Você está na página 1de 112

W.C.

Chew
ECE 350 Lecture Notes

1. Elements of Complex Algebra


Complex numbers are extensions of real numbers, and they make the
number elds complete in the sense that an n-th order polynomial has n-th
roots in a complex eld while it is not always true in the real eld. Complex
numbers are also very useful in time harmonic analysis of engineering and
physical systems, because they considerably simplify the analysis.
A complex number can be represented in cartesian form as

c = a + jb
(1)
p
where j = ;1. a is the real part of c while b is the imaginary part of c.
On the complex plane, c is represented by a point c or sometimes an arrow
oc as shown.
Imaginary Axis
c

b
|c|

Real Axis
0
a
Sometimes it is more convenient to represent c in polar form, i.e.

c = a + jb = jcj ej = jcj cos  + j jcj sin 


p
where jcj = a2 + b2 is the magnitude or the absolute value of c.

(2)

From (2), it is seen that

tan  = ab )  = tan;1 ab

(3)

where  is the phase of c.

Addition and Subtraction


Addition and subtraction of complex numbers are carried out in Cartesian
forms.
1

W.C.Chew
ECE 350 Lecture Notes

2. Review of Vector Analysis


A vector A can be written as

A = x^Ax + y^Ay + z^Az :

(1)

Similarly, a vector B can be written as

B = x^Bx + y^By + z^Bz :

(2)

In the above, x^ y^ z^ are unit vectors pointing in the x y z directions respectively. Ax Ay and Az are the components of the vector A in the x y z
directions respectively. The same statement applies to Bx By , and Bz .

Addition
A + B = x^(Ax + Bx) + y^(Ay + By ) + z^(Az + Bz ):

(3)

Multiplication
(a) Dot Product (scalar product)

A B = AxBx + Ay By + Az Bz 
A B = B A
commutative property
A (B + C) = A B + A C distributive property
A B = A B cos :
In (7),  is the angle between vectors A and B.



jj

(4)
(5)
(6)
(7)

(b) Cross Product (vector product)









x^ y^ z^ 
A B = Ax Ay Az  =^x(Ay Bz Az By ) + y^(Az Bx AxBz )
Bx By Bz 
+ z^(AxBy Ay Bx)


(8)

A B = u^ A B sin 

(9)
where u^ is a unit vector obtained from A and B via the right hand rule.
A  (B + C) = A  B + A  C distributive property (10)
A  (B  C) 6= (A  B)  C non-associative property
(11)
A  B = ;B  A anti-commutative property
(12)


jj

Vector Derivatives
Del
Gradient
Divergent
Curl

@ + y^ @ + z^ @ 
= x^ @x
@y @z

(13)

@  + y^ @  + z^ @ 
 = x^ @x
@y
@z
@
@
@
A = @x Ax + @y Ay + @z Az 



 x
^
y
^
z
^



(14)

r

r 

A =  @x@

@
@y
y

(15)

@ 
@z 
z

Ax A A



@
@
@
@
= x^ @y Az @z Ay + y^ @z Ax @x Az


@
@
(16)
+ z^ @x Ay @y Az :


Divergence Theorem
I

r

AdV = A n^ dS:


(17)

Stokes Theorem
I

(r  A)  n^dS =

A dl:

(18)

Some Useful Vector Identities


a (b c) = b (c a) = c (a b)
a (b c) = b (a c) c (a b)
a a = 0
a (a b) = 0





(r) = 0
r  (r  A) = 0
r  ( A) = A  r +  r  A
r  (A  B) = B  r  A ; A  r  B
r  r  A = r(r  A) ; r  rA
@2 + @2 + @2 :
2
r = rr =
@x2 @y2 @z2
r

(19)
(20)
(21)
(22)
(23)
(24)
(25)
(26)
(27)
(28)

W.C.Chew
ECE 350 Lecture Notes

3. Wave Equation from Maxwell's Equations


Lossless Medium
In a source free region, Maxwell's equations are

r  H = @@tD
r  E = ; @B 
r  B = 0
r  D = 0

(1)
(2)
(3)
(4)

@t

where B = H and D = E. Taking the curl of (2), we have


@
r  (r  E) = ; @t
r  H:

(5)

Substituting (1) into (5), we obtain


2

r  r  E = ; @t@ 2 E:

(6)

Making use of the vector identity that

r  r  E = r(r  E) ; r2 E
we have

r(r  E) ; r2E = ; @t@ 2 E:


Since the region is source free, and r  E = 0, we have
2

r2 E =  @t@ 2 E

(7)
(8)
(9)

which is the vector wave equation in freespace where r  E = 0.


Similarly, we can show that
2

r2H =  @t@ 2 H

if r  H = 0, which is, of course, true in free space.


1

(10)

Plane Wave Solutions to the Vector Wave Equations


The condition for arriving at Equation (9) is that r E = 0. We can have
solutions of the form
E = x^Ex(z t)
(11)
E = y^Ey (z t)
(12)
but not
E = z^Ez (z t)
(13)
because (13) violates r  E = 0 unless Ez is independent of z. If E is of the
form (11), then

r E = x^
2

with both

@2
@x2

and

@2
@y2

 @2

@x2

@2
@y 2

@2
@2
E
(
z
t
)
=
x
^
Ex
x
@z 2
@z 2

(14)

equal to zero. Then (9) becomes


@2
@2
E
(
z
t
)
;

E (z t) = 0:
@z 2 x
@t2 x

(15)

Similarly, if H = y^Hy (z t), (10) becomes


@2
@2
H
(
z
t
)
;

Hy (z t) = 0:
y
@z 2
@t2

(16)

Equations (15) and (16) are scalar, one dimensional wave equations of the
form
1 @ 2 y(z t) = 0
@2
y
(
z
t
)
;
(17)
@z 2
v 2 @t2

where v = 1=p. The solution to (17) is of the form y = f (z + at). We can


show that
@f
@
f = f (z + at)
= af (z + at)
(18)
@z
@t
0

@2
f
@z 2

@ 2f
@t2

= f (z + at)
00

Substituting (19) into (17), we have


f (z + at) ;
00

= a2f (z + at):
00

a2
f (z + at) = 0
v2
00

(19)
(20)

which is possible only if a = v. Hence, the general solution to the wave
equation is
y = f (z ; vt) + g (z + vt)
(21)
where f and g are arbitrary functions.
2

The solution f (z ; vt) moves in the positive z-direction for increasing t.


f(z)

f(z-vt)
t>0

t=0

z=0

z=vt

g(z)

g(z+vt)

t=0

t>0
z

z=0

z=-vt

The solution g(z + vt) moves in the negative z-direction for increasing t.
The shapes of the functions f and g are undistorted as they move along.
We can observe wavelike behavior in a pond when we drop a pebble into it.
Solutions to (9) and (10) that correspond to a plane wave is of the form
E = x^f1 (z ; vt) H = y^f2(z ; vt):
(22)
The wave is propagating in the z-direction, but the electric and magnetic
elds are transverse to the direction of propagation. Such a wave is known
as the Transverse Electro Magnetic wave or TEM wave.
If one substitutes (22) into Equation (2), one has
@
@
Ex = ; H
(23)
r  E = y^ @z
@t
or
@
@
f1 (z ; vt) = ; f2 (z ; vt)
(24)
@z
@t
or
f1 (z ; vt) = vf2 (z ; vt)
(25)
or
r
f2 (z ; vt) =
f1 (z ; vt):
(26)

Hence, for a plane TEM wave,
r
0

The quantity

Ex
Hy




= 377  for free space.

= 
is also known as the intrinsic impedance of free-space.
Z

(27)
(28)

W.C.Chew
ECE 350 Lecture Notes

4. Using Phasor Techniques to Solve Maxwell's Equations


For a time-harmonic (simple harmonic) signal, Maxwell's Equations can
be easily solved using phasor techniques. For example, if we let
H = <e H~ ej!t ]
(1)

E = < E~ j!t]
e

and substituting into (3.1), we have

~ j!t ] = <
< r  H
e

(2)

@
@t

E~ j!t
e

(3)

We could replace @t@ by j ! since the signal is time harmonic. Furthermore,


we can remove the <e operator and obtain
~ ej!t = j !E~ ej!t
rH
(4)
where ej!t cancels out on both sides.
Equation (4) implies Equation (3). Also, any time dependence cancels out in
the problem. Hence,
~ = j !E~ :
rH
(5)
Similarly,
~
r  E~ = ;j !H
(6)
~ = 0
r  H
(7)
r  E~ = 0:
(8)
Taking the curl of (6) and substituting (5) into it, we have

~ = 2 E~
r  r  E~ = ; r  H
(9)
Again, making use of the identity rr E~ = r(r E~ ) ;r2E~ , and r E~ = 0,
we have
r2E~ = ; 2 E~
(10)
Similarly,
~ = ; 2 H~
r2 H
(11)
These are the Helmholtz's wave equations.
j !

! 

! 

Lossy Medium (Conductive Medium)


1

! 

Phasor technique is particularly appropriate for solving Maxwell's equations in a lossy medium. In a lossy medium, Equation (3.1) becomes

rH= E +J


(12)

@t

where J is the induced currents in the medium, and hence,

J= E


(13)

Applying phasor technique to (12), we have

~=
rH
=

We can dene the quantity

E~ + E~ 

j !

j!

E~

(14)

~ =  ; j !

(15)

to be the complex permittivity of the medium, and (14) becomes

~ = ~E~
rH
(16)
Notice that the only dierence between (16) and (5) is the complex permittivity versus the real permittivity. If one goes about deriving the Helmholtz
wave equations for a lossy medium, the results are
r2E~ = ; 2 ~E~
(17)
j !

! 

~ = ; 2 ~H~
r2 H
(18)
Hence, a lossy medium is easily treated using phasor technique by replacing
a real permittivity with a complex permittivity.
If we restrict ourselves to one dimension, Equation (17), for instance,
becomes of the form
2
~ ( ) ; 2 ~x ( ) = 0
(19)
2 x
where
r 

p
=
~=
;
= +
(20)
! 

dz

j!



j!

 E

j
:

The general solution to (19) is of the form


~x(z) = C1e;z + C2 e+z :
E

(21)

In real space time,


( ) = <e E~x(z)ej!t ]
= <e C1e;z ej!t] + <e C2ez ej!t]

Ex z t

(23)

If C1 = jC1 j ej1 

C2

= jC2j ej2 

= + j
, then

( ) = jC1 j cos(!t ;
z + 1 )e;z + j C2 j cos(!t +
z + 2 )ez : (24)

Ex z t

Note that one of the solutions in (24) is decaying with z while another solution
is growing with z. The function cos(!t 
z + ) can be written as cos 
(z 
! t) + ]. Hence, it moves with a velocity

v

= !
:

(25)

Depending on its sign, it moves either in the positive or negative z direction.


In the above,  is the propagation constant, is the attenuation constant
while
is the phase constant.

Intrinsic Impedance
The intrinsic impedance can be easily derived also in the phasor world.
The phasor representation of Equation (3.23) is
d

~ = ;j !H~ y :

(26)

~ = ;j !E~x:

(27)

Ex

dz

A corresponding one for H~ y is


d
dz

Hy

If we now let E~x = E0e;z , H~ y = H0e;z , and using them in (26) yields
;z

 E0 e

= ;j !H0e;z :

The above implies that


=H =
E0

j !





(28)
(29)

For a lossy medium, we replace  by the complex permittivity and the intrinsic
impedance becomes


~=


j
!

The above is obviously a complex number.

j !

+ j ! :

(30)

W.C.Chew
ECE 350 Lecture Notes

5. Transmission Lines
=

= GENERAL

STRIP LINE

COAXIAL

Examples of Transmission lines

Symbol of a Transmission Line

Symbol of a Transmission Line

Another place where wave phenomenon is often encountered is on transmission lines. A transmission line consists of two parallel conductors of arbitrary cross-sections that can carry two opposite currents or two opposite
charges. A transmission line has capacitances between the two conductors,
and the conductors have inductances to them. We can characterize this capacitance by a line capacitance C which has the unit of farad m;1, and the
inductance by a line inductance L, which has the unit of henry m;1 . Hence
a transmission line can be approximated by a lumped element equivalent as
1

shown
LZ

LZ V+V LZ
I

CZ

I+I

CZ

V+2V
I+2I

CZ
I
Z

We can derive the voltage equation between nodes (1) and (2) to get
V

or


; (V + V ) = L z @I
@t

(1)

V = ;L z @I
:
@t

(2)

:
; I = C z @ (V +@t V ) ' C z @V
@t

(3)

Similarly, the current relation at node (3) says that

In the limit when we let our discrete or lumped element model become very
small, or z ! 0, we have
@V
@I
=
;
L 
(4)
@z
@t
and

@I
@z

= ;C @V
:
@t

(5)

The above are known as the telegrapher's equations. Wave equations can be
easily derived from the above

and

@ 2V
@z 2

; LC @@tV2 = 0

@ 2I
@z 2

; LC @@tI2 = 0:

(6)
(7)

Comparing with Equation (3.17), we deduce that the velocity of the current
and voltage waves on a transmission line is
1
v=p :
(8)
LC

The solution to (6) may be of the form


V (z t) = f (z ; vt):

(9)

Substituting into (4), we have

;L @I
= f 0(z ; vt)
@t

or

I (z t) =

Hence,

V (z t)
I (z t)

(10)

1 f (z ; vt):
Lv

(11)

= Lv =

(12)

rL
C

for a forward going wave. The quantity


Z0

rL

(13)

is the characteristic impedance of a transmission line.

Lossy Transmission Line


Often time, a transmission line has loss to it. For example, the conductor
has a nite conductivity and hence is a little resistive. The insulation between
the conductors may have current leakage, thus not forming an ideal capacitor.
A more appropriate lumped element model is as follows.
RZ

LZ

GZ

RZ

CZ

LZ

GZ

RZ

CZ

LZ

GZ

CZ

The above circuit is more easily treated using phasor techniques. If we


have applied phasor technique to (4) and (5), we would have obtained
dV~
dz

= ;j!LI~

(14)

dI~
= ;j!C V~ :
dz

(15)

Note that j!L is the series impedance per unit length of the lossless line
while j!C is the shunt admittance per unit length of the lossless line. In the
lossy line case, the series impedance per unit length becomes
Z = j!L + R
(16)
while the shunt admittance per unit length becomes
Y = j!C + G
(17)
where R and G are line resistance and line conductance respectively. The
telegraphers equations become
dV~
= ;Z I~
(18)
dz
dI~
= ;Y V~ 
dz

(19)

and the corresponding Helmholtz wave equations are


d2 V~
; ZY V~ = 0
dz 2

(20)

d2 I~
; ZY I~ = 0:
dz 2

(21)

Similarly, the characteristic impedance, is


Z0

j!L
j!C

) Z0 =

j!L + R
j!C + G

rZ
Y

Equations (20) and (21) are of the same form as (4.22) or


d2 V~
;  2V~ = 0
dz 2
where

d2 I~
;
 2 I~ = 0
2
dz

(22)

(23)
(24)

= ZY = (j!L + R)(j!C + G) =  + j:


(25)
The general solution is of the form (4.23). For example,
V~ (z ) = V+ e;z + V; e+z
= V+e;z;jz + V;ez+jz :
(26)
If V+ = jV+j ej+  V; = jV;j e+j; , then the real time representation of
V is
V (z t) = <eV~ (z )ej!t ]
= jV+j e;z cos(!t ; z + 1 ) + jV;j ez cos(!t + z + 2 ): (27)


The rst term corresponds to a decaying wave moving in the positive zdirection while the second term corresponds to a wave decaying and moving
in the negative z-direction. Hence, e;z corresponds to a positive going wave,
while e+z corresponds to a negative going wave.
If the transmission line is lossless, i.e., R = G = 0, then, the attenuation
constant  = 0, and the propagation constant  becomes  = j . In this
case, there is no attenuation, and (26) becomes
V~ (z ) = V+ e;jz + V; e+jz 
(28)
and (27) becomes
V (z t) = jV+ j cos(!t ; z + 1 ) + jV; j cos(!t + z + 2 ):

(29)

The wave propagates without attenuation or without decay in this case.


