TIME-VARYING FIELDS 221
amplitude is kept constant as a reference. In other words, the ampli-
tude of any voltage or current, whether of a source or of a branch, can
be kept constant as a reference. In the more general terminology used
here, any one voltage or current can be required to be a function of 7,
independent of a, thereby maintaining fixed the “shape” of its time
dependence over the entire family. It is entirely a matter of conven-
ience which voltage or current is used as a reference.
The same considerations apply to electromagnetic fields. Any elec-
tromagnetic quantity can be used as a reference, and required to be a
function of r alone, provided, of course, that the constraint introduced
by this requirement is consistent with the laws of electromagnetism.
This question of consistency and, therefore, the choice of a reference
quantity cannot be discussed profitably in general terms. We shall
return to this question in later sections, in connection with specific
illustrations.
6.3 The Power Series in the Time-Rate Parameter
We saw in the preceding section how a family of electromagnetic
fields can be defined, for any given system, by changing the time
scale with the help of an independent time-rate parameter a. We saw
also that, in such a family of electromagnetic fields, each electromag-
netic quantity is a function of the parameter a, of the family time r,
as well as of the space coordinates. Such quantities must satisfy the
laws of electromagnetism, as expressed in particular by Eqs. 6.19,
6.20, and 6.21.
Let us consider these equations from the point of view of coupling
between electric and magnetic field, as discussed in Sec. 6.1. The two
derivatives of B and D with respect to 7 are multiplied by a, so that
the coupling resulting from them is small for small values of «. In
the limit, when a approaches zero, the coupling disappears entirely,
and the equations reduce to those of static fields. This makes good
physical sense because, in such a limiting case, all electromagnetic quan-
tities vary at an infinitely slow rate, and, therefore, they must obey
the static laws. It follows that, for a = 0, the electric field can be
evaluated independently of the magnetic field, as discussed in Sec.
6.1.
For a small, although different from zero, the coupling introduced
by the derivatives with respect to r is correspondingly small. Then,
since E and H can be expected to be continuous functions of a, they
cannot differ much from their values for a = 0. As a matter of fact,222 ELECTROMAGNETIC FIELDS, ENERGY, AND FORCES
we have approximately, for small values of a,
()
E= (Eyota a) (6.22)
0
H= (Mota (=), (628)
where the subscript 0 indicates that the quantity in parentheses is evalu-
ated for a = 0. These two equations are valid at each point in space
and for each value of 7; thus the values of E and H evaluated for
a= 0, and their derivatives with respect to a, also evaluated for
a = 0, are all functions of z, y, z, and r.
The derivatives with respect to a in Eqs. 6.22 and 6.23 can be
evaluated directly from the field laws. In fact, differentiating Eqs.
6.19, 6.20, and 6.21 with respect to a, and setting a equal to zero yields,
é. a
eurl (=) - - xn (B)o (6.24)
a. ods a
curl (S)- (2) + x (D)o (6.25)
(9)
div (%). are (ps)o (6.26)
The quantities that are differentiated with respect to 7 on the right-
hand side of these equations can be obtained from the static solutions
and, therefore, can be considered as known. It follows that (dE/ da)
and (8H/da)y can be determined independently in succession just like
static fields. Thus, both terms on the right-hand side of Eq. 6.22 and
6.23 can be evaluated without having to solve the field equations
simultaneously.
If the linear approximations given by Eqs. 6.22 and 6.23 are not
satisfactory, square terms in a can be added, proportional to the second
derivatives of E and H with respect to a, evaluated for a = 0. Next,
cubic terms can be added, proportional to the third derivatives, and
so forth. The equations that the successive derivatives with respect
to @ must satisfy are more readily derived by approaching the problem
more directly, as shown below.
Since the electric field and the magnetic field are functions of the
parameter « as well as functions of x, y, 2, and 7, they can be expressedTIME-VARYING FIELDS 223
as power series in a,
E(, y, 2, 7,0) = Eg(a, y, 2,7) + ai(a, y, 2,7)
+ oP E(x, y, 2,7) +0 (6.27)
H(@, y, 2,7, 0) = Hols, y, z, 7) + alhi(@, y, 2, 7)
+ oP Hala, y, 2,7) +++ (6.28)
where
E,(z, y, 2,7) = (E(x, y, 2,7, @lamo (6.29)
aE(e, y, 2 7)
E,(z, y, 2,7) = a (6.30)
da ano
(6.31)
1 [a*EG, y, 2,7,
Ex(«, y, 2,7) = al |
kl la—0
dak
and Ho, Hy, Hj, are defined in a similar manner. All the other electro-
magnetic quantities can be expressed, similarly, as power series in a.
