
e,n
u
n
e
n
= k
1
e,n
u
n
k
2
e,n
e
n
= 0, r S
Consider the integration of u
n
h
n
over an arbitrary part of S, denoted
S
_
S
u
n
h
n
ds = k
1
e,n
_
S
u
n
e
n
ds = k
1
e,n
_

e
n
u

d
where is the closed contour around S and u

is the unit tangent

vector along the contour. The right-hand side is zero. Thus u
n
h
n
= 0
since S is arbitrary. Therefore h
n
satises (9) and the corresponding
eigenvalue is k
2
e,n
. If h
m
is another eigenfunction corresponding to e
m
belonging to category II, then
_
V
h
m
h
n
dv = (k
e,m
k
e,n
)
1
_
V
e
m
e
n
dv
= (k
e,m
k
e,n
)
1
_
S
u
n
e
m
e
n
ds+(k
e,n
/k
e,m
)
_
V
e
m
e
n
dv
=
mn
Therefore the eigenfunctions h
n
in category II can be derived from the
eigenfunction e
n
in category II. Conversely if h
n
is in category II, one
can dene e
n
by
h
n
= k
h,n
e
n
(14)
Progress In Electromagnetics Research, PIER 78, 2008 227
and a similar discussion shows that e
n
is an eigenfunction of (4) with
k
h,n
being the eigenvalue. So the completeness of the two sets are still
guaranteed if the eigenfunctions belonging to category II in {e
n
} and
{h
n
} are related through either (13) or (14). From now on, (13) and
(14) will be assumed to hold, and k
e,n
= k
h,n
will be denoted by k
n
.
Note that the complete set {e
n
} is most appropriate for the expansion
of electric eld, and {h
n
} for the expansion of the magnetic eld.
3. TRANSIENT FIELDS IN A METAL CAVITY FILLED
WITH LOSSY MEDIUM
If the cavity contains an impressed electric current source J and a
magnetic current source J
m
, the elds excited by these sources satisfy
H(r, t) =
E(r, t)
t
+ E +J(r, t), r V
E(r, t) =
H(r, t)
t
J
m
(r, t), r V
(15)
and can be expanded in terms of the vector modal functions as
E(r, t) =

n
e
n
(r)
_
V
E(r, t) e
n
(r)dv +

v
e
v
(r)
_
V
E(r, t) e
v
(r)dv
=

n
V
n
(t)e
n
(r) +

v
V
v
(t)e
v
(r)
H(r, t) =

n
h
n
(r)
_
V
H(r, t)h
n
(r)dv+

(r)
_
V
H(r, t) h

(r)dv
=

n
I
n
(t)h
n
(r) +

(t)h

(r)
(16)
E(r, t) =

n
h
n
(r)
_
V
E(r, t) h
n
(r)dv
+

(r)
_
V
E(r, t) h

(r)dv
H(r, t) =

n
e
n
(r)
_
V
H(r, t) e
n
(r)dv
+

v
e
v
(r)
_
V
H(r, t) e
v
(r)dv
(17)
where the subscript n denotes the modes belonging to category II, and
the Greek subscript v and for the modes belonging to category I or
228 Geyi
III, and
V
n(v)
(t) =
_
V
E(r, t) e
n(v)
(r)dv
I
n()
(t) =
_
V
H(r, t) h
n()
(r)dv
(18)
Making use of the following calculations
_
V
E h
n
dv =
_
V
E h
n
dv +
_
S
(E h
n
) u
n
ds = k
n
V
n
_
V
E h

dv =
_
V
E h

dv +
_
S
(E h

) u
n
ds = 0
_
V
H e
n
ds =
_
V
H e
n
dv +
_
S
(H e
n
) u
n
ds = k
n
I
n
_
V
H e
v
ds =
_
V
H e
v
dv +
_
S
(H e
v
) u
n
ds = 0
(17) can be written as
E =

n
k
n
V
n
h
n
, H =

n
k
n
I
n
e
n
Substituting the above expansions into (15) leads to

n
k
n
I
n
e
n
=

n
e
n
V
n
t
+

v
e
v
V
v
t
+

n
e
n
V
n
+

v
e
v
V
v
+J

n
k
n
V
n
h
n
=

n
h
n
I
n
t

t
J
m
Thus
V
n
t
+

V
n

k
n

I
n
=
1

_
V
J e
n
dv
V
v
t
+

V
v
=
1

_
V
J e
v
dv
I
n
t
+
k
n

V
n
=
1

_
V
J
m
h
n
dv
I

t
=
1

_
V
J
m
h

dv
(19)
Progress In Electromagnetics Research, PIER 78, 2008 229
From the above equations one may obtain

2
I
n
t
2
+ 2
I
n
t
+
2
n
I
n
=
n
S
I
n

2
V
n
t
2
+ 2
V
n
t
+
2
n
V
n
=
n
S
V
n
(20)
where
n
= k
n
c, = /2 and
S
I
n
= c
_
V
J e
n
dv
1
k
n

t
_
V
J
m
h
n
dv
c
k
n
_
V
J
m
h
n
dv
S
V
n
=

k
n

t
_
V
J e
n
dv c
_
V
J
m
h
n
dv
To nd I
n
and V
n
, one may use the retarded Greens function dened
by
_

2
G
n
(t, t

)
t
2
+ 2
G
n
(t, t

)
t
+
2
n
G
n
(t, t

) = (t t

)
G
n
(t, t

t<t

= 0
(21)
It is easy to show that the solution of (21) is [8]
G
n
(t, t

) =
e
(tt

)
_

2
n

2
sin
_

2
n

2
(t t

)H(t t

) (22)
Therefore the general solution of I
n
is given by
I
n
(t) =

G
n
(t, t

)
n
S
I
n
(t

)dt

+
_
G
n
(t, t

)
I
n
(t

)
t

I
n
(t

)
G
n
(t, t

)
t

_
t

=
t

=
(23)
If the source is turned on at t = 0, one may let V
n
(0

) = I
n
(0

) = 0
due to causality. Considering the third equation of (19), the term in
the square bracket vanishes and the above equation reduces to
I
n
(t) =

