Vous êtes sur la page 1sur 11

Physics 506 Winter 2006

Homework Assignment #3 Solutions

Textbook problems: Ch. 8: 8.18, 8.19


Ch. 9: 9.3, 9.6

8.18 a) From the use of Greens theorem in two dimensions show that the TM and TE
modes in a waveguide defined by the boundary-value problems (8.34) and (8.36)
are orthogonal in the sense that
Z
Ez Ez da = 0 for 6=
A

for TM modes, and a corresponding relation for Hz for TE modes.


Orthogonality is a general property of the eigenfunctions of the wave equation.
The general two-dimensional equation is given by

[2t + 2 ] = 0

where either
|S = 0 TM modes
or

=0 TE modes
n S
To prove orthogonality, note that and satisfy the equations

[2t + 2 ] = 0, [2t + 2 ] = 0

Multiplying the first by and the second by and subtracting gives

(2 2 ) = 2t 2t

Integrating this over the cross-sectional area, and using Greens theorem yields
Z Z
2 2
( ) da = [ 2t 2t ] da
A A
I  

= dl
C n n
where we have used an inward pointing normal direction. We now note that
the right hand side vanishes for either TM or TE boundary conditions. Thus,
provided 2 6= 2 , we end up with
Z
da = 0 (2 6= 2 )
A
For non-degenerate eigenvalues, we conclude that
Z
da = 0 6
for =
A

For degenerate eigenvalues, we note that linearity of the wave equation guarantees
that we may find an orthogonal basis using, e.g., a Gram-Schmidt orthogonaliza-
tion process.
b) Prove that the relations (8.131)(8.134) form a consistent set of normalization
conditions for the fields, including the circumstances when is a TM mode and
is a TE mode.
We start with relation (8.131), which states
Z
E~ t, E
~ t, da = ,
A

where E~ t, may be either a TM or a TE mode. To handle this expression, we


note that the transverse fields for TM and TE modes are given by

TM: ~ t = ik
E ~ t Ez , ~ t = 1 z E
H ~t Z=
k
2 Z 
(1)
TE: ~ t = i z
E ~ t Hz , ~ t = 1 z E
H ~t Z=

2 Z k
Hence for two TM modes, we end up with
Z 2 Z
~ t, E
~ t, da = k ~ t Ez,
~ t Ez, da
E
A 2 2 A
k2
 I Z 
Ez, 2
= 2 2 Ez, dl Ez, t Ez, da
S n A

The surface term vanishes, while 2t Ez, = 2 Ez, . Hence we arrive at

k2
Z Z
~ ~
Et, Et, da = 2 Ez, Ez, da = 0 for 6= (2)
A A
When properly normalized for = , this gives (8.131) for two TM modes. The
case of two TE modes is similar. We have
Z 2 2 Z
E ~ t, da =
~ t, E ~ t Hz, ) (z
(z ~ t Hz, ) da
A 2 2 A
2 2
Z h i
= 2 2 ~ t Hz,
~ t Hz, (z
~ t Hz, )(z
~ t Hz, ) da
A
2 2
Z
= 2 2 ~ t Hz,
~ t Hz, da
A
(3)
we we have noted that z ~ t = 0 identically (since the transverse gradient is
orthogonal to z). The proof of orthogonality of two TE modes then follows using
the same integration method that was used above for the TM modes (but with
Ez replaced by Hz , and with Hz /n vanishing on the boundary). Finally, for
one TE mode and one TM mode, we have
Z Z
~ ~ k ~ t Ez, (z
~ t Hz, ) da
Et, Et, da = 2 2
A
A
Z
k ~ t Ez, ~ t Hz, ] z da
= 2 2 [
A
Z
k ~ t (Ez, ~ t Hz, ) z da
= 2 2
A
I
k ~ t Hz, d~l = 0
= 2 2 Ez,
S

This integral vanishes because Ez, vanishes on the boundary. As a result, all
TE modes are orthogonal to all TM modes. Proper normalization then results in
(8.131).
We now turn to relation (8.132), which states
Z
~ t, da = 1 ,
~ t, H
H
A Z2

The best way to prove this is to note from (1) that

~ t, = 1 z E
H ~ t,
Z

for either TM or TE modes, provided Z is chosen accordingly. In this case


Z Z
~ t, H
~ t, da = 1 ~ t, )(z E
~ t, ) da
H (z E
A Z Z A
Z h
1 ~ t, E~ t, (z E
~ t, )(z E
i
~ t, ) da
= E
Z Z A
Z
1 ~ t, E~ t, da = 1 , = 1 ,
= E
Z Z A Z Z Z2