The velocity of propagation is v = != .
Furthermore, we can derive the current that corresponds to the voltage
in (26) using Equation (18). Hence
1 dV~ =  V e;Z ;  V e+Z :
I~ = ;
(30)
+
;
Z dz
Z
Z
But


Z

rY

= Z1

(31)

= I+e;Z + I;eZ 

(32)

V;
I;

(33)

where Z0 is the characteristic impedance given by Equation (22). Hence,


I~ =

where

V+ ;Z V; Z
e
;Z e
Z0
0
V+
I+

= Z0 

= ;Z0 :

W.C.Chew
ECE 350 Lecture Notes

6. Terminated Uniform Lossless Transmission Lines


Zo, v

lossless

Z L LOAD

z=0

z = l

Consider a lossless transmission line terminated in a load of impedance


.
L A wave traveling to the right will be reected at the termination. In
general, there will be both positive going and negative going waves on the
line. Hence,
~ (z) = V0 e;jz + V1e+jz :
V
(1)
Here,  = j  ,  = 0, because of no loss. The corresponding current, as in
(5.32), is
~(z) = V0 e;jz ; V1 e+jz 
I
(2)
Z
Z
Z

qL

where Z0 =
At z = 0,

and  = !

for a lossless line.

LC

~ (z = 0)
V0 + V1
=
ZL =
Z0 :
~(z = 0)
V0 ; V1
I
We can solve for V1 in terms of V0 , i.e.
V

V1

; Z0 V :
= ZZL +
0
Z

(4)

; Z0 
= ZZL +
Z

(5)

If we dene
v

(3)

then V1 = v V0 , and Equation (1) becomes


~ (z) = V0e;jz + v V0e+jz :
V

(6)

In the above, v is the ratio of the negative going voltage amplitude to the
positive going voltage amplitude at z = 0, and it is known as the voltage
reection coecient.
1

The current reection coecient is dened as the ratio of the negative


going current to the positive going current at z = 0, and it is
i=

The current can be written as


~(z) =
I

I1
I0

V0
Z0

= ; VV1 = ;v :

(7)

;jz ; 

V0
Z0

jz :

(8)

The voltage and current in (6) and (8) are not constants of position. We can
dene a generalized impedance at position z to be
;jz + v e+jz
~ (z)
V
e
(9)
Z (z ) =
=
Z0
;jz ; v e+jz :
~(z)
e
I
At z = ;l, this becomes
jl + v e;jl
e
Z (;l ) = Z0
(10)
jl ;  e;jl :
e
v

With v dened by (5), we can substitute it into (10) to give after some
simplications,
ZL + j Z0 tan l
:
(11)
Z (;l ) = Z0
Z + j Z tan l
0

Shorted Terminations
If ZL is a short, or ZL = 0, then,
Z (;l ) = j Z0 tan l = j X:

(12)

x
inductive
2

3
2

5
2

capacitive

Open-Circuit Terminations
If ZL is an open circuit, ZL = 1, then
Z (;l ) = ;j Z0 cot l = j X:
2

(13)

inductive

3
2

5
2

capacitive

Standing Waves on a Lossless Transmission Line


The positive going wave in Equation (6) is
( ) = V0 e;jz 

(14)

V+ z

and the negative going wave in Equation (6) is


; (z ) = v V0 e

+jz

(15)

We can dene a generalized reection coecient to be the ratio of V+(z)


to V;(z) at position z. Hence,
;(z) = VV;((zz)) = v e2jz :

(16)

(z) = V0 e;jz 1 + ;(z)]:


The magnitude of V (z) is then

(17)

Hence,

j ( )j = j 0 j j1 + ;( )j
V

(18)

A plot of jV (z)j is as shown.


Im Axis

dmin

+z
z=0
z

Re Axis

(z)

d1

1+

(z)
3

|V(z)|
Vmax

Vmin
d1
2

dmin

d1

z=0

We can use the triangular inequality and show that


jV0j (1 ; j;(z)j)  jV (z)j  jV0j (1 + j;(z)j):
(19)
From (16), j;(z)j = jv j, hence (19) becomes,
jV0j (1 ; jv j)  jV (z)j  jV0j (1 + jv j):
(20)
The voltage standing wave ratio is dened to be Vmax =Vmin, and from (20),
it is
jv j :
(21)
VSWR = 11 +
; jv j
If v = 0, then VSWR= 1, and we have no reected wave. We say that
the load is matched to the transmission line. Note that v = 0 when ZL = Z0 .
If jv j =1, then VSWR= 1, and we have a badly matched transmission
line. In a passive load,
0  jv j  1:
(22)
jv j =1 only when ZL = 0, or ZL = 1 according to Equation (5). Hence,
1  VSWR < 1:
(23)
VSWR is an indicator of how well a load is being matched to the transmission
line. We can solve (21) for jv j in terms of VSWR, i.e.
VSWR ; 1 :
jv j = VSWR
(24)
+1
Therefore, given the measurement of V S WR on a terminated transmission
line, we can deduce the magnitude of v . Furthermore, if we know the phase
of v , we would be able to derive ZL from (5), or
1 + v 
(25)
ZL = Z0
1 ; v
or
1 + jv j ejv 
ZL = Z0
(26)
1 ; j j ejv
v

where

Determining

= jv j ejv :

(27)

from j ( )j
V

can be determined from the voltage standing wave measured. The


voltage standing wave pattern is proportional to j1 + ;(z)j, but ;(z) is related
to v as
;(z) = v e2jz :
(28)
Writing the polar representation of v , we have,
v

;(z) = jv j ej(2z+v ):

(29)

However, we know that the rst minimum value of V (z) occurs when ;(z) is
purely negative, or the phase of ;(z) is ; . This occurs at z = ;dmin rst.
In other words,
;2dmin +
v = ; :
(30)
Since dmin can be obtained from the voltage standing wave pattern measurement, and that  = 2 = , we deduce that
4 d :

= ; +
(31)
v

min

Transmission Coecients
It is sometimes useful to dene a transmission coecient on a transmission line. The transmission coecient may be dened as the ratio of the
voltage on the load to the amplitude of the incident voltage. Since
V

(z) = V0e;jz + v V0e+jz :

(32)

The voltage at the load is V (z = 0), and it is given by


V

(0) = V0(1 + v ):

Since the amplitude of the incident voltage is V0 , we have


V (0)
2ZL :
v =
=
1
+
v =
V0
ZL + Z0

(33)
(34)

W.C.Chew
ECE 350 Lecture Notes

7. The Smith Chart


We have seen from Equation (6.9) that a generalized impedance can be
de ned as
V~ (z )
e;jz + v e+jz
Z (z ) =
=
Z0 ;jz
(1)
e
; v e+jz :
I~(z )
The above can be written as
1 + v e2jz = Z 1 + ;(z) 
Z (z ) = Z0
(2)
0
1 ; v e2jz
1 ; ;(z)
where ;(z) is as de ned in (6.16). When z = 0, Z (0) = ZL, and ;(0) = v ,
and (2) becomes (6.25). Hence (6.25) is a special case of (2). We can introduce
a normalized generalized impedance to be
Z (z ) 1 + ;(z )
= 1 ; ;(z) :
(3)
Zn (z ) =
Z0
Similarly,
; 1:
(4)
;(z) = ZZn((zz)) +
1
n

Given ;(z), we can solve for Zn(z) in (3), and given Zn(z), we can solve for
;(z) in (4). It turns out that the mapping of Zn(z) to ;(z) and the mapping
of ;(z) to Zn(z) are one-to-one. We shall next discuss a graphical method to
solve (3) and (4) rapidly using the Smith Chart.
Zn = Rn + jXn
Xn

Rn = .5 Rn = 1

Im
Xn = 1

Rn = 2

Rn = 0

Xn = 1

Rn = 1
Rn = .5

Rn = 2

Xn = 0

Rn = 0
0

0.5

Rn

Re

|| = 1
circle

Xn = 1

0.5

Xn = 1
plane

Zn plane

is a complex number and can be represented by a point on the Zn-plane,


and ; is a complex number and can be represented by a point on the complex
; plane.

Zn

We noted that from Equation (4) that:


(i)
(ii)
(iii)
(iv)
(v)
(vi)

When Zn = 0 ; = ;1.
When Zn = 1, or Rn = 1 Xn = 0 ; = 0.
When Zn ! 1 in any direction, ; ! 1.
When Zn = jXn j;j = 1.
When Zn = j , or Rn = 0 Xn = 1 ; = j .
When Zn = ;j , or Rn = 0 Xn = ;1 ; = ;j .

If one works out the mapping from Zn-plane to ;-plane completely, one
nds that the Rn = 0 line on Zn-plane maps onto the unit-circle on the ;plane. Furthermore, the other Rn = constant lines map into circles as shown.
The Xn = constant lines map into arcs like the Xn  1 lines as shown. Hence,
if one puts grids on the ;-plane, one can read o the Rn and Xn associated
with the corresponding ; immediately, and, given the value of ;, one can
read o the values of Rn and Xn immediately.
The mappings (3) and (4) are known as bilinear transforms. A bilinear
transform always maps a circle onto a circle.

Properties of a Smith Chart


(i) The normalized admittance Yn = 1=Zn, or the reciprocal of Zn, can be
found easily from a Smith Chart, because
;=

Zn ; 1
Zn + 1

1 ; Z1n 1 ; Yn
Yn ; 1
=
=
;
:
1 =
Yn + 1
1 + Zn 1 + Yn

(5)

(ii) The change of impedance along the line is obtained by adding or subtracting phase to ;(z) via the relationship
(iii)

;(z) = v e2jz :

(6)

VSWR = 11;+j jv jj = Rn max 

(7)

since the Smith Chart is a graphical tool to solve Equation (7), and jv j
is real, corresponding to a number on the Xn = 0 line. Notice that
1 < VSWR < 1 always.

W.C.Chew
ECE 350 Lecture Notes

8. Examples on Using the Smith Chart


(a) Find the voltages at A on the transmission line.
Z0 = 50 W, v = 1.5 108 ms1

20
Vs =
10 sin t
volts

Zs

B
ZL (30+j25)

z = l = 1 m

25 MHz

z=0

The voltage source sets up a forward going and a backward going wave
on the transmission lines. Hence,
V (z) = V0 e;jz + v V0 ejz :
(1)
The corresponding current is
(2)
I (z) = ZV0 e;jz ; v ZV0 ejz :
0
0
In impedance at position Z is
;jz + v ejz
1 + ;(z) 
Z (z) = VI ((zz)) = Z0 ee;jz ;
=
Z
(3)
0
v ejz
1 ; ;(z)
where
;(z) = v e2jz  v = ZZL +; ZZ0 :
(4)
l

We can use the Smith Chart to nd Z (;l). To use the Smith Chart, we have
to normalize all the impedances with respect to the characteristic impedance
of the line. Hence,
ZnL = ZZL = 30 +50j 25 = 0:6 + j 0:5:
(5)
0
We can locate ZnL on the Smith Chart which is the complex ; plane. ;(0)
or v can also be deduced from the Smith Chart. Since ;(z) is given by (4),
at z = ;l, we have
;(;l) = v e;2jl :
(6)
1

At f = 25MHz, and with v = 1:5  108 ms;1 ,  = v=f = 6m. Then


l = 2 l = 3 l. Therefore,
(7)
;(;l) = v e;j 23 l :
2

At z = ;l = ;1m ;(;1) = v e;j 3 . From the Smith Chart, we can read
Zn(;1) = 2:15 ; j 0:3 or Z(;1) = 107:5 ; j15 :
(8)
So, an equivalent circuit for the point A is:
20

Zs

Vs = 10 sin t

Z(1) (107.5j15)

In phasor representation, VS = 10e;j 2 = ;j 10. Hence,


;j 7:9
107
:
5
;
j
15
108
:
54
e
Z
(
;
1)
VA = VS Z + Z (;1) = ;j 10 127:5 ; j 15 = 128:38e;j6:7 e;j90 10V
S
= 8:5e;j91:2 V:
(9)
Since
VA = V (;1) = V0 ej 1 + ;(;1)]
(10)
we can nd Vo from the above. Once Vo is found, we can nd VB from
VB = V (0) = Vo 1 + v ]:
(11)
j1
j0.5

j2

j0.2

zL
o

120
0

0.2

0.5

(-l)
Z(-l)

j0.2

j0.5

j2
j1

(b) Find ZL from VSWR and dmin using a Smith Chart


The voltage on the transmission line is

V (z) = Vo(e;jz + v e+jz ) = Voe;jz 1 + ;(z)]:

(12)

If V (z) = jV (z)jej(z) , the real time voltage can be written as

V (z t) = <ejV (z)jej(z)ej!t ] = jV (z)j cos  !t + (t)]:

(13)

Hence the amplitude of the real time voltage is proportional to jV (z)j which
is the voltage standing wave pattern.
|V(z)|
Vmax

Rnmax = 2.5
= VSWR

(z) for
voltage min.

z = dmin
Vmin
5/16
dmin

0
load

2.5

Re

toward
load

For example, we may be given that the VSWR = 2:5 on the line, Zo =
75, and dmin = 5=16, in order to nd ZL.
First, we note that jV (z)j / j1 + ;(z)j where ;(z) = v e2jz . Note that
Vmin occurs when ;(z) is purely negative. When z varies, ;(z) traces out a
constant circle on the Smith Chart, since j;(z)j = jv j is independent of z.
Since the j;(z)j circle must intersect the real ; axis at Rn = 2:5 since the
VSWR= 2:5, we can deduce that magnitude of j;(z)j = jv j. Since z = ;dmin
point corresponds to ;(z) as shown above, and the load is 5=16 from the
dmin point, we can gure out v 's location on the Smith Chart. We can read
o ZnL = 1:4 + j 1:1 on the Smith Chart. Hence ZL = (105 ; j 82:5).
3

j1
j0.5

j2

j0.2

Z=-dmin
0

0.2

0.5

R n=2.5=VSWR

5
16
j0.2

j0.5

j2
j1

W.C.Chew
ECE 350 Lecture Notes

9. Complex Power on a Transmission Line


Complex Power
Since we are dealing with phasors, it is convenient to dene a complex
power which has an imaginary part as well as a real part. We shall dene
the meaning of complex power.
A complex power is dened as
P~ = V~ I~ 
(1)
i.e. the product of a voltage phasor and a current phasor at a given point. If
V~ = jV~ jej  I~ = jI~jej 
(2)
then
P~ = jV~ jjI~j cos ( V ; I ) + sin ( V ; I )] :
(3)
The corresponding real time voltage and current are
V (t) = jV~ j cos ( !t + V ) I (t) = jI~j cos ( !t + I ) :
(4)
Then, the instantaneous power is
P (t) = V (t)I (t) = jV~ jjI~j cos (!t + V ) cos (!t + V + I ; V )
= jV~ jjI~j cos2( !t + V ) cos (I ; V )
; cos ( !t + V ) sin ( !t + V ) sin (I ; V )]:
(5)
The time average of P (t), dened as
V

Z T
1
hP (t)i = hV (t)I (t)i = Tlim
dtP (t)
!1 T 0
= jV~ jjI~j hcos2 ( !t + V )i cos (I ; V )
; hcos ( !t + V ) sin ( !t + V )i sin (I ; V )]:

Since
we have

(6)

hcos2 ( !t + V )i = 21  hcos( !t + V ) sin(!t + V )i = 0 

(7)

hP (t)i = 21 jV~ I~j cos( I ; V ) :

(8)

Comparing with (3), we see that

hP (t)i = 12 <e P~ ] :

(9)

The imaginary part of the complex power is proportional to the second term
in (5), and hence, the imaginary part of the complex power is proportional to
a part of the instantaneous power that averages to zero. Consequently, the
imaginary part of the complex power is called reactive power. For example,
a purely reactive device dissipates no power on the average, but instantaneous
power is being constantly absorbed and released by a reactive device. The
current and voltage through a reactive device is 90 out-of-phase, and the
complex power is purely imaginary or purely reactive.