G
n
(t, t

)
n
S
I
n
(t

)dt

230 Geyi
=

n
_

2
n

2
t
_
0

e
(tt

)
sin
_

2
n

2
(t t

_
_
c
_
V
J e
n
dv
1
k
n

_
V
J
m
h
n
dv
_
_
dt

(24)
Similarly
V
n
(t) =

G
n
(t, t

)
n
S
V
n
(t

)dt

=

n
_

2
n

2
t
_
0

e
(tt

)
sin
n
(t t

_
_

k
n

_
V
J e
n
dv c
_
V
J
m
h
n
dv
_
_
dt

(25)
Note that
V
v
(t) =
1

e
t
t
_
0

dt

e
t

_
_
_
V
J e
v
dv
_
_
I

(t) =
1

t
_
0

dt

_
V
J
m
h

dv
(26)
In order to validate the theory, let us consider a cavity excited by an
innitesimal electric and magnetic dipole at r
0
[3, pp. 538]
P = P
0
f(t)(r r
0
)
M = M
0
f(t)(r r
0
)
The elds in the cavity satisfy
H(r, t) =
0
E(r, t)
t
+P
0
f

(t)(r r
0
), r V
E(r, t) =
0
H(r, t)
t
M
0

0
f

(t)(r r
0
), r V
(27)
Comparing (27) with (15), the following identications may be made
J(r, t) = P
0
f

(t)(r r
0
), r V
J
m
(r, t) = M
0

0
f

(t)(r r
0
), r V
(28)
Progress In Electromagnetics Research, PIER 78, 2008 231
Introducing these into (24) gives
I
n
(t) = cP
0
e
n
(r
0
)
t
_
0

sin
n
(t t

)f

(t

)dt

c
0

M
0
h
n
(r
0
)
t
_
0

cos
n
(t t

)f

(t

)dt

= c
n
P
0
e
n
(r
0
)
t
_
0

cos
n
(t t

)f(t)dt

M
0
h
n
(r
0
)
_
_
f(t)
n
t
_
0

sin
n
(t t

)f(t

)dt

_
_
(29)
If the excitation waveform is sinusoidal which is turned on at t = 0,
i.e., f(t) = H(t) sint, then (29) may be written as
I
n
(t) =

2
n
[c
n
P
0
e
n
(r
0
) cos t + M
0
h
n
(r
0
) sint]
+

2
n
[c
n
P
0
e
n
(r
0
) cos t+M
0
h
n
(r
0
) sin
n
t] (30)
Similarly one can obtain the expressions of V
n
(t).
From the time-harmonic theory in which f(t) = sint, one would
obtain [3, 7.130b]
I
n
(t) =

2
n
[c
n
P
0
e
n
(r
0
) cos t + M
0
h
n
(r
0
) sint] (31)
for a lossless cavity. Thus the second term of the right-hand side of
(30) does not occur in (31). Also note that (31) is sinusoidal but (30)
is not. Furthermore (30) does not approach to (31) as t , which
contradicts our usual understanding. In fact, the response (30) based
on the time-domain analysis is not sinusoidal if the frequency of the
excitation sinusoidal waveform does not coincide with any resonant
frequencies, whereas the response (31) based on the time-harmonic
theory is always sinusoidal. Therefore the time-harmonic analysis for
a metal cavity suers a theoretical drawback that its solution does not
correspond to any practical situation in which a source is always turned
on at a nite instant of time. Apparently this theoretical drawback is
due to the lack of causality in the time-harmonic theory.
It can also be seen from (31) that I
n
(t) becomes singular when
approaches
n
, which implies that the elds are innite everywhere
232 Geyi
inside the cavity. This phenomenon is discussed in Bladels book [5]
and is compared to a lossless resonant LC network. In Collins book
[3], these singularities do not occur because of the introduction of losses
in the metal cavity. However, the time-domain solution (30) has no
singularities even for a lossless cavity. In fact (30) may be rewritten as
I
n
(t) =
_

n
t
+
n
cP
0
e
n
(r
0
) sin
+
n
2
t

2
t
+
n
M
0
h
n
(r
0
) cos
+
n
2
t
_
sin

n
2
t

n
2
t
As approaches
n
, the above becomes
I
n
(t) =

n
t
2
[cP
0
e
n
(r
0
) sin
n
t M
0
h
n
(r
0
) cos
n
t] (32)
for a nite time t, and no singularities appear in (32). The
above phenomenon can be explained by the uniqueness theorem of
electromagnetic eld. For a bounded region (such as a cavity), the
time-harmonic Maxwell equations have a unique solution if and only if
the region is lled with lossy medium, while the time-domain Maxwell
equations always have a unique solution even if the medium is lossless
[9, 10]. Therefore introducing losses in the metal cavity is required by
the time-harmonic electromagnetic theory, which guarantees that the
solution is unique and has no singularities.
Therefore the time-domain solution gives a more reasonable
picture for the physical process inside a metal cavity. More examples
in Section 5 will demonstrate this point.
4. TRANSIENT FIELDS IN A WAVEGUIDE CAVITY
FILLED WITH LOSSY MEDIUM
The evaluation of modal functions in an arbitrary metal cavity is
not an easy task. When the metal cavity consists of a section of a
uniform metal waveguide, the analysis of the transient process in the
metal cavity can be carried out by means of the time-domain theory
of waveguide [6].
4.1. Field Expansions in a Waveguide Filled with Lossy
Medium
Consider a waveguide cavity with a perfect electric wall of length L,
as shown in Figure 2. The transient electromagnetic elds inside the
Progress In Electromagnetics Research, PIER 78, 2008 233
L
2
z 1
z
n
u
z
J m
J

Figure 2. A metal cavity formed by a waveguide of cross-section .
waveguide cavity with current source J and J
m
can be expressed as
[6]
E(r, t) =

n=1
v
n
(z, t)e
tn
() +u
z

n=1
e
tn
()
k
cn
e
zn
H(r, t) =

n=1
i
n
(z, t)u
z
e
tn
() +u
z
1

u
z
H

n=1
e
tn
()
k
cn
h
zn
(33)
where = (x, y) is the position vector in the waveguide cross-section
; e
tn
are the transverse vector modal functions, and
v
n
(z, t) =
_