Here we have made use of the fact that z E ~ t vanishes because E


~ t is transverse
to the z direction. The last line follows from applying (8.131), which we proved
above.
The power flow relation (8.133)
Z
1 ~ t, ) z da = 1 ,
~ t, H
(E
2 A 2Z
follows similarly. Specifically, we have
Z Z
1 ~ ~ 1 ~ t, (z E ~ t, )] da
(Et, Ht, ) z da = z [E
2 A 2Z A
Z h
1 ~ t, E~ t, (z E
~ t, )(z E
i
~ t, ) da
= E
2Z A
Z
1 ~ t, E ~ t, da = 1 , = 1 ,
= E
2Z A 2Z 2Z

The relation (8.134) essentially normalizes the modes for the TM and TE case.
Examination of (2) for TM modes and (3) for TE modes indicates that the proper
normalization is
2
Z
TM: Ez, Ez, da = 2 ,
A k
(4)
2 2
Z
TE: Ez, Ez, da = 2 2 , = 2 2
A k Z

8.19 The figure shows a cross-sectional view of an infinitely long rectangular waveguide with
the center conductor of a coaxial line extending vertically a distance h into its interior
at z = 0. The current along the probe oscillates sinusoidally in time with frequency
, and its variation in space can be approximated as I(y) = I0 sin[(/c)(h y)]. The
thickness of the probe can be neglected. The frequency is such that only the TE10
mode can propagate in the guide.
a) Calculate the amplitudes for excitation of both TE and TM modes for all (m, n)
and show how the amplitudes depend on m and n for m, n  1 for a fixed
frequency .
Before calculating the amplitudes, we work out the normalization for the rectan-
gular waveguide normal modes. For the TM modes, we recall that
mx ny
Ez = E0 sin sin
a b
(5)
~ t = ikmn
E ~ t Ez Ey =
ikmn n
E0 sin
mx
cos
ny
2
mn 2
mn b a b
where
2 m2 n2
 
2 2 2 2
kmn = 2 mn , mn = +
c a2 b2
We have explicitly written out the y component above, since it will be needed to
calculate the amplitude for excitation. Noting that sin2 averages to 1/2, the TM
normalization condition (4) gives

ab 2 2 2imn
E0 = mn
2
E0 =
4 kmn kmn ab
Similarly, the TE modes are
mx ny
Hz = H0 cos cos
a b
i i0 m mx ny (6)
~ t = 0 z
E ~ t Hz Ey = H0 sin cos
2
mn 2
mn a a b

with normalization

ab 2 2 2imn
H0 = 2mn2 H0 =
4 0 0 ab

except that if m = 0 we should take a 2a and if n = 0 we should take b 2b


in the square root, since the constant mode averages to 1 instead of 1/2. (This is
not an issue for TM modes, since in that case the m = 0 or n = 0 possibilities are
disallowed.) Before proceeding, we note here that the functional behavior of the
Ey components are identical for TM and TE modes; they only differ by constant
factors.
Now that we have written out the normal modes, we turn to the excitation
amplitude computation. The amplitudes are
Z
Zmn
()
Amn = J~ E
~ () d3 x
mn
2 V

where Zmn = kmn /0 for TM modes and Zmn = 0 /kmn for TE modes. The
source current density may be written as

J~ = yI0 sin[ c (h y)](x X)(z)(h y)

Hence Z h
Zmn
A()
mn = I0 ()
sin[ c (h y)]Ey,mn (X, y, 0) dy
2 0

where Ey,mn is given by (5) for TM modes and (6) for TE modes. Since we only
need the electric field at z = 0, we see that this expression is independent of
whether we choose a left-moving or a right-moving mode. As a result, the (+)
and () amplitudes will be equally excited. For the TM mode of (5), we evaluate

mX h
Z
Zmn ikmn n ny
A()
mn = I0 2 E0 sin sin[ c (h y)] cos dy
2 mn b a 0 b
mX h
Z
kmn n ny
= I0 sin sin[ c (h y)] cos dy
0 mn ab b a 0 b

The integral may be performed by use of the trig identity

sin cos = 21 [sin( + ) + sin( )]


The result is
Z h  1  
ny  2  n 2 nh h
sin[ c (h y)] cos dy = cos cos
0 b c c b b c

which gives
 1  
kmn n mX  2  n 2 nh h
A()
mn = I0 sin cos cos
0 cmn ab b a c b b c

According to the specifications of this problem, all the TM modes are cutoff
modes. For large m, n  1 the wavenumber is imaginary, kmn imn . Hence
for fixed , we see that
1
A()
mn (TM)
n
provided the trig functions are O(1) and do not vanish. For the TE mode of (6),
we note that the amplitude calculation is identical, except for a different constant
factor. The TE result is
1 
0 2
 