Complex Power on a Transmission Line


The voltage on a transmission line could be written as
;

V~ (z) = V0 e;jz + v ejz
= V0e;jz 1 + ;(z)] :
The current on the line could be written as
I~(z) = ZV0 e;jz 1 ; ;(z)] :
0
The complex power is given by

(10)
(11)

P~ = V~ I~ = jVZ j 1 + ;(z)] 1 ; ;(z) ] 

(12)

P~ = jVZ j 1 ; j;(z)j2 + ;(z) ; ;(z) ] 

(13)

P~ = V~ I~ = jVZ j 1 ; jv j2 + j 2=m;(z)] :

(14)

hP (z t)i = 21 <e P~ (z)] = j2VZj (1 ; jv j2) 

(15)

which reduces to
2

or

The time average power, dened to be

for a lossless transmission line. If v = 0, or when the load is matched to the


transmission line, (i.e., ZL = Z0 ), all the power carried in the forward going
2

wave is dumped into the load. Otherwise, part of the power is reected. The
power carried by the forward going wave is
2
hP+i = j2VZj 
(16)
0
and the power carried by the backward going wave is
2
hP;i = j2VZj jv j2 :
(17)
0
Note that hP (z t)i is independent of z because of energy conservation.
hP i = hP+i ; hP;i
(18)
is everywhere the same on the lossless transmission line because the total
power leaving the source all arrive at the load end with no loss on the lossless
transmission line. The transmission line can only absorb reactive power.
Hence, the reactive power in (14) is not a constant of position.

Power Delivered to the Load on a Transmission Line


Z0
Vs

VA

Z0

Zs

ZL

Vs

VA IA
Z(l)

Z(l)
z = l

z=0

To nd the power delivered to the load on a lossless transmission line, we


can rst nd Z (;l) using formula (6.11). Then, we can replace the transmission line circuit with the equivalent circuit for nding VA, and IA. The real
power delivered to Z (;l) would be the same as the real power delivered to
ZL.


2
2
 Z (;l) 2 jVS j2
j
V
j
A

~

P = VAIA = Z  (;l) =  Z + Z (;l)  Z (;l) = jZZ (+;lZ)j(V;Slj)j2 : (19)
s
s
The time-average power delivered to the load is
2
hP i = 21 <e P~ ] = 21 jR + jX R+(;Rl();jVlS) j+ jX (;l)j2 
(20)
s
S
where we have assumed that ZS = RS + jXS , and Z (;l) = R(;l) + jX (;l).
To optimize hP i, with respect to X (;l), we choose X (;l) = ;XS , hence,
2
hP i = 21 jRR(+;lR)j(V;Slj)j2 :
(21)
s
3

The above is maximum when R(;l) = RS . Hence, maximum power is delivered to the load when
Z (;l) = ZS :
(22)

W.C.Chew
ECE 350 Lecture Notes

10. Impedance Matching on a Transmission Line.


We note that when the impedance of a load is the same as the characteristic impedance of the transmission line, there is no reected wave, and all
the forward going power is dissipated in the load. There are various ways to
achieve this impedance matching and we will discuss some of them below.
(a) Quarter-Wave Transformer
A quarter wave transformer, like low-frequency transformers, changes the
impedance of the load to another value so that matching is possible.
Z0

ZT

Z in

ZL
/4

A quarter-wave transformer uses a section of line of characterstic


impedance ZT of 4 long. To have a matching condition, we want Zin = Z0 .
From Equation (6.11) we have

2
Z
Z
L + jZT tan 2
T
(1)
Zin = ZT Z + jZ tan  = Z 
T
L
L
2
since tan l = tan 2 4 = tan 2 = 1. In order for Zin = Z0 , we need that

(2)
ZT2 = Z0 ZL ) ZT = Z0 Zl :
If Z0 and ZL are both real, then ZT is real, and we can use a lossless line
to perform the matching. If ZL is complex, it can be made real by adding a
section of line to it.
Z0

ZT
Zin

Z0
Z1

/4
1

ZL

Example

Given that ZL = (30 + j 40), Z0 = 50, nd the shortest l and ZT so that
the above circuit is matched. Assume that ZT is real and lossless.
We want Z1 to be real and Zin to be Z0 = 50 in order for ZT to be real
and the matching condition satised. We nd that ZnL = 0:6+ j 0:8. In order
to make Zn1 real, the shortest l from the Smith Chart is 8 . Then Zn1 = 3:0,
and Z1 = 150. Since Zin = 50, we need

ZT = ZinZ1 = 50  150 = 86:6


in order for matching condition to be satised.
Note that the quarter wave transformer only matches the circuit at one
frequency. Often time, it has a small bandwidth of operation, i.e., it only
works in the frequencies in a small neighborhood of the matching frequency.
Sometimes, a cascade of two or more quarter-wave transformers are used in
order to broaden the bandwidth of operation of the transformer.
j1
j0.5

j2

Z nL

l= 8

j0.2

0.2

0.5

Z n1

j0.2

j0.5

j2
j1

(b) Single Stub Tuning


Another device for performing matching is a single stub (either shorted or
opened at one end) which is shunted across the transmission line at z = ;d
from the load.
2

ZS

Y(d)
VSWR > 1 ZL

Zin

VS

Shorted
Stub

l, Z0

The location d is chosen so that the admittance Y (;d) looking toward


the load is Y0 + jB (Y0 = Z10 ). The length l of the shorted stub is chosen so
that its admittance is ;jB . Hence, when the stub is connected in parallel to
the transmission line at z = ;d, the impedance Zin = Z0, so that matching
condition is achieved.
A shorted stub has impedance and admittance given by

Zs = jZ0 tan l

(3)

Ys = ;jY0 cot l:

(4)
An open-circuited stub can also be used, and the impedance and admittance
are given by
Zop = ;jZ0 cot l
(5)
Yop = jY0 tan l:
(6)
j1
j0.5

j2

Y(-d)
j0.2

0.216

0.2

Z nL

0.5

Yshort

YnL

j0.2

0.99
j0.5

j2
j1

Example
3

Ystub

Let ZL = (100+ j 85), nd the minimum d and l that will reduce the VSWR
of the main line to 1. Assume that Z0 = 50.
We nd that the normalized load ZnL = 2 + 1:7j as shown on the Smith
Chart. Since this problem involves parallel connections, it is more convenient
to work with admittances. YnL = Z1nL is as shown. When we move toward
the generator, Yn(z) traces out a locus on the Smith Chart as shown. It
intersects the G = 1 circle as shown, after moving through 0:216. Therefore,
d = 0:216.
Now, Yn(;d) = 1 + j 1:4. Hence, Ynstub = ;j 1:4. From the Smith Chart,
we note that the admittance for a short is innity, and is at the right end of
the Smith Chart. To get a Ynstub = ;j 1:4, we move toward the generator for
0:099. Hence, l = 0:099.
Often time, it is not easy to change d, but quite easy to change l. We
note that both in the quarter wave transformer and the single stub tuner, we
have to change 2 parameters for tuning. We can provide these 2 degrees of
freedom by using two stubs, changing their length, but not their positions.
(c) Double Stub Tuning (optional reading)
Both single stub tuning and quarter wave transformer matching require
changing the location of the stub or the transformer. In practice, this is
dicult, and a double stub tuning removes the diculty.
3

Z0

Z0

A
Y1

B
Y2

1, Z 0

2, Z 0

Stub 1

Stub 2

ZL

All possible values of Yn2


by changing 2.
YnL

Rotation by

C2

All possible values of Yn1


by changing 1.

C1
Yn2
All possible values of Yn2 by
transforming from all possible
values of Yn1 by 3.

P
Yn1 = 1
Yn1 Ynstub1

C3
Q

(1) In order to have a matched circuit, we should have Y1 = Y0 so that


Yn1 = 1. However, if we change l1 , the possible values of Yn1 trace out a
circle C1 as shown.
(2) If YnL is as shown, by changing l2 , the possible values of Yn2 trace out a
circle C2 as shown.
(3) When l3 is added, all the possible values of Yn1 at A is transformed to B
by a rotation according to the length of l3. This constitute a circle C3
which is all the possible values of Yn2 obtained from Yn1. There are only
two points, P and Q that the two circles C2 and C3 intersect. If we pick
P , then this point should correspond to the value of Yn2.

Yn2 = Ynl + Ynstub2

(25:1)

We can gure out Ynstub2 and hence the length l2 .


(4) The length l3 rotates the point P to the point R. Then R has the
impedance Yn1 ; Ynstub1 = 1 ; Ynstub1 . We can gure out Ynstub1 from
the Smith Chart and hence the length l1.
(d) Ferranti Eect
Z o = 50
R L = 25

VS
= 10 V
z=

z=0

. Find VSWR on the line, and if l is allowed to vary arbitrarily, nd the
maximum voltage on the line.
5

We can nd VSWR from the Smith Chart or by calculator.


; 50 = ; 1 
P (0) = Pv = 25
25 + 50
3
+ jPv j = 43 = 2:
VSWR = 11 ;
jPv j 23
|V(z)|
Vmax

/2
Vs

Vmin
z=

dmin

The voltage at Z = ;l is always xed to be Vs. Hence, we can see that


jV (z)j on parts of the transmission line can be longer than jVsj. If l is chosen
so that Vs is at Vmin, then

Vmax = VSWR  Vmin = 10 volts  2 = 20 volts:

This amplication of voltage on a line is known as the Ferranti's eect. If


the VSWR on the line is very high, Vmax can be so large that it reaches the
breakdown voltage of the line. This is something one should be cautious of
in designing transmission line circuits.

W.C.Chew
ECE 350 Lecture Notes

11. Lossy Transmission Lines.


When R and G are not zero, we have a lossy transmission line. In this
case,
V (z) = V0(e;z + v e+z )
(1)
where
p
p
 = ZY = (j!L + R)(j!C + G) =  + j:
The current is derived using the telegrapher's equation to be
I (z) = ZV0 (e;z ; v ez )
(2)
0
where
s
r
j!L + R :
Z0 = YZ = j!C
+G
When RL = GC , then Z0 becomes frequency independent, and Z0 =

L
C

. Also,

 12

p 
R
G
(3)
1 + j!L = j! LC 1 + j!L
q
p
From (3), we see that  = R CL = ZR0 while  = ! LC . Since  is
1
is also frequency independent,
frequency independent, and the v = ! = pLC
the transmission line is a distortionless line because any pulse that propagates
on the line will not be distorted. This is because a pulse can be thought of
as a superposition of Fourier harmonics. Each Fourier harmonic is a time
harmonic signal. On a distortionless line, all the Fourier harmonics propagate
at the same velocity and su er the same attenuation. Hence the pulse is not
distorted but only diminished in amplitude.

R
 = j! LC 1 + j!L

 12 

If we divide (1) by (2), we get

where

;(z) 
Z (z) = VI ((zz)) = Z0 11 +
; ;(z)

(4)

;(z) = v e2z :

(5)

Note that (4) also implies that


Z0 = Zn(z) ; 1 :
;(z) = ZZ ((zz)) ;
(6)
+ Z0 Zn(z) + 1
Equations (4) and (6) can be solved using a Smith Chart. However, now we
have
j;(z)j = jv j e2z :
(7)
The amplitude of j;(z)j is diminishing when we move from the load to the
source. From (5), we note that ;(z) ! 0 when z ! ;1, Z (z) ! Z0 when
z ! ;1. Hence, a long lossy transmission line is always matched. The locus
traced out by (7) is a spiral converging on the origin of the Smith Chart when
we move from the load to the source.
Also, the voltage standing wave pattern is given by
jV (z)j = jV0j e;z j1 + ;(z)j :
(8)
A plot of ;(z) and jV (z)j are as shown. Furthermore, we can dene an ad
hoc VSWR given to be
+ j;(z)j = 1 + jv j e2z 
VSWR = 11 ;
(9)
j;(z)j 1 ; jv j e2z
which is dependent on z.

x ZnL

V(z) , VSWR

V(z)

VSWR

Power on a Lossy Line


With V (z) and I (z) given by (1) and (2), one can dene a complex power
on a lossy line to be
P (z) = V (z)I ?(z)
(10)
where
;

V (z) = V0e;z 1 + ;(z) 
(11)
and
;z ;

I (z) = V0Ze 1 ; ;(z) :
(12)
0
Hence,
2
;

j
V
0 j ;z ; ? z ;
P (z ) = Z ? e
1 + ;(z) 1 ; ;? (z) 
(13)
0

which is equal to

2


P (z) = jVZ0?j e;2 z 1 ; j;(z)j2 + 2j =m;(z) :
0

Since j;(z)j = jv j e2 z , we have


2

j
V
0 j ;2 z 
P (z ) = Z ? e
1 ; jv j2 e4 z + 2j =m;(z) :
0
3

(14)
(15)

We see that both the real part and the imaginary part of the complex power
are functions of position. This is because real power is dissipated as the wave
propagates. Hence, the real power at one point is not equal to the real power
at another point.

W.C.Chew
ECE 350 Lecture Notes

12. Transients on a Transmission Line.


When we do not have a time harmonic signal on a transmission line, we
have to use transient analysis to understand the waves on a transmission line.
A pulse waveform is an example of a transient waveform.
We have shown previously that if we have a forward going wave for a
voltage on a transmission line, the voltage is
V

(z t) = f (z ; vt):

(1)

The corresponding current can be derived via the telegrapher's equation


1 f (z ; vt):
I (z t) =
(2)
Z0

If instead, we have a wave going in the negative direction,


V

(z t) = g(z + vt)

then the current from the telegrapher's equations, is


1 g(z + vt):
I (z t) = ;
Z0

(3)
(4)

Hence, in general, if
(z t) = V+(z t) + V;(z t)


I (z t) = Y0 V+ (z t) ; V; (z t) 

(5)
(6)

where Y0 = Z10 , and the subscript + indicates a positive going wave, while
the subscript ; indicates a negative going wave.

(a) Reection of a Transient Signal from a Shorted Termination


Z 0, v
+
V0

z=0

z=L

If we switch on the voltage of the above network at t = 0, the voltage at


z = 0 has the form
V

(z = 0 t) = V0 U (t):

(7)

The voltage on the transmission line is zero initially, the disturbance at t = 0


will create a wave front propagating to the right as t increases.
L
t< v

V(z, t)
V+

V0

z=0

z = vt

z=L

I(z, t)
V0 Y0

I+
v

z=0

z = vt

z=L

When the wave reaches the right end termination, the voltage and the
current wave fronts will be reected. However, the short at z = L requires
that V (z = L t) = 0 always. Hence the reected voltage wave, which is
negative going, has an amplitude of ;V0 . The corresponding current can be
derived from (4) and is as shown.
2

V+(z, t)

V+

t>L
v

V0

z=0

z=L

z
t>L
v

V(z, t)

V0

V
V(z, t) = V+ + V

L
t> v

V0

z=L
I+(z, t)

I+

Y0 V0

z=L

z=0
I

I(z, t)

Y0 V0

z=L
I(z, t)

2Y0 V0
Y0 V0

z=0

z=L

When the signal reaches the source end, it is being reected again. A
voltage source looks like a short circuit because the reected voltage must
cancel the incident voltage in order for the voltage across the voltage source
remains unchanged. Hence the negative going voltage and current are again
reected like a short. Hence, if one is to measure the voltage at z = 0, it will
always be V0 . However, the current at z = 0 will increase indenitely with
time as shown.
I(z = 0, t)

7 Y0 V0
5 Y0 V 0
3 Y 0 V0

3 Y 0 V0
Y0 V0
0

t = 2L/v

t = 4L/v

t = 6L/v

The current will eventually become innitely large because the transmission line will become like a short circuit to the D.C. voltage source. Therefore,
the current becomes innite.
(b) Open-Circuited Termination
If we have an open-circuited termination at z = L, then the current has
to be zero always. In this case, the reected current is negative that of the
incident current such that I (z = L t) = 0 always. For example, if the source
waveform looks like as shown below, the reected waveform will behave as
shown.