E e
tn
d, i
n
(z, t) =
_

H u
z
e
tn
d
h
zn
(z, t) =
_

H
_
e
tn
k
cn
_
d, e
zn
(z, t) =
_

u
z
E
_
e
tn
k
cn
_
d
Similar to the time-domain theory of waveguide lled with lossless
medium [6], the modal voltage and current for TEM mode satisfy the
one-dimensional wave equation

2
v
TEM
n
z
2

1
c
2

2
v
TEM
n
t
2

c
v
TEM
n
t
=

c

t
_

J e
tn
d

z
_

J
m
u
z
e
tn
d

2
i
TEM
n
z
2

1
c
2

2
i
TEM
n
t
2

c
i
TEM
n
t
=
_

J
m
u
z
e
tn
d

z
_

J e
tn
d+
1
c

t
_

J
m
u
z
e
tn
d (34)
234 Geyi
Once v
TEM
v
(or i
TEM
n
) is determined, i
TEM
n
(or v
TEM
n
) can be
determined by time integration of v
TEM
n
(or i
TEM
n
). The modal voltage
v
TE
n
satises the following hyperbolic equation

2
v
TE
n
z
2

1
c
2

2
v
TE
n
t
2

c
v
TE
n
t
k
2
cn
v
TE
n
=

c

t
_

J e
tn
d

z
_

J
m
u
z
e
tn
d
+k
cn
_

(u
z
J
m
)
_
u
z
e
tn
k
cn
_
d (35)
When = 0, the above equation reduces to Klein-Gordon equation.
(35) will be called modied Klein-Gordon equation. The modal current
i
TE
n
can be determined by a time integration of v
TE
n
/z
i
TE
n
(z, t) =

c
t
_

v
TE
n
(z, t

)
z
dt

c
t
_

_
_
_
_

J
m
(r, t

) [u
z
e
tn
()]d()
_
_
_
dt

(36)
The modal current i
TM
n
also satises the modied Klein-Gordon
equation.

2
i
TM
n
z
2

1
c
2

2
i
TM
n
t
2

c
i
TM
n
t
k
2
cn
i
TM
n
=
_

J
m
u
z
e
tn
d

z
_

J e
tn
d
+
1
c

t
_

J
m
u
z
e
tn
d k
cn
_

u
z
J
_
e
tn
k
cn
_
d (37)
The modal voltage v
TM
n
can then be determined by a time integration
of i
TM
n
/z
v
TM
n
(z, t) = c
t
_

i
TM
n
(z, t

)
z
dt

c
t
_

_
_
_

J(r, t

) e
tn
()d()
_
_
dt

(38)
Progress In Electromagnetics Research, PIER 78, 2008 235
4.2. Retarded Greens Function of Modied Klein-Gordon
Equation
Since the tangential electric eld on the electric conductor must be
zero, the time-domain voltage satises the homogeneous Dirichlet
boundary conditions
v
n
(z, t)|
z=z
1
= v
n
(z, t)|
z=z
2
= 0 (39)
Making use of the following relation [6]

i
n
(z, t)
z
+ k
cn
h
zn
(z, t) =
1
c
v
n
(z, t)
t
+ v
n
(z, t)
+
_

J(, z, t) e
tn
()d()
and considering the boundary condition that the normal component of
the magnetic eld on an electric conductor must be zero, the time-
domain current must satisfy the homogeneous Neumann boundary
conditions
i
n
(z, t)
z

z=z
1
=
i
n
(z, t)
z

z=z
2
= 0 (40)
In order to solve (34), (35) and (37) subject to the boundary conditions
(39) and (40), one may introduce the following retarded Greens
functions for the modied Klein-Gordon equation
_

_
_

2
z
2

1
c
2

2
t
2

t
k
2
cn
_
G
v
n
(z, t; z

, t

) = (z z

)(t t

)
G
v
n
(z, t; z

, t

)|
t<t
= 0
G
v
n
(z, t; z

, t

)|
z=z
1
= G
v
n
(z, t; z

, t

)|
z=z
2
= 0
(41)
and
_

_
_

2
z
2

1
c
2

2
t
2

t
k
2
cn
_
G
i
n
(z, t; z

, t

) = (z z

)(t t

)
G
i
n
(z, t; z

, t

)|
t<t
= 0
G
i
n
(z, t; z

, t

)
z

z=z
1
=
G
i
n
(z, t; z

, t

)
z

z=z
2
= 0
(42)
for the modal voltage and modal current respectively. Note that the
retarded Greens functions satisfy the causality condition. Taking the
236 Geyi
Fourier transform with respect to time

G
v,i
n
(z, ; z

, t

) =

G
v,i
n
(z, t; z

, t

)e
jt
dt
gives
_

2
z
2
+
2
n
_

G
v,i
n
(z, ; z

, t

) = e
jt

(z z

) (43)
where
2
n
= (/c)
2
k
2
cn
j/c. The above equations can be solved
by the method of eigenfunctions, i.e.,

G
v
n
(z, ; z

, t

) =

m=1
g
v
m
_
2
L
sin
m
L
(z z
1
)

G
i
n
(z, ; z

, t

) =

m=1
g
i
m
_

m
L
cos
m
L
(z z
1
)
where L = z
2
z
1
, and
m
= 1(m = 0),
m
= 2(m = 0). Substituting
these into (43) leads to
g
v
m
=
1