These expressions are then introduced into Eqs. 6.19, 6.20, and
6.21. It is assumed in this regard that all series can be differentiated
term by term with respect to the independent variables x, y, z, and 7,
so that, for instance,
curl E = curl Ey + a curl Ey + a? curl Ey +++
Then, if all terms in each equation are combined on one side, Eqs.
6.19, 6.20, and 6.21 take the form of power series in a equated to zero.
For instance, Eq. 6.19 becomes
OBo oB,
curl Ey + (cur E,+—)+e (curt E,+ *) +++-=0 (6.32)
or or
This equation must be satisfied for all values of a; i.e., the infinite sum
of the terms on the left-hand side must be equal to zero for all values
of a. This condition can be met only if the coefficients of all the
powers of @ are separately equal to zero. The same is true for the
corresponding equations obtained from Eqs. 6.20 and 6.21. It follows
that each of these equations separates into a series of equations. We
obtain for the zero-order set.
curl Ey = 0 (6.33)
curl Hp = Jyo (6.34)
div Jjo = 0 (6.35)224 ELECTROMAGNETIC FIELDS, ENERGY, AND FORCES
for the first-order set
Bo
curl Ey = — — (6.36)
or
aD,
curl Hy = Jy, + —" (6.37)
or
a,
div J, = - 2% (6.38)
or
and, in general, for the kth-order set,
cy
curl E, = — — By_4 (6.39)
ar
a
curl Hy = Jyn + ee Duy (6.40)
ir
. é
div Jj, = — — prer—1y (6.41)
or
where all quantities involved are, in general, functions of z, y, z, and r.
The other two field equations, namely, Eqs. 6.3 and 6.4, do not in-
volve explicitly the parameter a, and therefore yield series of equations
involving only terms of the same order. We have in general
div Dy = pyr (6.42)
div B, = 0 (6.43)
The same is true for the constituent relations, as long as they depend
only on the space coordinates.
The zero-order equations (Eqs. 6.33 to 6.35) are clearly of the static
type, as expected. Thus Ey and Hp can be determined independently
as static fields, although they are both functions of the family time r.
They can be regarded, from a physical point of view, as slowly varying
static fields. The first-order equations are also solvable by static-field
methods because the derivatives with respect to 7 of Bo, Do, and pyo
can be obtained from the zero-order fields. The kth-order equations
are similarly solvable by static-field methods because the quantities
that are differentiated with respect to + can be obtained from the
fields of order k — 1. Thus, all the terms in the series expansions of
E and H can be evaluated in succession by solving only equations of
the static type.
The actual electromagnetic field of interest is the member of theTIME-VARYING FIELDS 225
family of fields corresponding to a = 1. Thus, for instance, the actual
electric field and the actual magnetic field are given by the sums of
the coefficients of the powers of # in Eqs. 6.27 and 6.28, with ¢ substi-
tuted for 7. The number of terms in each series that must be evaluated
in order to obtain a satisfactory approximation depends, of course, on
the particular problem considered. Similarly, little can be said in
general about the convergence of the series involved. Therefore these
questions will be dealt with, when necessary, in connection with specific
physical problems.
In concluding, let us summarize the basic ideas involved in the
method of successive approximations developed above. To start with,
we enlarged the scope of the problem under consideration by consider-
ing a family of electromagnetic fields generated from the desired field
by changing the time scale of its sources. This was done, formally,
by substituting a family time 7 = at for the real time ¢, where was a
time-rate parameter describing the resulting family of electromagnetic
fields. Thus, a value of a larger than one corresponded to compressing
the time scale, i.e., to speeding up the electromagnetic phenomenon in
question. Conversely, a value of a smaller than one corresponded to
stretching the time scale, Le., to slowing down the electromagnetic
phenomenon in question. The advantage resulting from the definition
of such a family of electromagnetic fields was that each electromagnetic
quantity became a function of the parameter o as well as a function of
the space coordinates and of the family time. This gave us the oppor-
tunity to start from static field solutions corresponding to « = 0, and
work our way toward the desired electromagnetic field along the path
provided by the new variable a; this was accomplished by computing
successive corrections to the original static fields as the value of a was
increased. The intuitive idea of working our way from a = 0 toa = 1
in successive steps was then identified with the mathematical process
of representing all electromagnetic quantities as power series in a.