VS(t)
V0

t1

0
V(z, t)

t < L/v

V+

V0

z
z=0
I(z, t)
V0 Y0

z=0

z = v(tt1)

z = vt

z=L

I+

z = v(tt1)

t < L/v

z = vt

z=L

I(z, t)

I+

t > L/v

Y0 V0

z
z=L

0
I
V(z, t)

t > L/v

2 V0

V0

V+
I
z

(c) Resistive Termination


We can think of transient signals as superpositions of time harmonic
signals. This is a consequence of Fourier analysis. We see that the voltage
reection coecient is ;1 for a shorted termination for all frequencies. Hence,
the voltage reection coecient is ;1 for a transient signal. By a similar
argument, the voltage reection coecient for an open-circuited termination
is +1.
When the termination is resistive on a lossless transmission line, we recall
that the voltage reection coecient is

; Z0 = RL ; Z0 :
= ZZL +
Z
R + Z
L

(8)

Hence, the reection coecient is frequency independent. All frequency components in a transient signal will experience the same reection. Hence, v is
also the reection coecient for a voltage pulse.
R

Z0 , v

+
V0

Z in

z=0

z=L

Consider, for example, a transmission line being driven via a source resistance R and a load termination R. If R = 12 Z0, let us see what happens
when we turn on the switch.
For t < VL , the transmission line appears to be innitely long to the
source. Hence, Zin looks like Z0 to the source. Hence, VA = Z0Z+0 R V0 = 23 V0
for R = 21 Z0. Hence, we have a wavefront of 23 V0 propagating to the right for
L
t <
V.
6

2L/v > t > L/v

V(z, t)
V+

2 V0/3

4 V0/9
z
z=L
V = 2 V0/9
8 Y0 V0/9

0
2 V0/9
I(z, t)
2 Y0 V0/3

I+

0
I = 2 V0 Y0/9

z=L

For t > VL , a reected voltage wave is generated at the termination and


its amplitude is 23 v V0. v = ; 31 for this termination.
2L/v > t > L/v

V(z, t)
V+

4 V0
9

2 V0 /3
0
2 V0 /9

z=L

V =

I(z, t)
2 Y0 V0 /3

2 V0
8 Y0 V0 /9
9
I+

I = 2 V0 Y0 /9 z = L

For t > 2 VL , a voltage source looks like a short to the transient signal. The
reection from the left is again ; 31 for the voltage and + 13 for the current.
7

14 V0
27

V(z, t)
2 V0 /3

2
V1+ = 3 V0

V2+ = 2 V0/27

z
V = 2 V0 /9

I(z, t)

4 V0
9

26 Y V0
27 0

z=L
8YV
9 0 0

I 1+ = 2 Y0 V0 /3
z
z
=
L
I = 2 Y0 V0 /9

0
I 2+ = 2 Y0 V0 /27

When t ! 1, the voltage and current on the line will settle down to a
steady state. In that case, we have only DC signal on the line, and we need
only to use DC circuit analysis to nd the steady state solution. At DC, the
transmission line becomes rst two pieces of wires, VA = VB = 2RR V0 = 12 V0 .
The current through the circuit is ZV00 . If one is to measure VA as a function
of time, it will look like
VA(t)
2 V0/3

2 V0
3

14 V0/27
V0/2
t

2L/v

4L/v

6L/v

IA(t)

V0 Y0
2 V Y
3 0 0

26 V0 Y0/27

t
0

2L/v

4L/v

6L/v

Transient analysis has important application to computer circuitry. We


note that when we switch on a circuit with a delay line, we do not immediately
arrive at the desired steady state value when we have a transmission line or
a delay line. The settling time depends on the length of the line involved.

W.C.Chew
ECE 350 Lecture Notes

13. Properties of Fields in a Transmission Line.


The eld or wave in a transmission line is TEM (Transmission ElectroMagnetic) because both the H-eld and the E-eld are transverse to the
direction of propagation. If the wave is propagating in the z^-direction, then
both Ez and Hz are zero for such a wave. In such a case, the elds are

E = Es H = Hs


(1)

where we have used the subscript s to denote elds transverse to the direction
of propagation. We can also dene a del operation such that

r = rs + ^
z

@
@z

(2)

where rs is transverse to the z^-direction, and in Cartesian coordinate, it is


rs = x^ @x@ + y^ @y@ . From
(3)
r  H =  @@Et 
or


@
(4)
rs + z^@ z  Hs =  @@Et :
Since rs  Hs points in the z^-direction, z^ @z@  Hs is z^-directed, we have

rs  Hs = 0
(^  Hs) = Es

(5)
(6)

@z

@t

Similarly, from rs  Es = ; @@tHs , we can show that

rs  Es = 0
(^  Es) = ; Hs


@z

@t

(7)
(8)

Equations (5) and (7) shows that the transverse curl of the elds are zero.
This implies that the elds in the transverse directions of a transmission
line resembles that of the electrostatic elds. Furthermore, Equations (6)
and (8) couple the Es and Hs elds together. These two equations are the
electromagnetic eld analogues of the telegrapher's equations.
1









H

Contour
C

I
a
b

A current in a coaxial cable will produce a magnetic eld polarized in the


 direction. From Ampere's Law, we have

or

Hs  = J  =
dl

ds

2

(9)

I

 = I:

(10)

 dH

Hence,

( ):
(11)
2
If we assume that the inner conductor in theH coaxial line is charged up with
the line charge Q in coulomb=m, then from E  n^ ds = Q, we have
 ( z t) =

I z t

2E = Q
or

=

(12)

2 :
R
Since the potential between a and b is ab E d, we have
E

Hence,
Q
V

 d =

2 ln
Q

 
b

(z t) = Q(z t) :
b
2
 ln( )
a
is the capacitance per unit length, and it is
E

The ratio

 ( z t) =

= 2b :
ln( a )
2

(13)

(14)
(15)

(16)

If Es = ^E, Hs = ^H, equations (6) and (8) become


@
@z
@
@z

= ; @@Et 

 = ;

:

@H
@t

Substituting (11) for H and (15) for E, we get


@
2 @ V 
I (z t) = ;
@z
ln( ab ) @ t
and

(17)
(18)
(19)

b
 ln( ) @ I
a
(
z t) = ;
(20)
@z
2 @ t :
This is just the telegrapher's equations derived from Maxwell's equations.
C is given by (16) while the inductance per unit length L is obtained by
comparing (20) with the telegrapher's equations
@

ln( b )
=  2a :
Note that the velocity of the wave on a transmission line is
1 = 1 
v = p
L

LC



(21)
(22)

which is independent of the dimensions of the line. This is because all TEM
waves have velocity given by p1 .

W.C.Chew
ECE 350 Lecture Notes

14. Skin Depth and Plane Wave in a Lossy Medium.


We learn earlier that in a lossy medium, J = E, and from

r  H =  @@tE + J =  @@tE + E:

(1)

r  H = j!E + E = j!E

(2)


= ;j 
!

(3)

r  E = ;j!H

(4)

Using phasor technique, we can convert the above to


where

is the complex permittivity. Furthermore, using that


and that r  H = 0, r  E = 0, we can show that

r2 E = ;!2E
(5)
2
2
(6)
r H = ;! H:
Refer to x 4 for details]. If we assume that E = x^Ex(z), then, we can show
that

where

d2
Ex(z ) ;  2 Ex(z ) = 0
2
dz

(7)

 = j!  = + j
:

(7a)

Ex(z ) = c1 e;z + c2 ez :

(8)

The general solution to (7) is of the form


If we assume that c2 = 0, we have only

Ex(z ) = c1 e;z :

(9)

We can convert the above into a real time quantity using phasor techniques,
or
Ex(z t) = jc1 j <ee;z;jz+j1+j!t ]
= jc1j e;z cos(!t ;
z + 1 )
1

(10)

where we have assumed that c1 = jc1 j ej1 . Hence, we see that Ex(z t) is
a wave that propagates to the right with velocity v = ! and attenuation
constant . We can nd from equation (7a), and
r 
r 


 = + j
= j!   ; j
=
j!  1 ; j
:
!
!

(11)

The rst term on the RHS of (1) is the displacement current term, while the
second term is the conduction current term. From (2), we see that the ratio

! is the ratio of the conduction current to the displacement current in a lossy
medium. ! is also known as the loss tangent of a lossy medium.
(i) When !  1, the loss tangent is small, and the conduction current compared to the displacement current is small. The medium behaves more
like a dielectric medium. In this case, we can use binomial expansions to
approximate (11) to obtain


p

1
 = j!  1 ; j
= 1

2 !

where


p
+
j! 


1  
= !p:
= 
2 

(12)

(13)

(ii) When !  1, the loss tangent is large because there is more conduction current than displacement current in the medium. In this case, the
medium is conductive. According to equation (11), when !  1, we
have
r


 = j! ;j
!

r
p
= j! = (1 + j ) ! :

Hence
=
=

1:
=
2

!

(14)

(15)

If we substitute =
= 1 into (10), we have
Ex (z t) = jc1 j e

;z


cos

z
!t ;

+ 1 :

(16)

Ex(z,t), t=0.,0.5,1.0,1.5,2.0,2.5
1
0.8
0.6
1.5
0.4

2.

Ex(z,t)

0.2
2.5

1.

.5
t=0.

0.2
0.4
0.6

omega=1, delta=0.2, phi=0.25pi

0.8
1
0

0.1

0.2

0.3

0.4

0.5
z

0.6

0.7

0.8

0.9

This signal attenuates to e;1 of its original strength at z = . Hence


is also known as the penetration depth or the skin depth of a conductive
medium. For other media, the penetration is 1 , but for a conductive medium,
it is
r
r
2
1 :
=
(17)
=
!
f
This skin depth decreases with increasing frequencies and increasing conductivities.

(iii) When !  1, it is a general lossy medium, and we have to resort to


complex arithmatics to nd and
.
If we square (11), we have
2 ;
2 + 2j
= ;! 2 ( ; j

or

2 ;
2 = ;! 2 
2
= !:


)
!

Squaring (19a) and adding the square of (19b) to it, we have


( 2 ;
2 )2 + (2
)2 = ( 2 +
2 )2 = !42 2 + !22 2
or
p
2 +
2 = ! ! 2 2 +  2 :
Combining with (19a), we deduce that
1 p
2 = (! ! 2 2 +  2 ; ! 2 )
2
3

(18)
(19a)
(19b)
(20)
(21)
(22a)

1 (!p!22 + 2 + !2)
2
Notice that when  = 0, = 0.

2 =

(22b)

W.C.Chew
ECE 350 Lecture Notes
15. Group and Phase Velocities.

If we have two waves that are slightly dierent in frequency ! and phase
constant  , a linear superposition of them is still a solution of the wave
equation
Ex = E0 cos(!1t ; 1 z) + E0 cos(!2t ; 2 z):
(1)
If !1 = ! ; !, 1 =  ;  , !2 = ! + !, 2 =  +  , then
Ex = E0 cos !t ; z ; ( !t ; z)] + E0 cos !t ; z + ( !t ; z)]: (2)
Using the fact that cos(A ; B ) + cos(A + B ) = 2 cos A cos B , we have
Ex = 2E0 cos(!t ; z) cos( !t ; z)
(3)
or
 




!

!
Ex(z t) = 2E0 cos   t ; z cos   t ; z :
(4)
At t = 0, we have Ex = 2E0 cos z cos z which is sketched below.
Ex(z,t), t=1., omega=0.5, beta=0.4, domega=0.01, dbeta=0.02
2

1.5

Ex(z,t)

0.5

0.5

1.5

2
0

50

100

150
z

200

250

300

The rst factor in (4) is rapidly varying while the second factor is slowly
varying. The slowly varying term amplitude-modulates the rapidly varying
term giving rise to the picture as shown.
We have learnt that a function of the form f (vt ; z) propagates in the
positive z-direction with velocity v. From (10.5), we see that the rapidly
1

varying term propagates with velocity ! . Since this represents the propagation of the phases in the rapidly oscillating part in the gure, this is also
known as phase velocity,
vp = ! :
(5)
The slowly varying part propagates with the velocity ! , which is d!
d in the
limit that ! and  ! 0. This represents the velocity on the envelope in
the picture and hence, it is known as the group velocity,
;1 d :
vg = d!
or
v
g =
d
d!

(6)

If  = !p, the phase velocity vp = ! = p1 , the group velocity from (6)
is also p1 . Hence, the group and the phase velocities are the same is  is a
linear function of !.
If  is not a linear function of !, then, the phase velocity and the group
velocities are functions of frequencies, and the medium is known to be dispersive. In a dispersive medium, a pulse propagates with subsequent distortions
because the dierent harmonics in the pulse propagate with dierent phase
velocity. pExample of a dispersive medium is a conductive medium where
 = 1 = !
2 , is not a linear function of ! .
In a distortionless line, the phase velocity is made to be frequency independent so that a pulse propagates without distortions.
Furthermore, a phase velocity can be larger than the velocity of light
while the group velocity is always less than the speed of light. This is because
energy propagates with the group velocity so that special relativity is not
violated.

W.C.Chew
ECE 350 Lecture Notes

16. Real Poynting Theorem.


Since E  H has the dimension of watts=m2 , we can study its divergence
property and its conservative property. Using the vector identity in (1.26),
we have,
r  (E  H) = H  r  E ; E  r  H:
(1)
From Maxwell's equations, we can replace r  E by ; @@tB and r  H by
@ D + J. Hence,
@t

r  (E  H) = ;H @@tB ; E  @@tD ; E  J
= ;H  @ H ; E  @ E ; E  J:

We can show that


Hence,

@t

1 @ jHj2 = H  @ H :
2 @t
@t

@t

(2)
(3)

1

@
1
2
2
r  (E  H) = ; @t 2  jHj + 2  jEj ; E  J:
(4)
We can dene
S = E  H Poynting vector (Power Flow Density watt m;2) (5)
(6)
UH = 21  jHj2 Magnetic Energy Density (joule m;3)
UE = 21  jEj2 Electric Energy Density(joule m;3)
(7)
E  J = Energy Dissipation Density(watt m;3):
(8)
UH and UE represent the energy stored in the magnetic eld and electric eld
respectively. Alternatively, (4) becomes
r  S = ; @t@ (UH + UE ) ; E  J:
(9)

Using the divergence theorem, (9) can be written in integral form,


I
Z
Z
@
S  n^ dA = ; @t (UH + UE ) dV ; E  J dV:
(10)
A
V
V

S
A

The equation says that the LHS will be positive only if there is a net
outow of the ux due to the vector eld S. If there is no current inside V so
that E  J = 0, then this is only possible if the stored energy UH + UE inside
V decreases with time.

2
If J = E, then the last term
H is ;  jEj dV is always negative. Hence,
the last term tends to make S S  n^ dA negative, because energy dissipation
has to be compensated by power ux owing into V . The Poynting theorems
(9) and (10) are statements of energy conservation. For example, for a plane
wave,
r
(11)
E = x^f (z ; vt) H = y^  f (z ; vt)

then

Also,
Therefore,

r
S = E  H = z^  f 2(z ; vt):

(12)

UE + UH = 21 f 2(z ; vt) + 12 f 2(z ; vt) = f 2(z ; vt)

(13)

S = z^ p1 f 2 (z ; vt) = z^v(UE + UH ):

(14)

Hence, the velocity times the total energy density stored equals the power
density ow in a plane wave.

W.C.Chew
ECE 350 Lecture Notes

17. Complex Poynting Theorem.


The complex Poynting vector is dened to be

S = E  H:

(1)

It has the dimension of watt=m2 and it denotes the


ow of complex power.
(We have used underbars to denote complex vectors).
Before we proceed further, let us look at Maxwell's equations for the
phasor eld. In phasor representation, Maxwell's equations become

r  H = J + j!E
r  E = ;j!H:

(2)
(3)

First, we study the divergence property of (1),

r  (E  H) = H  r  E ; E  r  H :

(4)

Substituting (2) and (3) into (4), we have

r  (E  H ) = ;j!H  H + j!E  E ; E  J
= ;j! jHj2 ;  jEj2 ] ; E  J:

(5)

Comparing with (16.4), (5) involves the dierence of the stored energy terms
rather than the sum.
We have shown that for two quantities,

A(z t) = <eA(z)ej!t ]
B(z t) = <eB(z)ej!t ]:
The time average of A(z t)B (z t), denoted by hA B i is given by
Therefore,

(6)
(7)

hA B i = 21 <eA(z)B(z)]:

(8)

hSi = hE  Hi = 21 <eE  H ] = 21 <eS]:

(9)

The imaginary part of S corresponds to instantaneous power that time averages to zero. It is also known as the reactive power. We can also convert (5)
into integral from using the divergence theorem,
I

(E  H )  n^ dA = ;j!  jHj ;  jEj ] dV ;




 jEj2 dV

(10)

where we have assumed that J = E. If , , and  are all real, then
I

and

<e(E  H )  n^ dA = ;  jEj2 dV




=m(E  H )  n^ dA = ;!  jHj2 ;  jEj2] dV:




(11)

(12)

We see that the real part of the power corresponds to power dissipated in
V while the imaginary part of the power corresponds to dierence in the
magnetic energy stored and the electric energy stored. Hence, if a system
has equal amount of magnetic and electric energy stored, it does not consume
any reactive power.