2
n
(m/L)
2
_
2
L
sin
m
L
(z

z
1
)e
jt

g
i
m
=
1

2
n
(m/L)
2
_

m
L
cos
m
L
(z

z
1
)e
jt

Thus

G
v
n
(z, ; z

, t

) =

m=1
1

2
n
(m/L)
2
2
L
sin
m
L
(zz
1
) sin
m
L
(z

z
1
)e
jt

G
i
n
(z, ; z

, t

) =

m=0
1

2
n
(m/L)
2

m
L
cos
m
L
(zz
1
) cos
m
L
(z

z
1
)e
jt

Taking the inverse Fourier transform

G
v,i
n
(z, t; z

, t

) =
1
2

G
v,i
n
(z, ; z

, t

)e
jt
d
one may obtain
G
v
n
(z, t; z

, t

) =

m=1
c
2
L
sin
m
L
(z z
1
) sin
m
L
(z

z
1
)
Progress In Electromagnetics Research, PIER 78, 2008 237

e
j(tt

)
d

2
(ck
cn
)
2
(mc/L)
2
j/
G
i
n
(z, t; z

, t

) =

m=0

m
c
2
2L
cos
m
L
(z z
1
) cos
m
L
(z

z
1
)

e
j(tt

)
d

2
(ck
cn
)
2
(mc/L)
2
j/
The integral in the summation can be evaluated by the residue theorem
[6]. The results are
G
v
n
(z, t; z

, t

) =

m=1
2c
L
sin
m
L
(z z
1
) sin
m
L
(z

z
1
)
e
(tt

)
sin
_
c(t t

)
_
k
2
cn
+ (m/L)
2

2
_
_
k
2
cn
+ (m/L)
2

2
H(t t

)
(44)
G
i
n
(z, t; z

, t

) =

m=0

m
c
L
cos
m
L
(z z
1
) cos
m
L
(z

z
1
)
e
(tt

)
sin
_
c(t t

)
_
k
2
cn
+ (m/L)
2

2
_
_
k
2
cn
+ (m/L)
2

2
H(t t

)
(45)
where = /2. If one of the ends of the waveguide cavity extends to
innity, say, z
2
, the discrete values m/L become a continuum.
In this case, (44) and (45) can be rewritten as
G
v
n
(z, t; z

, t

z
2

=
c

e
(tt

_
0
[cos k(z + z

2z
1
) cos k(z z

)]
sin
_
c(t t

)
_
k
2
cn
+ k
2

2
_
_
k
2
cn
+ k
2

2
dk
G
i
n
(z, t; z

, t

z
2

=
c

e
(tt

_
0
[cos k(z + z

2z
1
) + cos k(z z

)]
sin
_
c(t t

)
_
k
2
cn
+ k
2

2
_
_
k
2
cn
+ k
2

2
dk
238 Geyi
where k = /c. These integrations may be carried out by using

_
0
_
sinq

x
2
+ a
2

x
2
+ a
2
_
cos bxdx =

2
J
0
_
a
_
q
2
b
2
_
H(q b)
a > 0, q > 0, b > 0
and the results are
e
(tt

)
G
v
n
(z, t; z

, t

)|
z
2

c
2
J
0
_
_
k
2
cn

2
_
1/2
_
c
2
(t t

)
2
|z + z

2z
1
|
2
_
H[c(t t

) |z + z

2z
1
|]
+
c
2
J
0
_
_
k
2
cn

2
_
1/2
_
c
2
(t t

)
2
|z z

|
2
_
H[c(t t

) |z z

|]
(46)
e
(tt

)
G
i
n
(z, t; z

, t

)|
z
2

=
c
2
J
0
_
_
k
2
cn

2
_
1/2
_
c
2
(t t

)
2
|z + z

2z
1
|
2
_
H[c(t t

) |z + z

2z
1
|]
+
c
2
J
0
_
_
k
2
cn

2
_
1/2
_
c
2
(t t

)
2
|z z

|
2
_
H[c(t t

) |z z

|]
(47)
4.3. Solution of Inhomogeneous Klein-Gordon Equation
The retarded Greens functions can be used to solve the modied
Klein-Gordon equation. Consider the inhomogeneous Klein-Gordon
equations
_

2
z
2

1
c
2

2
t
2

t
k
2
cn
_
v
n
(z, t) = f(z, t), z
1
< z < z
2
_

2
z
2

1
c
2

2
t
2

t
k
2
cn
_
i
n
(z, t) = g(z, t), z
1
< z < z
2
with the boundary conditions (39) and (40). It is easy to show that
the solutions of the above equations are
Progress In Electromagnetics Research, PIER 78, 2008 239
v
n
(z, t) =
z
2
_
z
1
dz

f(z

, t

)G
v
n
(z, t; z

, t

)dt

, z (z
1
, z
2
)
i
n
(z, t) =
z
2
_
z
1
dz

g(z

, t

)G
i
n
(z, t; z

, t

)dt

, z (z
1
, z
2
)
(48)
Thus the solutions of (34), (35) and (37) can be obtained from (48) as
v
TE
n
(z, t) =