The coefficients of these series, being proportional to successive deriva-
tives with respect to a, evaluated for a = 0, were found to satisfy field
equations of the static type. Thus, the entire process turned out to
involve only determinations of static fields. The details of the process
are illustrated in the next section.
6.4 Example of Power Series Solution
The power series method developed in the preceding sections is illus-
trated below by applying it to the determination of an electromagnetic226 ELECTROMAGNETIC FIELDS, ENERGY, AND FORCES
-a x a
y
Fig. 6.3. Parallel-plate system analyzed in Sec. 6.4.
field of the simplest possible type, namely, a field which depends on a
single Cartesian coordinate. Fields of this type can be generated, in
principle, by connecting idealized voltage sources between parallel,
perfectly conducting sheets, and can be approximated in practice by
connecting physical sources between the terminals of parallel-plate
transmission lines. The particular example on which we shall focus
our attention is illustrated schematically in Fig. 6.3. Two parallel
conducting plates, normal to the y-axis are fed symmetrically by voltage
sources connected between their edges at z= a and z= —a. The
length 2a is very much greater than the spacing d between the two
plates, and the width of the plates in the direction of the z-axis is also
appreciably greater than d. Furthermore the plates have such a high
conductivity that they act as infinitely conducting sheets, at least for
the purposes of our analysis. Under these conditions, the electromag-
netic field between the plates is, for all practical purposes, independent
of x and y, except near the edges of the plates. We shall idealize this
situation in such a way as to make the field exactly independent of x
and y, by making the plates infinite in both the x and y directions and
by regarding the voltage sources as devices that force the electric field
to be uniform and parallel to the y-axis over the planes z = a and z =
—a; we shall confine our attention hereafter to the space between these
planes.
We wish to determine the electromagnetic field between the con-
ducting sheets in general terms, without specifying the time dependence
of the voltage sources, except for the fact that the two voltage sources
at opposite ends of the plates are identical. Because of the symmetry
of the system, it is convenient to use as a reference the electric field
on the plane z = 0, ie., on the plane of symmetry, rather than the
source voltage. Thus, we shall characterize the family of electromag-
netic fields by requiring the electric field on the plane z = 0 to be a
function of the family time, independent of the time-rate parameter.
More precisely, since the electric field is, by assumption, uniform andTIME-VARYING FIELDS 227
parallel to the y-axis, we set
(6.44)
where Vo(7), a function of the family time alone, represents the voltage
between the plates on the plane of symmetry of the system. This
implies that, when the electric field is expressed as a power series in a,
the zero-order term must assume the value given by Eq. 6.44 for z = 0,
and all higher order terms must vanish for z = 0. This follows from
the fact that the electric field is, by assumption, independent of a for
z=0.
The zero-order electric field must satisfy the static-field equations
curl Ey = 0 (6.45)
div Ey = 0 (6.46)
must be normal to the perfectly conducting plates, and must assume
the value given by Eq. 6.44 for z = 0. It follows that the zero-order
electric field is uniform and parallel to the y-axis throughout the space
between the conducting sheets,
v
Ey =i, um
where Eo,(r) is a scalar function of the variable 7 alone. The zero-
order surface charge on the upper conducting plate has, correspond-
ingly, a density
= 1,Eo,() (6.47)
oo = gE oy(r) (6.48)
and, therefore, is uniformly distributed over the plate. The charge on
the lower plate has the same magnitude and opposite sign. The sub-
script f indicating free charge or free current is dropped in this example
to simplify the notations in view of the absence of polarized matter.
The current flow in the system is restricted to the conducting plates,
and, therefore, can be represented by a surface current density K.
The zero-order component of this current density, because of Eq.
6.35, must satisfy the two-dimensional divergence equation, defined in
Sec. 3.2,
divs Ky = 0 (6.49)
and, therefore, it must be constant. On the other hand, the symmetry
of the system requires the current density to be an odd function of z.
It follows that the zero-order current is equal to zero, as expected in
view of the static character of the field.228 ELECTROMAGNETIC FIELDS, ENERGY, AND FORCES
The zero-order magnetic field must satisfy the static equations
curl Hy = Jo (6.50)
div Hy = 0 (6.51)
On the other hand, no zero-order currents are present anywhere. It
follows that the zero-order magnetic field vanishes everywhere.