() m mX  2  n 2 nh h
Amn = I0 sin cos cos
ckmn mn ab a a c b b c

for m, n  1, all modes are cut off, and we see that


m 1 1
A()
mn
n2 (m/a)2 + (n/b)2 N3

where m n N  1. On the other hand, the propagating TE10 mode has


amplitude
 
() 0 c X h 20 c X h
A10 = I0 sin 1 cos = I0 sin sin2 (7)
k10 2ab a c k10 ab a 2c

where we have made sure to use the normalization appropriate for an n = 0 mode.
b) For the propagating mode show that the power radiated in the positive z direction
is
c2 I02
   
2 X 4 h
P = sin sin
kab a 2c
with an equal amount in the opposite direction. Here k is the wave number for
the TE10 mode.
For an expansion in properly normalized normal modes, the radiated power in
the +z direction is given by

1 (+) 2

Pmn = A
mn
2Zmn
Using the amplitude coefficient (7) for the TE10 mode, we find

k10 (+) 2 0 c2 2 2 X h
P = A10 = I0 sin sin4 (8)
20 k10 ab a 2c

c) Discuss the modifications that occur if the guide, instead of running off to infinity
in both directions, is terminated with a perfectly conducting surface at z = L.
What values of L with maximize the power flow for a fixed current I0 ? What is
the radiation resistance of the probe (defined as the ratio of power flow to one-half
the square of the current at the base of the probe) at maximum?
If we place a perfectly conducting surface at z = L (we take L positive), then
the right-moving wave will be perfectly reflected at this surface. As a result,
the wave flowing out of the left end (z < 0) of the waveguide will be the linear
superposition of two components: the left-moving wave generated by the probe,
and the right-moving wave reflected off of the conducting surface at z = L. We
further note that, since Ek must vanish at the surface of a perfect conductor,
the reflected wave must come back 180 out of phase. We start by writing the
left-moving wave as
E~ () = A() E
~ t,10 eikz
10

The right-moving wave leaves the source as

~ (+) = A(+) E
E ~ t,10 eikz
10

and it is easy to see that the reflected wave must be

~ (refl) = A(+) E
E ~ t,10 eik(2Lz)
10

so that it satisfies the conductor boundary condition

~ (+) (z = L) + E
E ~ (refl) (z = L) = 0

Therefore, for z < 0, the total left-moving wave is given by

~ =E
E ~ (refl) = (A() A(+) e2ikL )E
~ () + E ~ t,10 eikz = A10 (1 e2ikL )E
~ t,10 eikz
10 10

The maximum amplitude case occurs when there is constructive interference

kL = (n + 12 )

Since the amplitude doubles, the power is increased by a factor of four compared
with (8). For this maximum power case, the radiation resistance is given by

40 c2 2 2 X h 1
P = I0 sin sin4 = I02 Rrad
k10 ab a 2c 2
or
80 c2 X h
Rrad = sin2 sin4
k10 ab a 2c

9.3 Two halves of a spherical metallic shell of radius R and infinite conductivity are sepa-
rated by a very small insulating gap. An alternating potential is applied between the
two halves of the sphere so that the potentials are V cos t. In the long-wavelength
limit, find the radiation fields, the angular distribution of radiated power, and the
total radiated power from the sphere.
In the long wavelength limit, we may appeal to the multipole expansion of the
source. In this case, the source is essentially a harmonically (eit ) varying
version of the electrostatic problem with hemispheres at opposite potential. The
long wavelength limit is also equivalent to the low frequency limit. In this case, it
is valid to think of the source as a quasi-static object. Using azimuthal symmetry,
the potential then admits an expansion in Legendre polynomials
 l+1
X R
(r, ) = l Pl (cos )
r
l

where Z 1
2l + 1
l = (R, cos )Pl (cos )d cos
2 1

For hemispheres at opposite potential V (times eit , which is to be under-


stood), the expansion coefficients are
Z 1
l = (2l + 1)V Pl (x) dx odd l only
0

The dipole is dominant, with 1 = 23 V . This gives rise to a dipole potential of


the form  2
3 R 3 z
= V P1 (cos ) = V R2 3
2 r 2 r
This makes it straightforward to read off an electric dipole moment

p~ = 40 ( 32 V R2 z) = 60 V R2 z

Working in the radiation zone, this electric dipole gives


2 ikr
~ = ck (r p~ ) e ck 2 eikr 3 eikr
H = 60 V R2 sin = Z01 V (kR)2 sin
4 r 4 r 2 r
and
ikr
~ = 3 V (kR)2 e sin
~ = Z0 r H
E
2 r
The angular distribution of dipole radiation gives

dP c2 Z0 4 2 2 c2 Z0 4 9
= 2
k |~
p | sin = 2
k 36 2 20 V 2 R4 sin2 = Z01 V 2 (kR)4 sin2
d 32 32 8

and the total radiated power is

P = 3Z01 V 2 (kR)4

9.6 a) Starting from the general expression (9.2) for A~ and the corresponding expression
for , expand both R = |~x ~x | and t = t R/c to first order in |~x 0 |/r to obtain
0 0

the electric dipole potentials for arbitrary time variation


 
1 1 1 ~
pret
(~x, t) = ~n p~ret + ~n
40 r2 cr t
~ x, t) = 0 ~
A(~
pret
4r t

where p~ret = p~(t0 = t r/c) is the dipole moment evaluated at the retarded time
measured from the origin.
We start with the scalar potential, which is given by