c
z
2
_
z
1
dz

_
_

J(

, z

, t

) e
tn
(

)d(

)
_
_
G
v
n
(z, t; z

, t

)dt

z
2
_
z
1
dz

_
_
_

J
m
(

, z

, t

) u
z
e
tn
(

)d(

)
_
_
G
v
n
(z, t; z

, t

)
z

dt

k
cn
z
2
_
z
1
dz

_
_
_
_

u
z
J
m
(

, z

, t

)
_
u
z
e
tn
(

)
k
cn
_
d(

)
_
_
_
G
v
n
(z, t; z

, t

)dt

(49)
i
TM
n
(z, t) =
z
2
_
z
1
dz

_
_
_

J(

, z

, t

)e
tn
(

)d(

)
_
_
G
i
n
(z, t; z

, t

)
z

dt

1
c
z
2
_
z
1
dz

_
_

J
m
(

, z

, t

) u
z
e
tn
(

)d(

)
_
_
G
i
n
(z, t; z

, t

)dt

+k
cn
z
2
_
z
1
dz

_
_
_
_

u
z
J(

, z

, t

)
_
e
tn
(

)
k
cn
_
d(

)
_
_
_
G
i
n
(z, t; z

, t

)dt

(50)
In deriving the above expressions it has been assumed that all sources
are conned in (z
1
, z
2
). It should be notied that if the magnetic
current J
m
approaches to z
1
or z
2
so that it is tightly pressed on the
electric wall z = z
1
or z = z
2
, the time-domain voltage and current
will not satisfy the homogeneous boundary conditions (39) and (40) at
z = z
1
or z = z
2
.
240 Geyi
5. APPLICATIONS
The time-domain theory developed above may be used to study the
transient process inside a cavity resonator. It will be shown that the
time-domain response of the cavity to an arbitrary excitation waveform
turned on at a nite instant of time will be severely distorted. For
example, the response of a cavity to a sinusoidal excitation turned on
at a nite instant of time is not sinusoidal in general, which is totally
dierent from the prediction of time-harmonic theory. To obtain a
sinusoidal oscillation in the cavity, the frequency of the excitation
sinusoidal wave must coincide with one of the resonant frequencies.
5.1. A Shorted Rectangular Waveguide
Let us investigate the transient process in a shorted rectangular
waveguide excited by a line current extending across the waveguide
centered at x = x
0
= a/2, z = z
0
J(r, t) = u
y
(x x
0
)(z z
0
)f(t) (51)
as shown in Figure 3. By the symmetry of the structure and excitation,
only TE
n0
mode will be excited, which are
e
tn
(x, y) = e
TE
n0
(x, y) = u
y
_
2
ab
_
1/2
sin
nx
a
, n = 1, 2, 3 (52)
with k
cn
= n/a. It follows from (46) and (49) that
v
TE
n
(z, t) =
b
2
_
2
ab
_
1/2
sin
n
a
x
0

_
t|zz
0
|/c
_

df(t

)
dt

J
0
_
k
cn
c
_
(t t

)
2
|z z
0
|
2
/c
2
_
dt

o
TE
n
v
TE
n
i
o
y
z
0
z
y
a
b
x
J
Figure 3. A shorted rectangular waveguide excited by a centered
current source.
Progress In Electromagnetics Research, PIER 78, 2008 241

t|z+z
0
|/c
_

df(t

)
dt

J
0
_
k
cn
c
_
(t t

)
2
|z + z
0
|
2
/c
2
_
dt

_
Assuming that f(t) = H(t) sint, the time-domain voltage may be
written as
v
TE
n
(z, t) =
b
2
_
2
ab
_
1/2
ka sin
n
a
x
0

_
ct/a
_
|zz
0
|/a
cos ka(ct/a u)J
0
_
k
cn
a
_
u
2
|z z
0
|
2
/a
2
_
du

ct/a
_
|z+z
0
|/a
cos ka(ct/a u)J
0
_
k
cn
a
_
u
2
|z + z
0
|
2
/a
2
_
du
_

_
The time-domain response v
TE
n
(z, t) may be divided into the sum of a
steady-state part and a transient part
v
TE
n
(z, t) = v
TE
n
(z, t)

steady
+ v
TE
n
(z, t)

transient
where
v
TE
n
(z, t)

steady
=
b
2
_
2
ab
_
1/2
ka sin
n
a
x
0

_
|zz
0
|/a
cos ka(ct/a u)J
0
_
k
cn
a
_
u
2
|z z
0
|
2
/a
2
_
du

_
|z+z
0
|/a
cos ka(ct/a u)J
0
_
k
cn
a
_
u
2
|z + z
0
|
2
/a
2
_
du
_

_
v
TE
n
(z, t)

transient
=
b
2
_
2
ab
_
1/2
ka sin
n
a
x
0

_
ct/a
cos ka(ct/a u)J
0
_
k
cn
a
_
u
2
|z z
0
|
2
/a
2
_
du

_
ct/a
cos ka(ct/a u)J
0
_
k
cn
a
_
u
2
|z + z
0
|
2
/a
2
_
du
_

_
242 Geyi
The transient part approaches to zero as t . The integrals in the
steady part can be carried out by use of the following calculations [11]

_
a
J
0
_
b
_
x
2
a
2
_
sincxdx =
_
0, 0 < c < b
cos
_
a

c
2
b
2
_
/

c
2
b
2
, 0 < b < c

_
a
J
0
_
b
_
x
2
a
2
_
cos cxdx =
_
_
_
exp
_
a

b
2
c
2
_
/

b
2
c
2
, 0 < c < b
sin
_
a

c
2
b
2
_
/

c
2
b
2
, 0 < b < c
Thus
v
TE
n
(z, t)

steady
=
b
2
_
2
ab
_
1/2
ka sin
n
a
1
_
|(ka)
2
(k
cn
a)
2
|

_
sin
_
ka
ct
a

|z z
0
|
a
_
(ka)
2
(k
cn
a)
2
_
sin
_
ka
ct
a

|z + z
0
|
a
_
(ka)
2
(k
cn
a)
2
_
, k > k
cn
cos
_
ka
ct
a
_
exp
_

|z z
0
|
a
_
(k
cn
a)
2
(ka)
2
_
cos
_
ka
ct
a
_
exp
_

|z + z
0
|
a
_
(k
cn
a)
2
(ka)
2
_
k < k
cn
In the region 0 < z < z
0
, the above equation may be rewritten as
v
TE
n
(z, t)

steady
=
1

2
_
b
a
_
1/2
sin
n
2
k

_
2 sin(
n
z) cos(t
n
z
0
), k > k
cn
cos(t) exp[
n
(z z
0
)]
cos(t) exp[
n
(z + z
0
)], k < k
cn
where
n
= (|k
2
k
2
cn
|)
1/2
. Therefore the time-domain voltage for the
TE
n0
mode in the shorted waveguide is a standing wave if the operating
frequency is higher than cut-o frequency of the TE
n0
mode, which is
a well-known result in time-harmonic theory.
The time-domain current can be determined by (36)
i
TE
n
(z, t) =
_
2b
a
_
1/2
sin
n
2

1
2
sin(t |z + z
0
|/c)
1
2
sin(t |z z
0
|/c)
_
Progress In Electromagnetics Research, PIER 78, 2008 243
+
_
2b
a
_
1/2
sin
n
2