Example of Reactive Power


Ig

I2
I1

Vg

We notice that in the complex Poynting theorem, the reactive power is


proportional to !(jHj2 ; jEj2 ). It is zero when jHj2 = jEj2, or when
the stored magnetic eld energy equals the stored electric eld energy. To
comprehend this further, we look at a simple LC circuit driven by a timeharmonic voltage source.
p
At the resonant frequency of the tank circuit, ! = 1= LC , its input
impedance is innite, and hence Ig = 0. Therefore, there is no power delivered from the generator, be it real or reactive. However, I1 = ;I2 6= 0 at
resonance, and as the tank circuit is resonating, the electric eld energy stored
in C is being converted into the magnetic eld energy stored in L. Therefore,
2

LjI j2 = 12 C jV j2 can be easily veried for a resonating tank circuit. This is


precisely the case mentioned above.
Away from resonance,
1 ) = j!CV (1 ; 1 ):
Ig = Vg (j!C + j!L
g
!2LC
Ig is at 90 out-of-phase with Vg , and the complex power, Vg Ig is purely
imaginary. This implies that there is no time average power delivered by the
source Vg , but it delivers nonzero reactive power. Away from resonance, the
magnetic and electric stored energies are not in perfect balance with respect
to each other, and we need to augment the system with external reactive
power.
1
2

W.C.Chew
ECE 350 Lecture Notes
18. Wave Polarization.

We learnt that

E = x^Ex = x^E1 cos(!t ; z )

(1)

is a solution to the wave equation because r  E = 0. Similarly,


E = y^Ey = y^E2 cos(!t ; z + )

(2)

is also a solution to the wave equation. Solutions (1) and (2) are known as
linearly polarized waves, because the electric eld or the magnetic eld are
polarized in only one direction. However, a linear superposition of (1) and
(2) are still a solution to Maxwell's equation
E = x^Ex(z t) + y^Ey (z t):

(3)

If we observe this eld at z = 0, it is


E = x^E1 cos !t + y^E2 cos(!t + ):

(4)

Ex = E1 cos !t Ey = E2 cos(!t + 90)

(5)

When !t = 0 Ex = E1 Ey = 0:


E1  E = ; pE2 :
When !t = 45 Ex = p
y
2
2

When !t = 90  Ex = 0 Ey = ;E2:
E1  E = ; pE2 :
When !t = 135 Ex = ; p
y
2
2

When !t = 180  Ex = ;E1  Ey = 0:

(6)
(7)
(8)
(9)
(10)

When  = 90,

If we continue further, we can sketch out the tip of the vector eld E. It
traces out an ellipse as shown when E1 6= E2. Such a wave is known as an
elliptically polarized wave.
1

y
t=225

t=270
t=315

E2

E1

t=0
E1

t=180
t=135

E2 t=90

t=45

When E1 = E2, the ellipse becomes a circle, and the wave is known as
a circularly polarized wave. When  is ;90 , the vector E rotates in the
counter-clockwise direction.
A wave is classi ed as left hand elliptically (circularly) polarized when
the wave is approaching the viewer. A counterclockwise rotation is classi ed
as right hand elliptically (circularly) polarized.
When  6= 90 , the tip of the vector E traces out a tilted ellipse. We
can show this by expanding Ey in (5).

Ey = E2 cos !t cos  ; E2 sin !t sin 


"  2 #
E
x
= E2 Ex cos  ; E2 1 ; E
sin :
E
1
1
1
2

(11)

Rearranging terms, we get

where

aEx2 ; bExEy + cEy2 = 1

(12)

1  b = 2 cos   c = 1 :
a = E 2 sin
2

E1E2 sin2 
E22 sin2 
1

(13)

Equation (12) is of the form

ax2 ; bxy + cy2 = 1


which is the equation of a tilted ellipse.
2

(14)

(x, y)

(x, y)

y
B

The equation of an ellipse in its self coordinate is


 0 2

x +  y0 2 = 1
A
B

(15)

where A and B are the semi-axes of the ellipse. However,

x0 = x cos  ; y sin 
y0 = x sin  + y cos 

(16)
(17)

we have
 2


 2

2 
2
sin

1
1
cos

2 cos 
2 sin 
x A2 + B 2 ; xy sin 2 A2 ; B 2 + y A2 + B 2 = 1:
(18)
Equating (14) and (18), we can deduce that


E1 E2 
 = 21 tan;1 2 cos
2
E2 ; E12


where

+
AR = 11 ;



 12

(19)
(20)

1

2
2 2
2
2 sin 
 = 1 ; 4EE1 E
:
(21)
2
2
1 + E2
AR is the axial ratio which is the ratio of the two axes of the ellipse. It is
de ned to be larger than one always.

W.C.Chew
ECE 350 Lecture Notes
19. Representation of a Plane Wave.

When r  E = 0, the electric eld satises the wave equation


2
2
r E +  E = 0
(1)
where  2 = !2. We have learnt that one of the many possible solutions to
the above equation is
E = x^E0 e;jz :
(2)
;
jz
The expression e , when viewed in three dimensions, has constant phase
planes or wave fronts which are orthogonal to the z-axis.
Constant phase
planes

0
z

To denote a plane wave propagating in other directions, we write it as


E = a^E0 e;jxx;jy y;jz z 
(3)
where a^ is a constant unit vector, and E0 a constant. If we substitute (3)
into (1), we obtain
;x2 ; y2 ; z2 +  2 ]E0 = 0:
(4)
In order for (3) to satisfy (1) and that E0 6= 0, we require that
x2 + y2 + z2 =  2 = !2:
(5)
If we dene a vector  = x^x + y^y + z^z , and r = x^x + y^y + z^z, then (3)
can be written as
E = a^E0 e;j r 
(6)
where the magnitude of  is
2
2
2 1
j j = x + y + z ] 2 = :
(7)
1

Equation (6) is a concise way to write a solution to (1). Since r  E = 0 using


(3), we note that
E = ;j ^xx + y^y + z^z ]  a^E0 e;j r :
Therefore, in order for r  E = 0, we require that

(8)

r 

  a^ = 0:
(9)
;
j


r
To explore further how the function e
look like, we assume  to be
pointing in a direction as shown in the gure. The value of   r is constant
on a plane that is orthogonal to .

r
r
A

Constant phase
planes

That is

  r = jj jrj cos  =  (OA)


(10)
for all r on the plane S that is orthogonal to . Hence, S is the constant
phase plane of e;jr = e;j(OA). As one moves progressively in the  direction, the function e;jr has a phase that is linearly decreasing with distance.
Therefore, e;jr denotes a plane wave that is propagating in the  direction.
When  is pointing in the z-direction, such that  = z^ , then e;jr = e;jz ,
which is our familiar solution of a plane wave propagating in the z-direction.

is

An example of a plane wave electric eld satisfying Maxwell's equations


E = y^E0 e;jxx;jz z 

(11)
where x + z =  . The corresponding magnetic eld can be derived using
Maxwell's equations.
r  E = ;j!H:
(12)
Hence,
 @

;1
@
H=
j! z^@x Ey ; x^ @z Ey
2

E0 e;jxx;jz z :
= (^z x ; x^z ) !
2

(13)

In general, when r operates on a plane wave phasor described by e;jr,


it transforms into ;j . This is obvious also from Equation (8). Therefore,
from (12), we can express
1   E:
H=
(14)
!

Therefore, H is orthogonal to both E and , or that H  E = 0, and that


H   = 0, in addition to E   = 0. Furthermore, E  H points in the direction
of . Therefore, for a plane electromagnetic wave, E, H, and  form a righthanded orthogonal system. It is also a transverse electromagnetic (TEM)
wave.
E

W.C.Chew
ECE 350 Lecture Notes

19a. Reection and Transmission of a Simple Plane Wave O an


Interface.
We have learnt that in an innite free space, a simple plane wave solution
exists that is given by

E = ^ x( ) = ^
H = ^ y( ) = ^

;j0 z 

xE

xE0 e

yH

y H0 e

;j0 z

(1)

= y^ E 0 e;j0z 
0

where 0 = 0 =0 is the intrinsic impedance, and 0 = !p0 0 is the


wavenumber. Also, 0 = 2= 0 where 0 is the free space wavelength.
p

Region 1
1, 1

Region 0
0, 0

When the simple plane wave is normally incident on a at material interface, we expect to have a re ected wave in Region 0, and a transmitted wave
in Region 1.
In Region 0, we can write the total elds as

E0 = ^

H0 = ^
y

;j0z + E ;e+j0 z  

E0 e

E0
0

;j0z ;
e

In Region 1, the total elds are

E0 = ^

E0

0

;j1 z 

xE1 e

+j0 z


:

(3)
(4)

H0 = ^ 1 ;j z
(5)
1
p . There are two unknowns in the
1 and 1 =
1 1
;
+
+
x

(2)

E


where 1 = 1=

!
 
above expressions, E0 and H0 . E0 is known because it is the amplitude
1

if the incident eld. We can set up two equations to nd two unknowns by
matching boundary conditions at z = 0. The requisite boundary conditions
are that the tangential components of the E eld and H eld should be
continuous.
By imposing tangential E continuous, we arrive at
+

E0

+ E0; = E1+

(6)

whereas imposing tangential H conditions yields


+

E0
0

E0

0

E1
1

(7)

Solving these two equations expresses E0; and E1+ in terms of E0+:
;=

1

0

+ 0
21 E +:
;
E1 =
0
1 + 0
We dene the re ection coecient to be
E0

1

E0 

; 0 
; = 1 +

1

(8)
(9)
(10)

and the transmission coecient to be


T

=  2+1
1

(11)

Notice that 1 + ; = T .
When there is a mismatch at the interface, we expect most of the wave
to be re ected. This occurs when 1  0 . In this case, ; ' ;1, and T ' 0.
It also occurs when 1  0 , for which case, ; ' +1, T ' 2.
The above derivation also holds true when Region 1 is a conductive lossy
region. In this case, we replace 1 with a comlex permittivity ~1 which is
given by

~1 = 1 ; j !1 :
(12)
p

Then 1p= 1=~ where 1 would be a complex number. Also, j 1 becomes


1 = j !
1 
~1 = 1 + j 1 which is a complex number also.
For a highlypconductive medium like copper, 1=!  1 , ~1 ' ;j 1=!,
and 1 = (1 + j ) !1 =(2 1). Consequently, 1  0 and ; ' ;1, T '= 0.

W.C.Chew
ECE 350 Lecture Notes
Date:November 7, 1997

20. Reections and Refractions of Plane Waves.


Hr
Ei

Er
medium 1

Hi
i

1 , 1



2 , 2

medium 2
z

t
Ht

Et

Perpendicular Case (Transverse Electric or TE case)


When an incident wave impinges on a dielectric interface, a re
ected wave
as well as a transmitted wave is generated. We can express the three waves
as
Ei = y^E0e;j r
(1)
;
j  r
Er = y^?E0e 
(2)
;
j  r
Et = y^?E0e :
(3)
The electric eld is perpendicular to the xz plane, and i, r , and t are their
respective directions of propagation. The 's are also known as propagation
vectors. In particular,
i = x^ix + z^iz 
(4)
r = x^rx ; z^rz 
(5)
tx = x^tx + z^tz :
(6)
Since Ei and Er are in medium 1, we have
(7)
ix2 + iz2 = 12 = !21 1 
2
2
2
2
rx + rz = 1 = ! 1 1 
(8)
i

and for Et in medium 2, we have


tx2 + tz2 = 22 = !222 :
(9)
(7), (8), and (9) are known as the dispersion relations for the components
of the propagation vectors. From the gure, we note that
ix = 1 sin i iz = 1 cos i
(10)
rx = 1 sin r  rz = 1 cos r 
(11)
tx = 2 sin t tz = 2 cos t :
(12)
To nd the unknown ? and ?, we need to match boundary conditions for
the elds at the dielectric interface. The boundary conditions are the equality
of the tangential electric and magnetic elds on both sides of the interface.
The magnetic elds can be derived via Maxwell's equations.
H = r  Ei = i  Ei = (^z  ; x^ ) E0 e;j r:
(13)
i

Similarly,

;j!1

!1

ix

iz

!1

?E0 e;j  r 
Hr = (^z rx + x^rz ) !

(14)

?E0 e;j  r:


Ht = (^z tx ; x^tz ) !

(15)

Continuity of the tangential electric elds across the interface implies


E0e;j x + ?E0e;j x = ?E0e;j x:
(16)
The above equation is to be satised for all x. This is only possible if
ix = rx = tx = x :
(17)
This condition is known as phase matching. From (10), (11), and (12), we
know that (17) implies
1 sin i = 1 sin r = 2 sin t:
(18)
The above implies that r = i. Furthermore,
p  sin = p  sin :
(19a)
1 1
i
2 2
t
rx

ix

If we dene a refractive index ni =

q 

i i

0 0

tx

, then (19a) becomes

n1 sin i = n2 sin t

(19b)
which is the well known Snell's Law. Consequently, equation (16) becomes
1 + ? = ?:
(20)
2

From the continuity of the tangential magnetic elds, we have

E0 +  ?E0 = ; ?E0 :
;iz !
rz
tz
!
!
1

(21)

Since r = i, we have iz = rz . Therefore, (21) becomes


1 ; ? = 1 tz ?:
2

(22)

iz

Solving (20) and (22), we have

; 1tz 
? = 2iz +
1tz
2 iz
2


? =   +2 iz  :
2 iz
1 tz

(23)
(24)

Using (10), (11), and (12), we can rewrite the above as


cos i ;
1 cos t 
? =

2 cos
i +
1 cos t
2
2

? =
cos 2 +cos
icos :
2

(25)
(26)

If the media are non-magnetic so that 1 = 2 = 0 , we can use (19) to


rewrite (25) as
q  2

2 cos i ;
1 1 ; 12 sin i
q
? =
:
(27)

2 cos i +
1 1 ; 12 sin2 i

If 12 sin i > 1, which is possible if 12 > 1, when i < 2 , then ? is of the
form
; jB 
(28)
? = AA +
jB
which always has a magnitude of 1. In this case, all energy will be re
ected.
This isqknown as a total internal reection. This occurs when i > c
where 12 sin c = 1. or

= sin;1

r
2
1  2 < 1 :

(29)

When i = c, t = 90 from (19). The gure below denotes the phenomenon.
less than
critical angle
larger than
critical angle

at critical angle
x
t
less than critical angle

When i > c, tz = 22 ; 12 sin2 i, or

 

p
1
2
tz = ! 0 2 1 ;  sin i :
1
2

(30)

The quantity in the parenthesis is purely negative, so that

tz = ;j tz 

(31)

a pure imaginary number. In this case, the electric eld in medium 2 is

Et = y^?E0e;j x; z :
x

(32)

tz

The eld is exponentially decaying in the positive z direction. We call such


a wave an evanescent wave, or an inhomogeneous wave as opposed to
uniform plane wave. The magnitude of a uniform plane wave is a constant
of space while the magnitude of an evanescent wave or an inhomogeneous
wave is not a constant of space. The corresponding magnetic eld is
?E0 e;j x; z :
Ht = (^z x + x^j tz ) !
x

tz

(33)

The complex power in the transmitted wave is

j2 jE0j2 e;2 z :
S = Et  Ht = (^xx + z^j tz ) j?!