The first-order electric field must satisfy Eqs. 6.36 and 6.42, which,
in view of the absence of any zero-order magnetic field, reduce to
curl E; = 0 (6.52)
div E; = 0 (6.53)
Again, only a uniform field, parallel to the y-axis, can satisfy these
equations and be normal to the conducting plates. The first-order elec-
tric field must vanish for z = 0, as stated above. It follows that this
field must vanish everywhere, and, as a consequence, there cannot be
any first-order charge on the plates.
The first-order current on the plates must satisfy Eq. 6.38, which
reduces to the two-dimensional equation
do
divs K, = ——° (6.54)
or
However, because of the one-dimensional character of the system, cur-
rent can flow only in the z direction. It follows that this equation re-
duces to aK, dE oy(r)
=F .
dz dr (6.55)
where the minus sign applies to the upper plate (y = 0) and the plus
sign applies to the lower plate (y = d). Thus, we obtain by integration
dEoy(r)
Ky. = Fez
Le b
(6.56)
where the constant of integration has been set equal to zero in view of
the symmetry of the system.
The first-order magnetic field must satisfy Eqs. 6.37 and 6.43. In
the space between the two conducting plates, these equations reduce to
oHy i OMe i dEo,(r)
= ie
dz ” 82 On dt
aH,
div H, = —“ =0 (6.58)
oz
curl Hy = —i,
(6.57)TIME-VARYING FIELDS 229
in view of the one-dimensional character of the field. Integration with
respect to z yields then
abou (*)
2 ———
dr
hy, = Hy, =0 (6.60)
le = &
(6.59)
where the integration constants have again been set equal to zero be-
cause of the symmetry of the system and because of the absence of
currents that can generate uniform fields in the y and z directions.
Thus, the first-order magnetic field in the space between the plates is
parallel to the x-axis, and varies linearly as an odd function of z. Its
magnitude is just equal to the magnitude of the surface current density
on the plates, given by Eq. 6.56. Thus, in accordance with Eq. 3.31,
the x component of the first-order magnetic field is discontinuous at
the two plates and vanishes in the space outside the plates.
Let us proceed next to the evaluation of the second-order terms.
The electric field must satisfy Eqs. 6.39 and 6.42, for k = 2. In the
space between the two plates, these equations reduce to
aE. ak. aH
curl Ey = —i, —" + iy —* = —izuo — (6.61)
oz
a OE a,
div E, = =0 (6.62)
dz
in view of the one-dimensional character of the field. We obtain then,
by integration, and with the help of Eq. 6.59,
he, = 2 PEo,(r)
ey = foo Fe
Boz = Eo, = 0
(6.63)
where the integration constants have been set again equal to zero be-
cause Hy, must vanish for z = 0 and because the entire electric field
must be normal to the conducting plates. Thus the second-order elec-
tric field is parallel to the y-axis, and proportional to z’.
The second-order charge on the plates has a density
2 2 PE oy (r )
og = eo Ho
6.64
2 dr? (6.64)
where the plus sign applies to the upper plate (y = 0), and the minus
sign to the lower plate (y = d). The second-order current, on the other230 ELECTROMAGNETIC FIELDS, ENERGY, AND FORCES
hand, is equal to zero, because the first-order charge is equal to zero.
Similarly, the second-order magnetic field is equal to zero because the
first-order electric field and the second-order current are equal to zero.
In fact, for k = 2, the right-hand side of Eq. 6.40, which includes all
the sources of Hz vanishes everywhere.
The pattern of the successive terms should be clear by now. We
have in general
0 k odd
Ex =|, # dE oy(r) (6.65)
i, — k even
kl dr*
fe
n= {iz iz ETFO) gaa
ie wo OL (6.66)
0 k even
oe = ek ny (6.67)
Kiz = Fire (6.68)
where
1
(6.69)
is the velocity of light in free space; the upper signs in Eqs. 6.67 and
6.68 apply to the upper plate (y = 0), and the lower signs to the lower
plate (y =d). Thus, the desired power-series expressions for the
families of electric fields and of magnetic fields are
O8 & ge iT
E(a,1,2)=i, 0 a() aE oy) (6.70)
k even k! dr*
co 1 faz\* d*Boy(r)
i =i /2r—-(2 71
(a7 2) =i za) dr* (6.71)
The corresponding family of source voltages is given by
V.(a, 7) = [E,(a, 7, 2)Jenad (6.72)
As expected, the source voltages are functions of a as well as of r be-
cause the electric field on the plane z = 0 has been used as a reference
by making it independent of a.
Let us suppose, more specifically, that the system is sinusoidally
excited in such a way that the electric field on the plane z = 0 is
[E(@®]--0 = i,A cos wt (6.73)