_
k
cn
(z+z
0
)
2
t|z+z
0
|/c
_
0
J
1
_
k
cn
c
_
(t t

)
2
|z + z
0
|
2
/c
2
_
_
(t t

)
2
|z + z
0
|
2
/c
2
sint

dt

k
cn
(zz
0
)
2
t|zz
0
|/c
_
0
J
1
_
k
cn
c
_
(t t

)
2
|z z
0
|
2
/c
2
_
_
(t t

)
2
|z z
0
|
2
/c
2
sint

dt

_
The steady state part of i
TE
n
(z, t) is given by
i
TE
n
(z, t)

steady
=
_
2b
a
_
1/2
sin
n
2

1
2
sin(t |z + z
0
|/c)
1
2
sin(t |z z
0
|/c)
_
+
_
2b
a
_
1/2
sin
n
2

_
k
cn
(z+z
0
)
2

_
|z+z
0
|/c
J
1
_
k
cn
c
_
u
2
|z + z
0
|
2
/c
2
_
_
u
2
|z + z
0
|
2
/c
2
sin(t u)du

k
cn
(zz
0
)
2

_
|zz
0
|/c
J
1
_
k
cn
c
_
u
2
|z z
0
|
2
/c
2
_
_
u
2
|z z
0
|
2
/c
2
sin(t u)du
_

_
Assuming that k > k
cn
and making use of the following calculations

_
a
sincx

x
2
a
2
J
v
_
b
_
x
2
a
2
_
dx =

2
J
v/2
_
a
2
_
c
_
c
2
b
2
_
_
J
v/2
_
a
2
_
c +
_
c
2
b
2
_
_

_
a
cos cx

x
2
a
2
J
v
_
b
_
x
2
a
2
_
dx =

2
J
v/2
_
a
2
_
c
_
c
2
b
2
_
_
N
v/2
_
a
2
_
c +
_
c
2
b
2
_
_
(a > 0, 0 < b < c)
244 Geyi
one may obtain
i
TE
n
(z, t)

steady
=
1

2
_
b
a
_
1/2
2 sin
n
2
cos(
n
z) sin(t
n
z
0
)
Let V
TE
n
(z) and I
TE
n
(z) be the phasors of v
TE
n
(z, t)|
steady
and
i
TE
n
(z, t)|
steady
respectively, then
V
TE
n
(z) =
1

2
_
b
a
_
1/2
k

n
2 sin
n
2
sin(
n
z)e
jnz
0
, k > k
cn
I
TE
n
(z) = j
1

2
_
b
a
_
1/2
2 sin
n
2
cos(
n
z)e
jnz
0
, k > k
cn
Since the current is assumed to be in positive z-direction, the
impedance at z (0, z
0
) is thus given by
Z
n
(z) =
V
TE
n
(z)
I
TE
n
(z)
= j
k

n
tan(
n
z), k > k
cn
which is a well-known result and validates the time-domain theory.
5.2. A Rectangular Waveguide Cavity
Let the shorted waveguide shown in Figure 3 be closed by a perfect
conducting wall at z = L with L > z
0
(Figure 4) and the excitation
source be given by (51). By symmetry, only TE
n0
mode will be excited.
It follows from (44) and (49) that
v
TE
n
(z, t) =
2
L
_
2b
a
_
1/2
sin
nx
0
a

m=1
sin
m
L
z sin
m
L
z
0
_
(n/a)
2
+ (m/L)
2

df(t

)
dt

sin
_
c(t t

)
_
(n/a)
2
+(m/L)
2
_
H(t t

)dt

(53)
L
o
y
z
0
z
y
a
b
x
J
Figure 4. A rectangular waveguide cavity excited by a current source.
Progress In Electromagnetics Research, PIER 78, 2008 245
Again if it is assumed that f(t) = H(t) sint, the above expression
may be written as
v
TE
n
(z, t) =
2
L
_
2b
a
_
1/2
sin
nx
0
a

m=1
sin
m
L
z sin
m
L
z
0
_
(n/a)
2
+ (m/L)
2

t
_

sin
_
c(t t

)
_
(n/a)
2
+(m/L)
2
_
cos t

dt

=
2k
L
_
2b
a
_
1/2
sin
nx
0
a

m=1
sin
m
L
z sin
m
L
z
0

cos kct cos

_
ct
_
(n/a)
2
+(m/L)
2
_
k
2
(n/a)
2
+(m/L)
2
(54)
where k = /c. The electromagnetic eld is given by (33)
E(r, t) = u
y
E
y
=

n=1
v
TE
n
(z, t)e
TE
n0
(x, y)
= u
y
4k
La

n=1

m=1
sin
nx
a
sin
nx
0
a
sin
m
L
z sin
m
L
z
0

cos kct cos

_
ct
_
(n/a)
2
+(m/L)
2
_
k
2
(n/a)
2
+(m/L)
2
(55)
0 10 20 30 40 50 60
0.5
0
0.5
ctoa
y
aE

0 0
/ ( 0.5 , 0.75 , 0.5 ) ct a x a z a x z a = = = =
Figure 5. Electric eld excited by a sinusoidal wave f(t) = H(t) sint
when k = /c =
_
(n/a)
2
+ (m/L)
2
(ka = ).
246 Geyi
where e
TE
n0
(x, y) are given by (52). As can be seen from (54) and
(55), one cannot separate the response of a closed cavity into a
transient part and a steady-state part. Figure 5 shows the normalized
electric eld aE(r, t)/ at x = 0.5a, z = 0.75a, excited by a source
dened by (51) with sinusoidal waveform f(t) = H(t) sint with
a = b = L. It is assumed that k is below any resonant wavenumbers
_
(n/a)
2
+ (m/L)
2
(m, n 1) with k = . It can be seen from
the plot that the electric eld does not approach to a pure sinusoidal
wave as t . It should be noted that (54), (55) are nite as k
approaches to any of the resonant wavenumbers
_
(n/a)
2
+ (m/L)
2
.
For example, when k approaches to
_
(/a)
2
+ (/L)
2
, the singular
term for (n, m) = (1, 1) in (55) becomes
cos kct cos
_
ct
_
(/a)
2
+ (/L)
2
_
k
2
(/a)
2
+ (/L)
2
=
ct sin
_
ct
_
(/a)
2
+ (/L)
2
_
2
_
(/a)
2
+ (/L)
2
(56)
which is a nite number for a nite t.
0 10 20 30 40 50 60
40
20
0
20
40
E1 ctoa ( )
ctoa
y
aE