(~x 0 , t |~x ~x 0 |/c) 3 0


Z
1
(~x ) = d x (9)
40 |~x ~x 0 |

We now use the expansion

|~x ~x 0 | r n ~x 0

as well as
|~x ~x 0 | r n ~x 0 n ~x 0
t0 = t t + = tret +
c c c c
where tret = t r/c. Since is a function of time t0 , we make the expansion

n ~x 0 (~x 0 , tret )
(~x 0 , t0 ) = (~x 0 , tret ) + +
c t

As a result, the expansion of (9) becomes


Z    
1 0 1 1
(~x ) = + n ~x + + d3 x0
40 r r c t
   
1 1 1 ~
p
= Q + n p~ + +
40 r r c t
where the retarded time dependence is to be understood, and where we have used
the fact that Z Z
3 0
Q = d x , p~ = ~x 0 d3 x0

Dropping the static Coulomb potential (which does not radiate) then gives
 
1 1 1 ~
p
(~x ) n p~ + n (10)
40 r2 cr t

For the vector potential, the expansion is even simpler. We only need to keep the
lowest order behavior
Z ~ 0 0
|~ |/c)
Z h
~ 0 J(~ x , t x ~
x 0 0 ~
i
A(~x ) = 0
3
d x = J + d3 x0
4 |~x ~x | 4r

Using integration by parts, we note that

x0i
Z Z Z Z
pi
3 0
Ji d x = 0 Jj d x = xi ( J~ ) d x = x0i d3 x0 =
3 0 ~ 3 0
xj t t

Hence
~ x 0 ) 0 ~
A(~
p
(11)
4r t

b) Calculate the dipole electric and magnetic fields directly from these potentials
and show that
2 p~ret
 
~ 0 1 ~
pret 1
B(~x, t) = 2 ~n 2 ~n
4 cr t c r t2
2 p~ret
    
~ 1 r 3~n(~n p~ret ) p~ret 1
E(~x, t) = 1+ + 2 ~n ~n
40 c t r3 c r t2

~ =
For the magnetic field, we use B ~ A,~ where the vector potential is given
by (11). It is important to note that the electric dipole p~ in (11) is actually a
function of retarded time
p~ret = p~(t r/c)
Application of the chain rule then gives

~
pret 1 ~
pret
=
r c t
~ = n, the magnetic field turns out to be
Since r

1 2 p~
   
~ 0 ~ 1 ~p 0 1 ~
p
B= = n 2 (12)
4 r t 4 r t cr t2
The expression for the electric field is a bit more involved. Using (10) and (11),
we obtain

~
~ A
~ =
E
t
0 2 p~
 
1 ~ ~x ~x ~ p
= p~ + 2
40 r3 cr t 4r t2
    
1 p~ ~x(~x p~ ) ~x ~p 1 ~ p 2~x ~p
= 3 4 ~x + 2 4 ~x
40 r3 r5 cr t cr t cr t
2 2
 
~x p~ 1 1 p~
2 3 ~x 2
c r t 40 c2 r t2
1 2
 
1 p~ 3n(n p~ ) 1
= + 2 (~ p 3n(n p~ )) + 2 p n(n p~ ))
(~
40 r3 cr t c r t2
2 p~
   
1 r 3n(n p~ ) p~ 1
= 1+ + 2 n n 2
40 c t r3 c r t
(13)
c) Show explicitly how you can go back and forth between these results and the
harmonic fields of (9.18) by the substitutions i /t and p~eikrit
p~ret (t0 ).
Making the substitution


p~ret p~eikr and i
t

the magnetic field (12) becomes


 
~ 1 1 1 2
H= n 2 (i)~ p ( )~ p eikr
4 r cr
2 ck 2 eikr
 
 c  ikr 1
= (n p~ ) 1 e = (n p~ ) 1
4cr ir 4 r ikr

while the electric field (13) becomes


  3n(n p~ ) p~ 
~ = 1 r 1
E 1 + (i) 3
+ 2 ( )n (n p~ ) eikr
2
40 c r c r
 ikr ikr

1 e 2e
= (1 ikr)(3n(n p~ ) p~ ) k n (n p~ )
40 r3 r

To go in the other direction, we simply read these equations backwards.