(34)

tz

We note that S x is pure real implying the presence of net time average power

owing in the x^-direction. However, S z is pure imaginary implying that the


power that is
owing in the z^-direction is purely reactive. Hence, no net time
average power is
owing in the z^-direction.

Parallel case (Transverse Magnetic or TM case)


In this case, the electric eld is parallel to the xz plane that contains the
plane of incidence.
Ei
Hi

Hr

i
r

Er

medium 1
1 , 1



i

as

2 , 2

medium 2
Et

Ht

The magnetic eld is polarized in the y direction, and they can be written

Hi = y^ E
0 e;j r
i

Hr = ;y^k E
0 e;j r
r

Ht = y^k E
0 e;j r:
t

(35)
(36)
(37)

We put a negative sign in the denition for k to follow the convention of


transmission line theory, where re
ection coecients are dened for voltages,
and hence has a negative sign when used for currents. The magnetic eld is
the analogue of a current in transmission theory.
5

In this case, the electric eld has to be orthogonal to  and y^, and they
can be derived using
 Hi
Ei = ; i!
to be

Ei = y^  i E0e;j r = (^xiz ; z^ix ) E 0 e;j r


i

Er = (^xrz + z^rx) kE0 e;j r

(38)
(39)

Et = (^xtz ; z^tx ) kE0 e;j r:

(40)

Imposing the boundary conditions as before, we have


1 + k = tz 1 k
2
1

(41)

iz

1 ; k =

k:

(42)

The above can be solved to give

and

; 2 iz =
2 cos t ;
1 cos i 
k = 1 tz +
1 tz
2 cos t +
1 cos i
2 iz

(43)

k =  2+2 iz 

2 =
cos2
2 +cos
icos :
2 iz
1 tz 1
2
t
1
i

(44)

In (43), k will be zero if

22 cos2 t =
12 cos2 i:
(45)
Using Snell's Law, or (19), cos2 t = 1 ;   sin2 i, and (45) becomes
1 ; 11 sin2 i = 12 cos2 i:
(46)
2 2
2 1
1 1
2 2

Solving the above, we get


sin i =

1 ; 21 12

1 1
2 2

! 12

;  

1 2
2 1

(47)

Most materials are non-magnetic in this world so that  = 0, then

r 
sin i =  +2  :
2
1
6

(48)

The angle for i at which k = 0 is known as the Brewster angle. It is


given by
r 
r
2
;
1
;
1
ib = sin  +  = tan 2 :
(49)
2

At this angle of incident, the wave will not be re


ected but totally transmitted. Furthermore, we can show that
sin2 ib + sin2 tb = 1

(50)

implying that

(51)
ib + tb = 2 :
On the contrary,? can never be zero for  = 0 or non-magnetic materials.
Hence, a plot of k as a function of i goes through a zero while the plot of
j?j is always larger than zero for non-magnetic materials.
, or

1 < 2

90

At normal incidence, i.e., i = 0 ? = k since we cannot distinguish


between
perpendicular and parallel
 
  polarizations. When i = 90, j?j =
k = 1. On the whole, j?j  k for non-magnetic materials.
The above equations are dened for lossless media. However, for lossy
media, if we dene a complex permittivity  =  ; j ! , Maxwell's equations
remain unchanged. Hence, the expressions for ?, ?, k, and k remain the
same, except that we replace real permittivities with complex permittivities.
q
For example, if medium 2 is metallic so that  ! 1, then,
2 = 22 ! 0,
and ? = ;1, and ? = 0. Similarly, k = ;1 and k = 0.

W.C.Chew
ECE 350 Lecture Notes

21. Innite Parallel Plate Waveguide.

x
y
z

x=0

x=b

We have studied TEM (transverse electromagnetic) waves between two


pieces of parallel conductors in the transmission line theory. We shall study
other kinds of waves between two innite parallel plates, or planes. We have
learnt earlier that for a plane wave incident on a plane interface, the wave
can be categorized into TE (transverse electric) with electric eld polarized in
the y-direction. Hence, between a parallel plate waveguide, we shall look for
solutions of TE type with E = y^Ey , or TM (transverse magnetic) type with
H = y^@Hy . We shall assume that the eld does not vary in the y-direction so
that @y = 0.
We have shown earlier that if r  E = 0, the equation for the E eld in a
source region is
(r2 + !2)E = 0:
If r  H = 0, the equation for the H eld is
(r2 + !2)H = 0:
Since @y@ = 0, r2 = @x@22 + @z@22 in these two equations.

(1)
(2)

I. TM Case, H = y^Hy .
In this case,

 @2 @2

2
@x2 + @z2 + !  Hy = 0:
1

(3)

If we assume that

Hy = A(x)e;jzz 
(4)
substituting (4) into (3), we have
 d2

2
2
(5)
dx2 + !  z A(x) = 0:
Letting x2 = !2 z2 , (5) becomes

 d2
2
(6)
dx2 + x A(x) = 0
where the independent solutions are
 cos  x
(7)
A(x) = sin  xx :
x
Hence, Hy is of the form
 cos  x
Hy = H0 sin  xx e;jz z 
(8)
x
where
x2 + z2 = !2 =  2 
(9)
which are the dispersion relation for plane waves. We can also dene
x =  cos , z =  sin  so that (9) is automatically satised.
;

To decide a viable solution from (8), we look at the boundary conditions


for the E-eld at the metallic plates. From r  H = j!E, we have
@H ; @H
j!Ex = @y
(10)
z
@z y
(where @y@ Hz = 0 in the above equation) or

 cos  x

z
Ex = ! H0 sin  xx e;jz z 
x

and

@H @H
j!Ez = @x
y
@y x
(where @y@ Hx = 0 in the above equation) or
 sin  x 

x
Ez = j! H0 cos x x e;jz z :
x
;

(11)
(12)

(13)

The boundary conditions require that Ez (x = 0) = Ez (x = b) = 0. Only the


rst solution gives Ez (x = 0) = 0. Hence, we eliminate the second solution,
or
x H sin( x)e;jz z :
Ez = ; j!
(14)
0
x
In order for Ez (x = b) = 0, we require that
sin xb = 0
(15)
or
xb = m  m = 0 1 2 3 : : :
(16)
and consequently,
x = m
(17)
b  m = 0 1 2 3 : : : :
This is known as the guidance condition for the waveguide. Finally, we
have
 ;j z

z
x
(18)
Hy = H0 cos m
b e 
z H cos m x e;jz z 
Ex = !
(19)
0
b
m H sin m x  e;jz z 
(20)
Ez = ; j!b
0
b
where


 m 2  21
2
z = ! 

b
;

(21)

which is the dispersion relation for the parallel plate waveguide. Equation
(18) can be written as
(22)
Hy = H20 ejxx + e;jxx]e;jz z = H20 ejxx;jzz + e;jxx;jzz ]:
The rst term in the above represents a plane wave propagating in the positive
z^-direction and the negative x^-direction, while the second term corresponds
to a wave propagating in the positive x and z directions. Hence, the eld in
between a parallel plate waveguide consists of a plane wave bouncing back
and forth between the two plates, as shown.

= ^x x+ ^z z

= x^ x + ^z z

Since we dene x =  cos , z =  sin , the wave propagates in a direction making an angle  with the x^-direction. Since the guidance condition
requires that x = mb =  cos , the plane wave can be guided only for
discrete values of .
From (21), we note p
that for dierent m's, z will assume dierent values.
When m = 0, z = ! , Ez = 0, and we have a TEM mode. When
m > 0, we have a TM mode of order m we call it a TMm mode. Hence,
there are innitely many solutions to Maxwell's equations between a parallel
plate waveguide with the eld given by (18), (19), (20), and the dispersion
relation given by (21) where m = 0 1 2 3 : : : .

II. Cuto Frequency


From (21), for a given TMm mode, if !p < mb , then z is pure imaginary. In this case, the wave is purely decaying in the z^-direction, and it is
evanescent and non-propagating. For a given TMm mode, we can always
lower the frequency so that this occurs. When this happens, we say that the
mode is cut o. The cuto frequency is the frequency for which a given
TMm mode becomes cuto when the frequency of the TMm mode is lower
than this cuto frequency. Hence,
m
mv :
!mc = bm
or
f
(23)
p
mc = p =

2b  2b
When
(m + 1)v > f > mv > (m ; 1)v > (m ; 2)v > : : : > 0
(24)
2b
2b
2b
2b
the TEM mode plus all the TMn modes, where 0 < n  m are propagating
or guided while the TMm+1 and higher order modes are evanescent or
cuto. For the parallel plate waveguide, there is one mode with zero cuto
frequency and hence is guided for all frequencies. This is the TEM mode
which is equivalent to the transmission line mode.
The wavelength that corresponds to the cuto frequency is known as the
cuto wavelength, i.e.,
mc = fv = 2mb :
(25)
mc

When < mc, the corresponding TMm mode will be guided. You can think
of as some kind of the \size" of the wave, and that only when the \size" of
the wave is less than mc can a wave \enter" the waveguide. Notice that mc
is proportional to the physical size of the waveguide.

IV. TE Case, E = y^Ey .


4

The eld for the TE case can be derived similarly to the TM case. The
electric eld is polarized in the y^-direction, and satises
 @2 @2

2
(26)
@x2 + @z2 + !  Ey = 0:
The elds can be shown in a similar fashion to be
Ey = E0 sin(xx)e;jzz 
(27)
z E sin( x)e;jzz 
Hx = ; !
(28)
0
x
x E cos( x)e;jzz :
Hz = ; j!
(29)
0
x
The boundary conditions are
Ey (x = 0) = 0 Ey (x = b) = 0:
(30)
This gives
x = m
(31)
b 
as before, where x2 + z2 = !2. Hence, the TEm modes have the same
dispersion relation and cut-o frequency as the TMm mode. However, when
m = 0, x = 0, and (27){(29) imply that we have zero eld. Therefore, TE0
mode does not exist. We say that TEm and TMm modes are degenerate
when they have the same cuto frequencies.
We can decompose (27) into plane waves, i.e.,
(32)
Ey = E2j0 ejxx;jzz ; e;jxx;jz z ]
and interpret the above as bouncing waves. Compared to (22), we see that
the two bouncing waves in (32) are of the opposite signs whereas that in (22)
are of the same sign. This is because the electric eld has to vanish on the
plates while the magnetic eld need not.

TM1 mode eld




H-field

E-field

TE1 mode eld


H-field


z

E-field

The sketch of the elds for TM1 and TE1 modes are as shown above.
For the TM mode, Hz = 0, and Ez 6= 0, while for the TE mode, Ez = 0,
and Hz 6= 0. Tangential electric eld is zero on the plates while tangential
magnetic eld is not zero on the plates. The above is the instantaneous eld
plots. E  H is in the direction of propagation of the waves.

III. Phase and Group Velocities.


The phase velocity in the z^-direction of a wave in a waveguide is dened
to be
!
1
vp = ! = h
(33)
1 =

i
 f 2 21 
; m
2 2 p
z
2
!  ; b
 1 ; mc
f
which is always larger than the speed of light for f > fmc. The group velocity
is

; m
2i 12

 d ;1 !2 b
d!
vg = d = d!z =
!
z
which is always less than the speed of light.
;

1;

 f 2  12
mc



TM2 and TE2 modes

=
z

TM1 and TE1 modes

2c

TM0 = TEM mode


1c

vg =
vp = / z

d
d z
z

0c

=C

(34)

i1

Since z = !c 1 ; !!mc 2 2 , a plot of ! versus z is as shown. When


z ! 0, the group velocity becomes zero while the phase velocity approaches
innity. When z ! 1, or ! ! 1, the group and phase velocities both
approach the velocity of light in free-space which is the TEM wave velocity.

W.C.Chew
ECE 350 Lecture Notes

22. Hollow Waveguide.

A hollow cylindrical waveguide of uniform and arbitrary cross-section


can guide waves. The elds inside a hollow waveguide can guide waves of
both TE and TM types. When the eld is of TE type, the electric eld is
purely transverse to the direction of wave propagation z Hence Ez = 0. For
TM elds, the magnetic eld is purely transverse to the z-axis and hence,
Hz = 0. Therefore, the eld components of TE elds are
Ex Ey  Hx Hy  Hz 
and for TM elds, they are
Hx Hy  Ex Ey  Ez :
We can hence characterize TE elds as having Ez = 0 Hz 6= 0, and TM
elds as Hz = 0 Ez 6= 0. Hence, the z-component of the H eld can be used
to characterize TE elds, while the z-component of the E eld can be used
to characterize TM elds in a hollow waveguide. Given Ez , and Hz , it will be
desirable to derive the transverse components of the elds. We shall denote a
vector transverse to z^ by a subscript s. In this notation, Maxwell's equations
become


@
rs + z^ @z  (Hs + z^Hz ) = j!(Es + z^Ez )
(1)


@
(2)
rs + z^ @z  (Es + z^Ez ) = ;j!(Hs + z^Hz )
where rs = x^ @x@ + y^ @y@ , and Es and Hs are the electric eld and the magnetic eld, respectively, transverse to the z directon. Equating the transverse
components in (1) and (2), we have
@ z^  H = j!E 
rs  z^Hz + @z
(3)
s
s
@ z^  E = ;j!H :
rs  z^Ez + @z
(4)
s
s
1

Substituting (4) for Hs into (3), we have




@
j
@
rs  z^Hz + @z z^  ! rs  z^Ez + @z z^  Es = j!Es:
(5)
Using the vector identity
A  (B  C) = B(A  C) ; C(A  B)
(6)
we can show that
z^  rs  z^Ez = rs(^z  z^Ez ) ; z^Ez (^z  rs) = rsEz 
(7)
and
z^  (^z  Es) = z^(^z  Es) ; Es(^z  z^) = ;Es :
(8)
Hence, (5) becomes
j @ r E ; j @ 2 E = j!E :
(9)
rs  z^Hz + !
s
@z s z ! @z2 s
If E is of the form Ae;jz z + Bejz z , then @z@22 = ;z2 and (9) becomes
@

1
Es = !2 ;  2 @z rsEz ; j!rs  z^Hz :
(10)
z
In a similar fashion, we obtain
@

1
(11)
Hs = !2 ;  2 @z rsHz + j!rs  z^Ez :
z
The above equations can be used to derive the transverse components of the
elds given the z^-components. Hence, in general, we only need to know the
z^-components of the elds.