0 0
/ ( 0.5 , 0.75 , 0.5 ) ct a x a z a x z a = = = =
Figure 6. Electric eld exited by a sinusoidal wave f(t) = H(t) sint
when k = /c =
_
(/a)
2
+ (/L)
2
.
Figure 6 shows the normalized electric eld aE(r, t)/ at x =
0.5a, z = 0.75a, excited by a sinusoidal wave f(t) = H(t) sint when
k =
_
(/a)
2
+ (/L)
2
with a = b = L. In this case a sinusoidal wave
will be gradually built up as t . Therefore the response of a metal
cavity is a sinusoidal wave if and only if the frequency of the exciting
sinusoidal wave coincides with one of the resonant frequencies of the
metal cavity.
Progress In Electromagnetics Research, PIER 78, 2008 247
If the excitation waveform is a unit step function, i.e., f(t) = H(t),
(53) becomes
v
TE
n
(z, t) =
2
L
_
2b
a
_
1/2
sin
nx
0
a

m=1
sin
m
L
z sin
m
L
z
0

sin
_
ct
_
(n/a)
2
+(m/L)
2
_
_
(n/a)
2
+(m/L)
2
(57)
and the electric eld is
E(r, t) = u
y
E
y
= u
y
4
La

n=1

m=1
sin
nx
a
sin
nx
0
a
sin
m
L
z sin
m
L
z
0

sin
_
ct
_
(n/a)
2
+(m/L)
2
_
_
(n/a)
2
+(m/L)
2
(58)
Figure 7 shows the normalized electric eld aE(r, t)/ at x = 0.5a, z =
0.75a, excited by the unit step waveform. Note that the response of
the metal cavity is not a unit step function.
0 1 2 3 4 5
5
0
5
E2 ctoa ( )
ctoa

y
aE

0 0
/ ( 0.5 , 0.75 , 0.5 ) ct a x a z a x z a = = = =
Figure 7. Electric eld excited by a unit step waveform f(t) = H(t).
5.3. A Coaxial Waveguide Cavity
A coaxial waveguide cavity of length L consisting of an inner conductor
of radius a and an outer conductor of radius b is shown in Figure 8. Let
the coaxial waveguide be excited by a magnetic ring current located at
248 Geyi
z = z
0
J
m
(r, t) = u

f(t)(z z
0
)(
0
), a <
0
< b, 0 < z
0
< L
where (, , z) are the polar coordinates and u

is the unit vector

in direction. According to the symmetry, only TEM mode and
those TM
0q
modes that are independent of will be excited. The
orthonormal vector mode functions for these modes are given by [12]
2a 2b
o
a
y
x
J
b
0
L z
z
Figure 8. Cross-section of a coaxial waveguide.
k
c1
= 0, e
t1
(, ) = u

e
1
(), e
1
() =
1

2 lnc
1
k
cn
=

n
a
, e
tn
(, ) = u

e
n
()
e
n
() =

n
a
J
1
(
n
/a)N
0
(
n
) N
1
(
n
/a)J
0
(
n
)
_
J
2
0
(
n
)/J
2
0
(c
1

n
) 1
, n 2
where c
1
= b/a, u

is the unit vector in direction, and

n
is the nth nonvanishing root of the eqauation J
0
(
n
c
1
)N
0
(
n
)
N
0
(
n
c
1
)J
0
(
n
) = 0. It follows from (45) and (50) that
i
TM
n
(z, t) =
2
c
e
n
(
0
)
0

df(t

)
dt

G
i
n
(z, t; z
0
, t

)dt

=
2
L

0
e
n
(
0
)

m=0

m
cos
m
L
z cos
m
L
z
0
_
(
n
/a)
2
+ (m/L)
2

t
_

df(t

)
dt

sin
_
c(t t

)
_
(
n
/a)
2
+ (m/L)
2
_
dt

(59)
Progress In Electromagnetics Research, PIER 78, 2008 249
Let f(t) = H(t) sint, the above expression may be written as
i
TM
n
(z, t) =
2kc
L

0
e
n
(
0
)

m=0

m
cos
m
L
z cos
m
L
z
0
_
(
n
/a)
2
+ (m/L)
2

t
_
0
sin
_
c(t t

)
_
(
n
/a)
2
+ (m/L)
2
_
cos t

dt

=
2k
L

0
e
n
(
0
)

m=0

m
cos
m
L
z cos
m
L
z
0

cos kct cos

_
c(t t

)
_
(
n
/a)
2
+ (m/L)
2
_
k
2
(
n
/a)
2
+ (m/L)
2
(60)
0 10 20 30 40 50 60
2
1
0
1
2
Hc ctoa (
ctoa
0 0
/ ( 2 , 1.5 , 3 , 1.5 , 2 ) ct a b a a z a a z a = = = = =
aH

Figure 9. Magnetic eld exited by a sinusoidal wave f(t) = H(t) sint
when ka = 3 (k =
_
(
n
/a)
2
+ (m/L)
2
).
The magnetic eld H(r, t) = u

in the coaxial waveguide

can be determined from (33). Figure 9 shows the magnetic eld at
= (a + b)/2 and z = 3a when ka = 3 with the assumption that
b = 2a, L = 2b,
0
= (a + b)/2, z
0
= L/2. In this case k is not equal
to any resonant wavenumbers
_
(
/
a)
2
+ (m/L)
2
(m, n 1) and the
eld response is not a sinusoidal wave. When the frequency of the
excitation waveform approaches to one of the resonant frequencies, say
ka =
2
, a sinusoidal wave will gradually build up inside the coaxial
waveguide cavity as shown in Figure 10.
250 Geyi
0 10 20 30 40 50 60
10
5
0
5
10
Hc ctoa (
ctoa
0 0
/ ( 2 , 1.5 , 3 , 1.5 , 2 ) ct a b a a z a a z a = = = = =
aH

Figure 10. Magnetic eld exited by a sinusoidal wave f(t) =
H(t) sint when ka =
2
= 3.123.
0 1 2 3 4 5
3
2
1
0
1
Hc ctoa (
ctoa
0 0
/ ( 2 , 1.5 , 3 , 1.5 , 2 ) ct a b a a z a a z a = = = = =
aH

Figure 11. Magnetic eld exited by a sinusoidal wave f(t) = H(t).
If the coaxial waveguide cavity is excited by unit step waveform
f(t) = H(t), (59) may be written as
i
TM
n
(z, t) =
2
L

0
e
n
(
0
)

m=0

m
cos
m
L
z cos
m
L
z
0
_
(
n
/a)
2
+ (m/L)
2
sin
_
ct
_
(
n
/a)
2
+ (m/L)
2
_
The eld response is shown in Figure 11, and is no longer a unit step
waveform.
Progress In Electromagnetics Research, PIER 78, 2008 251
6. CONCLUSION
This paper provides a thorough discussion on the time-domain theory
of metal cavities. A rigorous proof of the completeness of the vector
modal functions inside an arbitrary metal cavity has been presented on
the basis of the theory of symmetric operators. The major topic of this
paper is the waveguide cavity resonator, which has been examined by
using the time-domain theory of the waveguide developed previously.
The elds inside the waveguide cavity have been expanded in terms of
the vector modal functions of the corresponding waveguide, and the
eld expansion coecients are shown to satisfy the modied Klein-
Gordon equations subject to the homogeneous Dirichlet or Neumann
boundary condition, which can then be solved by the method of
retarded Greens function. Such an approach guarantees that the
solutions satisfy the causality condition. Some numerical examples
have been presented to demonstrate the time-domain theory. It has
been shown that, for a closed cavity, the time-domain response to a
sinusoidal waveform turned on at a nite instant of time is generally not
sinusoidal even when the time tends to innity. A sinusoidal response
can develop inside the cavity only when the frequency of the excitation
sinusoidal wave coincides with one of the resonant frequencies of the
metal cavity.
The traditional time-harmonic theory, however, always yields a
sinusoidal response if the excitation waveform is sinusoidal. Therefore
the time-harmonic theory fails to give a correct theoretical prediction
for a closed cavity whenever the source is turned on at a nite instant
of time which corresponds to all practical situations. In addition the
singularities occur in the time-harmonic theory for a lossless metal
cavity if the frequency of the excitation source coincides with one of
the resonant frequencies. In this case, the eld distributions are innite
everywhere inside the cavity, leading to an awkward situation. On the
other hand, the time-domain theory always gives a nite solution and
is thus more appropriate to describe what happens in a closed metal
cavity, which is the foundation for a number of important applications
in microwave engineering and is also helpful in studying various cavity-
related problems [1320].
REFERENCES
1. Kurokawa, K., An Introduction to the Theory of Microwave
Circuits, Academic Press, New York, 1969.
2. Goubau, G. (ed.), Electromagnetic Waveguides and Cavities,
Pergmon, London, 1961.
252 Geyi
3. Collin, R. E., Foundations for Microwave Engineering, 2nd
edition, IEEE Press, 2001.
4. Aksoy, S. and O. A. Tretyakov, The evolution equations in study
of the cavity oscillations excited by a digital signal, IEEE Trans.
Antennas and Propagation, Vol. 52, No. 1, 263270, 2004.
5. Bladel, J. V., Electromagnetic Fields, 1st edition, Hemisphere
Publishing Corporation, 1985.
6. Geyi, W., A time-domain theory of waveguide, Progress In
Electromagnetics Research, PIER 59, 267297, 2006.
7. Zeidler, E., Applied Functional Analysis-Applications to Mathe-
matical Physics, Springer-Verlag, 1995.
8. Byron, F. W. and R. W. Fuller, Mathematics of Classical and
Quantum Physics, Addison-Wesley, 1969.
9. Colton, D. and R. Kress, Integral Equation Methods in Scattering
Theory, John Wiley, 1983.
10. Harrington, R. F., Time-Harmonic Electromagnetic Fields,
McGraw-Hill Book Company, Inc., 1961.
11. Grandshteyn, I. S. and I. M. Ryzhik, Table of Integrals, Series,
and Products, Academic Press, New York, 1994.
12. Marcuvitz, N., Waveguide Handbook, McGraw-Hill, 1951.
13. Bopp, III, C. L. and C. M. Butler, Analysis of transmission
of a signal through a complex cylindrical/coaxial cavity
by transmission line methods, Progress In Electromagnetics
Research, PIER 56, 3351, 2006.
14. Mohsen, A. A., E. M. Elkaramany, and F. G. A. El-Hadeed,
Analysis of microwave cavities using LTL-FD method, J. of
Electromagn. Waves and Appl., Vol. 19, No. 2, 145162, 2005.
15. Yang, D., C. Liao, and W. Chen, Numerical solution on coupling
of UWB pulse into a rectangular cavity through slots, J. of
Electromagn. Waves and Appl., Vol. 19, No. 12, 16291638, 2005.
16. Kim, J. H. and H. J. Eom, Radiation from multiple annular slots
on a circular cylindrical cavity, J. of Electromagn. Waves and
Appl., Vol. 21, No. 1, 4756, 2007.
17. Xiao, J.-K., W.-S. Ji, S. Zhang, and Y. Li, A eld theoretical
method for analyzing microwave cavity with arbitrary cross-
section, J. of Electromagn. Waves and Appl., Vol. 20, No. 4,
435446, 2006.
18. Zhang, J.-P. and K.-M. Chen, Mode matching analysis of
the induced electric eld in a material sample placed within
an energized cylindrical cavity, Progress In Electromagnetics
Research, PIER 28, 295311, 2000.
Progress In Electromagnetics Research, PIER 78, 2008 253
19. Van Rienen, U., Frequency domain analysis of waveguide and
resonator with t on non-orthogonal triangular grids, Progress
In Electromagnetics Research, PIER 32, 357381, 2001.
20. Wang, C.-F., Y. Xu, and Y.-B. Gan, 3-dimnensional implemen-
tation of the eld iterative method for cavity modeling, Progress
In Electromagnetics Research, PIER 47, 2747, 2004.