I. Rectangular Waveguides
Rectangular waveguides are a special case of cylindrical waveguides with
uniform rectangular cross section. Hence, we can divide the waves inside the
waveguide into TM and TE types.
y
b

0
2

TM Case, Hz = 0 Ez 6= 0
Inside the waveguide, we have a source free region, therefore
r2 + !2]E = 0

(12)

r2 + !2]Ez = 0:
Equation (13) admits solutions of the form

(13)

or

 sin  x  sin  y
Ez = E0 cos xx cos y y e;jz z 
y
x

since





@ 2 sin x x =  2 sin xx 
x cos  x
@x2 cos x x
x




@ 2 sin y y = ; 2 sin y y  @ 2 e;jz z = ; 2 ejz z :
y cos  y
z
@y2 cos y y
@z2
y
Therefore
(r2 + !2)Ez = (;x2 ; y2 ; z2 + !2)Ez = 0:

(14)
(15)
(16)
(17)

This is only possible if

x2 + y2 + z2 = !2


(18)
which is the dispersion relation. The boundary conditions require that
Ez (x = 0) = 0

Ez (y = 0) = 0:

(19)

Hence, the admissible solution is

Ez = E0 sin(xx) sin(y y)e;jzz :

(20)

Also, we require that

Ez (x = a) = 0

Ez (y = b) = 0:

(21)

This is only possible if sin(xa) = 0 and sin(y b) = 0, or

xa = m m = 0 1 2 : : :

y b = n n = 0 1 2 3 : : : :

However, when m or n = 0, Ez = 0. Hence, we have


n  n  1
x = m

m

1


y=
a
b
3

(22)
(23)

which are the guidance conditions. To get the transverse E and H elds,
we use (10) and (11)
@ @ E = ;jxz E cos( x) sin( y)e;jzz 
Ex = !21;  2 @z
x
y
@x z x2 + y2 0
z
(24)
1
@
@
;
j

Ey = !2 ;  2 @z @y Ez =  2 +xz2 E0 sin(xx) cos(y y)e;jzz 
z
x
y
(25)
@ E = j!y E sin( x) cos( y)e;jzz 
Hx = !2j!;  2 @y
(26)
z
x
y
2 + 2 0
z

@ E = ;j!x E cos( x) sin( y)e;jzz :


Hy = !2;j!
x
y
2
; z @x z x2 + y2 0

(27)

We note that the electric elds satisfy their boundary conditions. From the
dispersion relation (18), we have

r
 2  n 2
z = !2 ; m
a ; b :

(28)

The solution that corresponds to a particular choice of m and n in (23)


is known as the TMmn mode. For a given TMmn mode, z will be pure
imaginary if
 m 2  n 2
2
!  < a + b 
(29)
or
 m 2  n 2 21
1
:
(30)
! < p a + b
In this case, the mode is cuto, and the elds decay in the z^-direction and
become purely evanescent. We dene the cuto frequency for the TMmn
mode to be

 m 2  n 2 12  2  n 2 12
1
!mnc = p a + b
= v m
:
a + b

(31)

The TMmn mode will not propagate if

! < !mnc or f < fmnc


(32)
!
where fmnc = !mnc
2 , f = 2 . The corresponding cuto wavelength is
 m 2  n 2; 12

mnc = 2 a + b
:
(31a)
Only when the wavelength
is smaller than this \size" can the wave \enter"
the waveguide and be guided as the TMmn mode.
4

To nd the power owing in the waveguide, we use the Poynting theorem.
Sz = ExHy ; Ey Hx
(33)
2
!y2 z
!

z
2
x
2
2
= ( 2 +  2 )2 jE0j cos (xx) sin (y y) + ( 2 +  2 )2 jE0j2 sin2 (xx) cos2(y y)
x
y
x
y
z
2
2
2
2
2
2
2
= ( 2!
j
E
(34)
0 j x cos (x x) sin (y y ) + y sin (x x) cos (y y )]:
2
2
x + y )
The total power
Zb Za
2
!z ab jE0j2 :
z ab jE0 j 2
2
(

+

)
=
(35)
Pz = dy dxSz = !
4(x2 + y2 )2 x y
4(x2 + y2 )
0
0
When f < fmnc, z is purely imaginary and the power becomes purely reactive. No real power or time average power ows down a waveguide when all
the modes are cuto.

TE Case, Ez = 0 Hz 6= 0.
In this case,

Hz = H0 cos(xx) cos(y y)e;jzz 


so that from equations (10) and (11), we have,
j! @ H = j!y H cos( x) sin( y)e;jzz 
Ex = ; !2
x
y
; z2 @y z x2 + y2 0
j! @ H = ;j!x H sin( x) cos( y)e;jzz 
Ey = !2
x
y
; z2 @x z x2 + y2 0
@ @ H = jxz H sin( x) cos( y)e;jzz 
Hx = !21;  2 @z
x
y
@x z x2 + y2 0
z
@ @ H = jy z H cos( x) sin( y)e;jz z 
Hy = !21;  2 @z
x
y
@y z x2 + y2 0
z

(36)
(37)
(38)
(39)

(40)
where x2 + y2 + z2 =  2 = !2. Matching boundary conditions for the
tangential electric eld requires that
n  n = 0 1 2 3 : : : :
x = m

m
=
0

1

2

3

:
:
:

(41)
y=
a
b
Unlike the TM case, the TE case can have either m or n equal to zero.
Hence, TEm0 or TE0n modes exist. However, when both m and n are zero,
Hz = H0e;jz z , Hx = Hy = 0, and r  H 6= 0, therefore, TE00 mode cannot
exist.
For the TEmn modes, the subscript m is associated with the longer side
of the rectangular waveguide, while n is associated with the shorter side. In
5

the case of TEm0 mode, y = 0, implying that Ex = 0, Ey 6= 0, Hy = 0,


Hx 6= 0, Hz 6= 0. The elds resemble that of the TEm mode in a parallel
plate waveguide. For the general TEmn mode, the dispersion relation is

 2  n 2
z = !2 ; m
(42)
a ; b :
Hence, the TEmn mode and the TMmn mode have the same cuto frequency
and they are degenerate.

Example: Designing a Waveguide to Propagate only the TE10 mode


The cuto frequency of a TMmn or a TEmn mode is given by

 2  2 12
1
n
:
!mnc = p m
+
a
b

(43)

Usually, a is assumed to be larger than b so that TE10 mode has the lowest
cuto frequency, which is given by
f10c = 2va or
10c = 2a
(44)
where v = p1 , and f10c = !210c . The next higher cuto frequency is either
f20c or f01c depending on the ratio of a to b.
f20c = va  f01c = 2vb :
(45)
If a > 2b, f20c < f01c, and if a < 2b, f20c > f01c. f20c = f01c if a = 2b. When
a = 2b, and we want a waveguide to carry only the TE10 mode between 10
GHz and 20 GHz. Therefore, we want f10c = 10 GHz, and f20c = f01c =
20GHz. If the waveguide is lled with air, then v = 3  108 ms , and we deduce
that
a = 2fv = 1:5cm b = 2fv = 0:75:
(46)
10c
01c
In such a rectangular waveguide, only the TE10 will propagate above 10 GHz
and below 20 GHz. The other modes are all cuto. Note that no mode could
propagate below 10 GHz.

W.C.Chew
ECE 350 Lecture Notes
23. Cavity Resonator.
y
b
d

A cavity resonator is a useful microwave device. If we close o two ends


of a rectangular waveguide with metallic walls, we have a rectangular cavity
resonator. In this case, the wave propagating in the z^-direction will bounce
o the two walls resulting in a standing wave in the z^-direction. For the TM
case, we have

Ez = E0 sin(xx) sin(y y)(e;jzz + ejz z )


(1)
;
j

(2)
Ex =  2 +xz2 E0 cos(xx) sin(y y)(e;jzz ; ejz z )
x
y
Ey = ;2j+y z2 E0 sin(xx) cos(y y)(e;jzz ; ejz z ):
(3)
x
y
For the boundary conditions to be satised, we require that Ex(z = 0) =
Ey (z = 0) = 0. Hence,  = 1, and
Ez = 2E0 sin(xx) sin(y y) cos(z z)
(4)
xz E cos( x) sin( y) sin( z)
Ex = ;22+
(5)
x
y
z
2 0
x y
Ey = ;22+yz2 E0 sin(xx) cos(y y) sin(z z):
(6)
x
y
Furthermore, Ex(z = ;d) = Ey (z = ;d) = 0, implying that
(7)
z = p
d  p = 0 1 2 3 : : : :
The guidance conditions for a waveguide demand that x = ma and y = nb ,
where for TM case, neither m or n can be zero. Now that z has to satisfy
(7), the TM mode in a cavity is classied as TMmnp mode. We note from (4)
1

that p can be zero while Ez 6= 0. Hence, the TMmn0 cavity mode can exist.
In order for (4), (5), and (6) to be solutions to the wave equation, we require
that
 m 2  n 2  p 2
2
2
2
2
!  = x + y + z = a + b + d :
(8)
For a given choice of m, n, and p, only a single frequency can satisfy (8).
This frequency is the resonant frequency of the cavity. It is only at this
frequency that the cavity can sustain a free oscillation. At other frequencies,
the elds interfere destructively and the free oscillation is not sustained. From
(8), we gather that the resonant frequency for the TMmnp mode is


1

 2  n 2  p 2 2
!mnp = p1 m
:
a + b + d
For the TE case, similar derivation shows that
Hz = H0 cos(xx) cos(y y) sin(z z)
y
H cos(xx) sin(y y) sin(z z)
Ex = j!
2 + 2 0
x

(9)
(10)
(11)

x
H sin(xx) cos(y y) sin(z z):
Ey = ; j!
2 + 2 0
x
y
Similarly, the boundary conditions require that
n   = p :
x = m


y=
a
b z d

(12)
(13)

When p = 0, Hz = 0, hence TEmn0 mode does not exist. However, TE0np or


TEm0p modes can exist. The resonant frequency formula is as given in (9).
If a > b > d, the lowest resonant frequency is the TM110 mode. In this case,


1

 2  2 2
!110 = p1 a + b

(14)
and Ez 6= 0, Hx 6= 0, Hy 6= 0, Ex = Ey = 0. A sketch of the eld is as shown.
y
b
H-field
E-field

TM110
mode
z

We can decompose the wave into plane waves bouncing o the four walls
of the cavity.
2

y
b

As an example, for a = 2 cm, b = 1 cm, d = 0:5 cm, the resonant


frequency of the TM110 mode is
s

52 = 3  108 p5Hz


2f110 = 3  108 4(10
;2 )2 2  10;2
or

(15)

p
(16)
f110 = 43  1010  5Hz = 1:68  1010 Hz = 16:8GHz:
Cavity resonators are useful as lters and tuners in microwave circuits, as LC
resonators are in RF circuits. Cavity resonators can also be used to measure
the frequency of an electromagnetic signal.

W.C.Chew
ECE 350 Lecture Notes

24. Dielectric Waveguides (Slab).


When a wave is incident from a medium with higher dielectric constant
at an interface of two dielectric media, total internal reection occurs
when the angle of incident is larger than the critical angle. This fact can
be used to make waves bouncing between two interfaces of a dielectric slab
to be guided
x
0, 0

d/2

1, 1

2, 2

d/2

region 0
z

region 1

region 2

Since total internal reection occurs for both TE and TM waves, guidance
is possible for both types of waves

I. TE Case E = y^Ey
Ey is a solution to the wave equation in each region. In region 0, we
assume a solution of the form
E0y = E0e;j0xx;jz z 
(1)
where
02x + z2 = !200 = 02 :
(1a)
In region 1, we assume a solution of the form
E1y = A1e;j1xx + B1ej1xx]e;jz z 
(2)
where
12x + z2 = !211 = 12 :
(2a)
In region 2, the solution is of the form
E2y = E2ej2xx;jzz 
(3)
where
22x + z2 = !222 = 22 :
(3a)
1

We assume that all the solutions in the three regions to have the same zvariation of e;jz z by the phase matching condition.
In region 1, we have an up-going wave as well as a down-going wave. The
two waves have to be related by the reection coecient ? for the electric
eld at the boundaries. ? is derived earlier in the course. Therefore at
x = d2 , we have
(4)
B1ej1x d2 = 10?A1e;j1x d2 
where 10? is the reection coecient at the regions 1 and 0 interface. At
x = ; d2 , we have
(5)
A1 ej1x d2 = 12?B1e;j1x d2 
where 12? is the reection coecient at the regions 1 and 2 interface. Multiplying equations (4) and (5) together, we have,

A1B1ej1xd = 12?10?A1B1e;j1xd:
(6)
A1 and B1 are non-zero only if
1 = 12?10?e;2j1xd :
(7)
The above is known as the guidance condition of a dielectric slab waveguide. If medium 3 is equal to medium 1, then 12? = 10?, and the guidance
condition becomes
1 = 210?e;2j1xd:
(8)
From before, for a wave incident at an angle ,
cos  ; 1 cos 00 :
10? = 0 cos
(9)
 + 1 cos 00
0
Since 1x = 1 cos , 0x = 0 cos 00 , (9) could be written as
0
1
; 10x :
1 1x ; 0 0x
10? = 0  + 1  = 0 1x +
(10)
10x
0 1x
1 1x 0 0x
Taking the square root of (8), we have

10?e;j1xd = 1:
When we choose the plus sign, B1 = A1 from (4), and from (2)
E1y = 2A1 cos(1xx)e;jz z ) even in x:
When we choose the minus sign in (11) we have B1 = ;A1, and
E1y = ;2jA1 sin(1xx)e;jz z ) odd in x:
2

(11)
(12)
(13)

Multiplying (11) by ej1x d2 and manipulating, we have


 
0  d tan  d = j d even solutions
0x
1 1x 2  1x 2 
2
0  d cot  d = j d odd solutions:
1x
0x
 1x 2
2
2

(14)
(15)

Subtracting (1a) from (2a) and solving for 0x , we have

0x = !2(00 ; 11 ) + 12x] 21 :


(16)
In order for (14) and (15) to be satised, 0x has to be pure imaginary. In
other words, the waves in region 0 and 3 have to be evanescent and decay
exponentially away from the slab. Hence
0x = ;j
0x = ;j !2(11 ; 00 ) ; 12x] 21 
(17)
and (14) and (15) become
s
 
0  d tan  d =
d = !2(  ;   ) d2 ;  d 2 even solutions
1x
1x
1 1
0 0
1 1x 2
2 0x 2
4
2
(18)
s


2
2
; 0 1x d2 cot 1x d2 =
0x d2 = !2(11 ; 00 ) d4 ; 1x d2 odd solutions:
1
(19)
We can solve the above graphically by plotting
 d

d
0
y1 =  1x 2 tan 1x 2 even solutions
(20)
1
 
d

0
(21)
y2 = ;  1x 2 cot 1x d2 odd solutions
1
"
 d 2# 12
2
d
y3 = !2(11 ; 00 ) 4 ; 1x 2
=
0x d2 :
(22)

TE0
/2

ev

y2

odd

y1

od
d

y3

y2

en

y1 even

(1 1 0 0) d/2

TE1

0
1

3
2

1x

d
2

y3 is the equation of a circle the radius of the circle is given by


!(11 ; 00 ) 12 d2 :
(23)
The solutions to (18) and (19) are given by the intersections of y3 with y1 and
y2. We note from (23) that the radius of the circle can be increased in three
ways (i) by increasing the frequency, (ii) by increasing the contrast 10 10 , and
(iii) by increasing the thickness d of the slab.
When 0x = ;j
0x, the reection coecient is




0 1x + j1
0x
;
1 1
0x

(24)
10? =   ; j
= exp +2j tan  
0 1x
1 0x
0 1x
and j10?j = 1. Hence there is total internal reections and the wave is
guided by total internal reections. Cut-o occurs when the total internal
reection ceases to occur, i.e. when the frequency decreases such that
0x = 0.
From the diagram, we see that
0x = 0 when
!(11 ; 00 ) 21 d2 = m
m = 0 1 2 3 : : :
(25)
2 
or
m
!mc =
 m = 0 1 2 3 : : : :
(26)
d(11 ; 00 ) 21
The mode that corresponds to the m-th cut-o frequency above is labeled
the TEm mode. TE0 mode is the mode that has no cut-o or propagates at
all frequencies.
At cut-o,
0x = 0, and from (1a),
z = !p0 0 
(27)
for all the modes. Hence, both the group and the phase velocities are that of
the outer region. This is because when
0x = 0, the wave is not evanescent
outside, and most of the energy of the mode is carried by the exterior eld.
When ! ! 1, 1x ! nd from the diagram for all the modes. From (2a),

q
z = !21 1 ; 12x  !p1 1 

! ! 1:

(28)

Hence the group and phase velocities approach that of the dielectric slab.
This is because when ! ! 1,
0x ! 1, and all the elds are trapped in the
slab and propagating within it.
Because of this, the dispersion diagram of the dierent modes appear as
below.
4

1 1

0 0

TE0

1c

2c

TE3

3c

II. TM Case H = y^Hy


For the TM case, a similar guidance condition analogous to (27) can be
derived
1 = 12k10k e;2j1xd
(29)
where  is the reection coecient for the TM eld. Similar derivations show
that the above guidance condition, for 2 = 0, 2 = 0, reduces to

s
 
0  d tan  d = !2(  ;   ) d2 ;  d 2
1x
1x
1 1
0 0
1 1x 2
2
4
2
s
 2
2
d
d
d

0
;  1x 2 cot 1x 2 = !2(11 ; 00 ) 4 ; 1x d2

even solution
(30)

odd solution:
(31)
Note that for equations (7) and (29), when we have two parallel metallic
plates, k = 1, and ? = 1, and the guidance condition becomes
1 = e;2j1xd ) 1x = m
(32)
d  m = 0 1 2 : : :
which is what we have observed before.
1

W.C.Chew
ECE 350 Lecture Notes

25. Vector Potential - Introduction to Antennas & Radiations


Maxwell's equations are

r  E = ;j!H
r  H = j!E + J
r  H = 0
r  E = :

(1)
(2)
(3)
(4)

Since r  (r  A) = 0, we can let


H = r  A
(5)
so that equation (3) is automatically satised. Substituting (5) into (1), we
have
r  (E + j!A) = 0:
(6)
Since r  r = 0, we have
E = ;j!A ; r:
(7)
Hence, knowing A and  uniquely determines E and H. We shall relate A
and  to the sources J and  of Maxwell's equations. Substituting (5) and
(7) into (2), we have
r  r  A = j!;j!A ; r] + J
(8)
or
r2 A + !2A = ;J + j!r + rr  A:
(9)
Using (7) in (4), we have
r  (j!A + r) = ;  :
(10)

The above could be simplied for the following observation. Equations (5)
and (7) give the same E and H elds under the transformation
A = A + r
(11)
 =  ; j!:
(12)
The above are known as the Gauge Transformation. With the new A
and  , we can substitute into (5) and (7) and they give the same E and H
elds, i.e.
r  A = r  A + r  r = r  A = H
(13)
;j!A ; r = ;j!A ; j!r ; r + j!r = E:
(14)
0

It implies that A and  are not unique. The vector eld A is not unique
unless we specify both its curl and its divergence. Hence, in order to make
A unique, we have to specify its divergence. If we specify the divergence of
A such that
r  A = ;j!
(15)
then (9) and (10) become
r2 A + !2 = ;J
(16)

r2  + !2 = ;  :
(17)

The condition in (15) is also known as the Lorentz gauge. Equations (16)
and (17) represent a set of four inhomogeneous wave equations driven by the
sources of Maxwell's equations. Hence given the sources  and J, we may
nd A and . E and H may in turn be found using (5) and (7). However,
as a consequence of the Lorentz gauge, we need only to nd A  follows
directly from equation (15).

Let us consider the relation due to an elemental current that can be


described by
J = z^Il (r) A=m2
(18)
where Il denotes the strength of this current, and (r) = (x) (y) (z). Equation (16) becomes
r2Az + !2Az = ;Il (r):
(19)
Taking advantage of the spherical symmetry of the problem, r2 has only r
dependence in spherical coordinates, we have
1 d r2 d A + 2 A = ;Il (r)
(20)
z
r2 dr dr z
where 2 = !2. Equations (19) and (20) are similar in form to Poisson's
equation with a point charge Q at the origin,
(21)
r2  = ; Q (r):
We know that (21) has the solution of the form
Q :
 = 4 r
(22)
Hence, we guess that the solution to (20) is of the form
Az = 4Il
(23)
r C (r):
It can be shown that
1 d r2 d f (r) = 1 d2 rf (r):
(24)
r2 dr dr
r dr2
2

Outside the origin, the RHS of (20) is zero, and after using (23) and (24) in
(20), we have
d2 C (r) + 2 C (r) = 0:
(25)
dr2
This gives
C (r) = e jr :
(26)
Since we are looking for a solution that radiates energy to innity, we choose
an outgoing solution in (26). Hence,
jr
(27)
Az (r) = 4Il
r e 
for a source directed at a z^-direction. From (16), we note that A and J
always point in the same direction. Therefore, for a point source directed at
l and located at r instead of the origin, the vector potential A is
l e j r r :
A(r) = 4 jI
(28)
r;rj


j ;


|r r|

By linear superposition, the vector potential due to an arbitrary source


J is
ZZZ

A = 4
dr jrJ;(r r) j e j r r :
(29)
Similarly, we can show that
Z ZZ
1
(r ) e j r r :
 = 4 
dr jr;
(30)
rj
0

j ;

j ;

W.C.Chew
ECE 350 Lecture Notes

26. The Fields of a Hertzian Dipole


A Hertzian dipole is a dipole which is much smaller than the wavelength
under construction so that we can approximate it by a point current distribution,
J(r) = z^Il(r):
(1)
The dipole may look like the following
q
+

metallic spheres charge reservoir


I
generator

l is the e ective length of the dipole so that the dipole moment p =


ql. The charge qdqis varying time harmonically because it is driven by the
generator. Since dt = I , we have
Il = dq
dt l = j!ql = j!p

(2)

for a Hertzian dipole. We already know that the corresponding vector potential is given by
;jr
A(r) = z^4Il
(3)
r e :
The magnetic eld is obtained, using cylindrical coordinates, as

@ A ; ^ @ A 
H = 1 r  A = 1 ^1 @
z
@ z
p

(4)

@ =  @.
@r @ = p 
where @@ = 0 r = 2 + z2 . In the above, @@ = @
@r
r @r
2 +z 2 @r
Hence,
 1


Il
1
^
H = ; r 4 ; r2 ; j
r e;jr :
(5)

r
y

In spherical coordinates, r = sin , and (5) becomes

Il (1 + j
r)e;jr sin :
H = ^ 4r
2

The electric eld can be derived using Maxwell's equations.


 1 @

1
1
1
@
^
E = j! r  H = j! r^r sin @ sin H ; r @r rH
Ile;jr hr^2 cos (1 + j
r) + ^ sin (1 + j
r ;
2 r2 )i :
= j!
4r3

Case I. Near Field,


r  1
 (^r2 cos + ^ sin )
r  1
E = 4 r
3
H  E when
r  1:

(6)

(7)

(8)

(9)

r could be made very small by making r small or by making ! ! 0. The


above is like the static eld of a dipole.

Case II. Far Field (Radiation Field),


r  1
In this case,
and

E = ^j! 4Ilr e;jr sin 

(10)

(11)
H = ^j
4Ilr e;jr sin :
p
=  = 0 . E and H are orthogonal to each other

Note that HE = !


and are both orthogonal to the direction of propagation, i.e. as in the case
of a plane wave. A spherical wave resembles a plane wave in the far eld
approximation.
2

The time average power ow is given by



Il 2
1
1

0
2

hSi = 2 <eE  H ] = r^2 0 jHj = r^ 2 4r sin2 :
(12)
The radiation eld pattern of a Hertzian dipole is the plot of jEj as a
function of at a constant r.
z

|E|

x, y

The radiation power pattern is the plot of hSr i at a constant r.


z

x, y

The total power radiated by a Hertzian dipole is given by

P=

2
0

Since

Z
0

d sin = ;
3

d r sin hSr i = 2

;1
1

(d cos )1 ; cos ] =


2

then

 2

0
3
d 2
Il
4 sin :

;1

dx(1 ; x2) = 34 

Il
4
P = 3  0 4 :
The directive gain of an antenna, D(  ), is dened as

D(  ) = hSPr i 
4 r2

(13)
(14)
(15)
(16)

where 4 rP 2 is the power density if the power P were uniformly distributed


over a sphere. Substituting (12) and (15) into the above, we have

D(  ) =

; Il 2

2 4 r
1 4
4 r2 3 0

sin2

; Il 2 = 2 sin2 :

(17)

The peak of D(  ) is known as the directivity of an antenna. It is 1:5 in this


case. If an antenna is radiating isotropically, its directivity is 1. Therefore,
the lowest possible values for the directivity of an antenna is 1, whereas it
can be over 100 for some antennas like reector antennas. A directive gain
pattern is a plot of the above function D(  ) and it resembles the radiation
power pattern.
If the total power fed into the antenna instead of the total radiated power
is used in the denominator of (16), the ratio is known as the power gain or
just bf gain. The total power fed into the antenna is not equal to the total
radiated power because there could be some loss in the antenna system like
metallic loss.
Dening a radiation resistance Rr by P = 21 I 2Rr , we have

 

l
2
P
(18)
Rr = I 2 = 0 6  where 0 = 377:
For example, for a Hertzian dipole with l = 0:1, Rr  8. For a small dipole
with no charge reservoir at the two ends, the currents have to vanish at the
tip of the dipole.

I(z)

a/2

a/2

The e ective length of the dipole is half of its actual length due to the
manner the currents are distributed. For example, for a half-wave dipole,
a = 2 , and if we use le = 4 in (18), we have

Rr  50:

(19)

However, a half-wave dipole is not much smaller than a wavelength and does
not qualify to be a Hertzian dipole. Furthermore, the current distribution
on the half-wave dipole is not triangular in shape as above. A more precise
calculation shows that Rr = 73 for a half-wave dipole.
4

W.C.Chew
ECE 350 Lecture Notes

27. Radiation Field Approximations


The vector potential due to a source J(r), can be calculated from the
equation
ZZZ
A(r) =
dr 4jJr(;r )r j e j 
(1)
0

jr;r j

where V is the volume occupied by J(r).


J(r)

|r r|
r

r
y

0
r r
x

When jrj  jr j, then jr ; r j = r ; r  r^. Equation (1) becomes


ZZZ

A(r) =
dr r;J(rr )r^e jr ej r^
V
jr Z Z Z
e
= 4r
dr J(r )ej r^
V
(2)
= e jr f (r ) = ^A + ^A + r^Ar :
In the above we have assumed that jr  r^j  r but  r  r^ is not small, since 
)
can be large. When r is large, f (
is a slowly varying function compared
r
(

)
to e jr . Hence, we can regard r almost to be a constant compared to
e jr . The magnetic eld can be derived to be


1
1
@
@
^
^
H =  r  A  ;   @r A ;  @r A :
(3)
However, @r@  ;j when r is large. Hence,
H = j (^A ; ^A ) when r ! 1:
(4)
0

r0 

r0 

Similarly,

1 r  H  ;j!^A + ^A ]:
E = j!

=



(5)

Linear Array of Dipole Antennas


If J(r ) is of the form
J(r ) = z^IlA0 (x ) + A1 (x ; d1 ) + A2 (x ; d2)
+    + AN 1 (x ; dN 1)] (y ) (z )
0

(6)

0
y
r
r r

r r

d0 r d1

d2

d3

d4

d N3

dN2 d N1

the vector potential on the xy-plane can be derived to be


ZZZ
Il
jr
A(r) = z^4r e
dr A0 (x ) + A1 (x ; d1) +    ] (y ) (z )e+j
jr 
e
A0 + A1e+jd1 cos  + A2 ejd2 cos  +    + AN 1ejdN
= z^4Il
r
0

If dn = nd, and An = ejn , then (7) becomes


jr 
j (d cos +)
2j (d cos +)
j (N
e
1
+
e
+
e
+



+
e
A(r) = z^4Il
r
which is of the form
N
N
X1
xn = 11;;xx :
;

Therefore,

n=0

r0  ^


(8)

1)(d cos +)

:
(7)

1 cos 

(9)

1 ; ejN (d cos +) :


(10)
1 ; ej(d cos +)
The electric eld on the xy-plane is E = ;j!A = +j!Az . Hence, jE j is of
the form


 1 ; ejN (d cos +) 

jE j = jE0j  1 ; ej(d cos +) 


 sin N (d cos  + ) 


= jE0j  21
:
(11)
sin 2 (d cos  + ) 

A(r) = z^ 4Il
r e

jr

Equation (11) is of the form


an example.

sin Nxj
jsin xj

. Plots of jsin 3xj and jsin xj are shown as


|sin x|
|sin 3x|

sin 3x
sin x

In equation (11), = 12 (d cos  + ). We notice that the maximum in


(11) would occur if = n, or if

n = 0
1
2
3    :

d cos  + = 2n

(12)

The zeros or nulls will occur at Nx = n, or

d cos  + = 2Nn 

n =
1
2
3     n 6= mN:

(13)

For example,

Case I. = 0;d= , principal maximum is at  =


2 if N = 5, nulls
are at  =
cos 1 25n , or  =
66:4 
36:9 
113:6 
143:1 .


y
113.6
143.1

66.4
36.9

x broadside
array
143.1
113.6

36.9
66.4

Case II. =  d;=n , principal


maximum is at  = 0 , if N = 4,

1
nulls are at  =
cos 2 ; 1 , or  =
120 
90 
60 .
;

y
120

90

60

120

90

60

The interference eects between the dierent antenna elements of a linear


array focus the power in a given direction. We can use linear array to increase
the directivity of antennas.
Note that equation (7) can also be derived by other means. We know
that the vector potential due to one dipole is

e j 
A = z^Il
(14)
4 jr ; r j
when the dipole is located at r and pointing in the z^-direction. Hence
for an array of dipoles of dierent phases and amplitudes, located at x =
x^d0 x^d1 x^d2     x^dN 1, the vector potential by linear superposition is
 j x^d0

j x^d1
j x^dN 1
e
e
e
Il
A(r) = z^ 4 jr ; x^d j A0 + jr ; x^d j A1 +    + jr ; x^d j AN 1 :
0
1
N 1
(15)
If we approximate jr ; x^dnj by r ; r^  x^dN = r ; dN cos , in the phase, and
by r in the denominator, then (15) becomes
jr 
A(r) = z^4Il
e
A0 + A1e+jd1 cos  + A2 ejd2 cos 
r

+    + AN 1ejdN 1 cos   (16)
which is the same as equation (7). The interference between the terms in
(16) can be used to generate dierent radiation patterns for dierent communication applications.
0

jr;r j

jr;

jr;

jr;

Let c = a + jb, and h = f + jg, then


and

c + h = (a + f ) + j (b + g)

(4)

c ; h = (a ; f ) + j (b ; g)

(5)

Multiplication and Division


ch = (a + jb)(f + jg) = (af ; bg) + j (bf + ag)
c = a + jb = (a + jb)(f ; jg) = af + bg + j bf ; ag :
h f + jg (f + jg)(f ; jg) f 2 + g2 f 2 + g2

(6)
(7)

Multiplication and division are more conveniently carried out in a polar form.
Let
c = jcj ej1  h = jhj ej2 
(8)
then
ch = jcj jhj ej(1 +2) 
(9)

c = jcj ej( ; ):
h jhj
1

(10)

Square Root of a Complex Number


It is most convenient to take the square root of a complex number in

polar form or by converting it to polar form.


p
c = jcj ej1 = a2 + b2ej tan;1 ab 
pc = jcj 21 ej 21 = (a2 + b2) 14 ej 21 tan;1 ab :
In fact

1

c m = jcj m ej m = (a2 + b2) m ej m tan; ab :


1

1
2

Phasor Representation of a Time-Harmonic Scalar


2

(11)
(12)
(13)

If V (t) is a time-harmonic signal such that


V (t) = V0 cos(!t + )
it could also be written as
V (t) = <efV0 ejej!tg:
The term V~ = V0ej is known as the phasor representation of V (t).
If U (t) = U0 cos(!t + 1 ), or the phasor representation of U (t) is
U~ = U0ej1 :
It can be shown easily that
V (t) + U (t) = <efV| 0{zej} + U| 0{zej}1 ]ej!t g:
V~

(14)
(15)

(16)
(17)

U~

Hence V~ + U~ is a phasor representation of V (t) + U (t).


Also
@V (t) = @ <efV ejej!t g = <efj! V ej ej!tg:
0
| 0{z }
@t
@t

(18)

V~

Therefore j!V~ is a phasor representation of @t@ V (t). However, as a word of


caution, V~ U~ is not a phasor representation of V (t)U (t). You can convince
yourself of this.

Exercise
1) Show that,
(a) c + c is always real,
(b) c ; c is always imaginary,
(c) c=c has magnitude equal to 1.
2) Consider z2 = 1 + 2j . It is a second order polynomial with two roots.
Find the two roots.
3) Obtain the phasor representation of the following
(a) V (t) = 10 cos(!t + 3 ),
(b) I (t) = ;8 sin(!t + 3 ),
(c) A(t) = 3 sin !t ; 2 cos !t,
(d) C (t) = 3 cos(!t + 4 ) + 4 sin(!t + 3 ):
4) Obtain C (t) in terms of ! from the following phasors:
(a) c = 1 + j ,
3

(b) c = 4 exp(
j 0:8),

j
(c) c = 3e 2 + 4ej0:8,
(d) c = j sin 3z.
5) (a) Using binomial theorem, show that

1 + ja '  1 + j a2 


if jaj  1:

(b) Show that


1 + ja ' (1 + j ) a2

 1

if jaj  1: