You are on page 1of 191

CONTENTS

1. INTRODUCTION .............................................................................................. 6
1.1 Description of an electromagnetic field ………………….….............................. 8
1.2 Wave equation ………………...........................................................…............... 13
1.3 Potentials ………….........................................................................…................. 14
2. ELECTROMAGNETIC WAVES IN FREE SPACE .................................… 17
2.1 Solution of the wave equation ....................................…..................................... 17
2.2 Propagation of a plane electromagnetic wave …………………………………. 18
2.3 Wave polarization ……………………………………………………………… 28
2.4 Cylindrical and spherical waves ……………………………………………….. 32
2.5 Problems ……………………………………………………………………….. 34
3. WAVES ON A PLANE BOUNDARY ………………………………………. 35
3.1 Perpendicular incidence of a plane wave to a plane boundary ………………… 35
3.2 Perpendicular incidence of a plane wave to a layered medium ………………... 42
3.3 Oblique incidence of a plane electromagnetic wave to a plane boundary ……... 45
3.4 Problems ……………………………………………………………………….. 54
4. SOLUTION OF MAXWELL EQUATIONS AT VERY HIGH
FREQUENCIES ………………………………………………………………. 56
5. GUIDED WAVES …………………………………………………………….. 61
6. TEM WAVES ON A TRANSMISSION LINE ………………………………. 65
6.1 Parameters of a TEM wave …………………………………………………….. 65
6.2 Transformation of the impedance along the line ………………………………. 71
6.2.1 An infinitely long line ………………………………………………………….. 71
6.2.2 A line of finite length …………………………………………………………… 72
6.2.3 A line terminated by a short cut and by an open end ……….………………….. 73
6.3 Smith chart ……………………………………………………………………… 76
6.4 Problems ……………………………………………...………………………… 87
7. WAVEGUIDES WITH METALLIC WALLS ……………………………... 88
7.1 Parallel plate waveguide ……………………………………………………….. 88
7.2 Waveguide with a rectangular cross-section …………………………………… 94
7.3 Waveguide with a circular cross-section ………………………………………. 103
7.4 Problems ………………………………………………………………………... 107
8. DIELECTRIC WAVEGUIDES ……………………………………………… 108
8.1 Dielectric layers ………………………………………………………………… 109
8.2 Dielectric cylinders ……………………………………………………………... 114
8.3 Problems ………………………………………………………………………... 115
9. RESONATORS ……………………………………………………………….. 116
9.1 Cavity resonators ……………………………………………………………….. 116
9.2 Problems ………………………………………………………………………... 120
10. RADIATION ………………………………………………………………….. 121
10.1 Elementary electric dipoles …………………………………………………….. 121
10.2 Elementary magnetic dipoles …………………………………………………... 126
10.3 Radiation of sources with dimensions comparable with the wavelength ………. 129
10.4 Antenna parameters …………………………………………………………….. 133
10.5 Antenna arrays ………………………………………………………………….. 134
10.6 Receiving antennas ……………………………………………………………... 137
4
book - 1
10.7 Problems ……………………………………………………………………….. 140
11. WAVE PROPAGATION IN NON-ISOTROPIC MEDIA ………………… 141
11.1 Tensor of permeability of a magnetized ferrite ………………………………… 142
11.2 Longitudinal propagation of a plane electromagnetic wave in a magnetized
ferrite …………………………………………………………………………… 146
11.3 Transversal propagation of a plane electromagnetic wave in a magnetized
ferrite …………………………………………………………………………… 151
11.4 Applications of non-reciprocal devices ………………………………………… 154
11.5 Problems ………………………………………………………………………... 155
12. APPLICATIONS OF ELECTROMAGNETIC FIELDS ………………….. 156
12.1 Introduction to microwave technology …………………………………………. 156
12.2 Antennas ………………………………………………………………………... 162
12.2.1 Wire antennas …………………………………………………………………... 164
12.2.2 Aperture antennas ………………………………………………………………. 165
12.2.3 Broadband antennas …………………………………………………………….. 167
12.2.4 Planar antennas …………………………………………………………………. 168
12.3 Propagation of electromagnetic waves in the atmosphere ……………………… 169
12.4 Optoelectronic ………………………………………………………………….. 173
12.4.1 Optical waveguides …………………………………………………………….. 173
12.4.2 Optical detectors ………………………………………………………………. 176
12.4.3 Optical amplifiers and sources …………………………………………………. 177
12.4.4 Optical modulators and sensors ………………………………………………… 178
13. MATHEMATICAL APPENDIX …………………………………………….. 180
14. BASIC PROBLEMS ………………………………………………………….. 191
15 LIST OF RECOMMENDED LITERATURE ………………………………. 194

5
book - 1
1. INTRODUCTION
This textbook is aimed at students of the Faculty of Electrical Engineering, Czech
Technical University taking a course in Waves and Transmission Lines. The textbook builds
on basic knowledge of time varying electromagnetic fields gained from courses in physics
and electromagnetic field theory. The textbook introduces all the basic knowledge that an
electrical engineer specializing in radio engineering and telecommunications should have and
that is necessary for further courses such as microwave engineering, antennas and
propagation, optical communications, etc. Sequential mastering of wave theory contributes to
the final objective of university studies, which is to enable graduates to do creative work.
The course in Waves and Transmission Lines studies the theory and applications of
classical electrodynamics. It is based on Maxwell’s equations. This course provides a basis
for understanding the behavior of all high frequency electric circuits and transmission lines,
starting from those applied to transmitting and receiving electric energy, processing signals,
microwave circuits, optical fibers, and antennas. The main applications lie in wireless
communications, radio engineering and optical systems.
The text follows the classical approach to macroscopic electrodynamics. All quantities
are assumed to be averaged over the material, which by its nature has a microscopic structure
consisting of atoms. This confines the description of electromagnetic effects using
macroscopic theory on the high frequency side, as the wavelength must be much longer than
the dimensions of the atoms and molecules. This boundary lies in the range of ultraviolet
light. Nevertheless, the spectrum of frequencies in which electromagnetic effects can be
treated using this macroscopic theory is really huge – over 17 orders. And this whole
spectrum really is used in a variety of different applications. Modern communication systems
use electromagnetic waves with ever shorter wavelengths. The spectrum of electromagnetic
waves is shown in Fig. 1.1.
First, we review the basic relations from electromagnetic field theory and introduce
potentials describing a time varying electromagnetic field. The concept of a plane
electromagnetic wave is carefully reviewed. In addition, a cylindrical wave together with a
spherical wave are briefly introduced. The behaviour of a plane electromagnetic wave on the
boundary between two different materials is studied in detail. Here we will start treating the
incidence of a wave perpendicular to the plane boundary and to a layered medium. Oblique
incidence is studied in general, and then special effects such as total transmission and total
reflection are treated. Specific aspects of solving Maxwell equations at very high frequencies
are discussed separately.
Transmission lines are designed to transmit guided waves. After introducing the
general properties of guided waves the TEM wave is treated, and the transformation of
impedances along a line is described. The Smith chart, a very effective graphical tool for
analysis and design of high frequency circuits, is described and its basic applications are
explained through particular problems. Waves propagating along waveguides with metallic
walls of rectangular and circular cross-sections are studied. Dielectric waveguides are treated
separately. They form the basis of optical fibers. Cavity resonators, unlike low frequency L-C
resonant circuits, are able to resonate on an infinite row of resonant frequencies. Several kinds
of such resonators are analyzed in the text.
Attention is paid to problems of radiation of electromagnetic waves. This covers
antenna theory. Elementary sources of an electromagnetic field, such as an electric dipole and
a magnetic dipole, are studied and compared. Then radiation from sources with dimensions
comparable with the wavelength is described. The basic antenna parameters are defined. The
basic idea of antenna arrays is built up. Finally, receiving antennas are dealt with, and the
effective antenna length and effective antenna surface are derived.

6
book - 1
Non-isotropic materials are introduced, and the tensor of permeability of a magnetized
ferrite is calculated. The propagation of a plane electromagnetic wave in this ferrite material
homogeneously filling an unbounded space is studied, in particular when the wave propagates
both in the direction parallel with the magnetizing field and in the direction perpendicular to
the magnetizing field. Some devices using non-isotropic materials are mentioned.
frequency wavelength classification applications
(Hz) (m)
1018
10-9 ultraviolet radiation
nm
360
violet
460
blue argon laser 490 nm
1015 green 560
visible light yellow
10-6 He-Ne laser 630 nm
red 660
infrared radiation 760

1012 quasi optical waves


-3
10
m
millimeter waves i radar, space investigation
10 -2 c
r
centimeter waves o
w radar, satellite commun.
a
v
109 decimeter waves e
s radar, TV, navigation
0
10
very short waves TV, FM radio, services
10
short waves radio, services
102
106
medium waves AM radio
103
long waves
104

103
phone, audio
106
The spectrum of electromagnetic waves.
Fig. 1.1
The applications of electromagnetic fields in particular branches of electrical
engineering are briefly introduced. An introduction to microwave technology is given. Here
scattering parameters are introduced. The paragraph on antennas represents a continuation of
Chapter 10, introducing basic types of antennas. Particular mechanisms of wave propagation
in the atmosphere are explained. The transmission formula and radar equations are derived.
The basics of optoelectronics are presented. This involves a characterization of optical
waveguides, detection and optical detectors, optical amplifiers, and the sources of optical
7
book - 1
radiation, namely lasers, and finally modulators of an optical beam and a short review of
optical fiber sensors.
The book has a mathematical appendix summarizing the necessary knowledge of
mathematics. Basic problems in the form of questions are summarized at the end of the text.
They help the students in preparing for their examinations. A list of suggested literature is
given.
The textbook treats time varying electromagnetic fields. Harmonic dependence on
time is assumed throughout most of the text. Such fields are described using symbolic
complex quantities called phasors, and time dependence in the form e jωt is assumed. These
phasors are not marked by special symbols. When we need to emphasize an instantaneous
value, we will mark it by showing dependence on time, e.g., E(t). Vectors will be marked by
bold characters, e.g., E.

1.1 Description of an electromagnetic field


The sources of an electromagnetic field are electric charge Q [C] and electric current I
[A], which is nothing else than the flow of charge. Charge is often distributed continuously in
a space, on a surface, or along a curve. It is convenient in this case to define the corresponding
charge densities. The charge volume density is defined as the charge amount stored in a unit
volume

∆q
ρ = lim [C/m3] . (1.1)
∆V →0 ∆V

The charge surface density is defined by analogy

∆q
σ = lim [C/m2] . (1.2)
∆S →0 ∆S

The charge linear density is defined as the charge stored along a line or a curve of unit
length

∆q
τ = lim [C/m] , (1.3)
∆l →0 ∆l

Electric current is created by a moving charge. This is defined as the passing charge
per time interval ∆t

∆q
I = lim [A=C/s] . (1.4)
∆t → 0 ∆t

It is useful to define current densities, which are vector quantities, as it is necessary to define
the current flow direction. Current density is defined as

∆I
J = lim i 0 [A/m2] , (1.5)
∆S → 0 ∆S

where i0 is a unit vector describing the current flow direction. It is sometimes useful to use the
abstraction of a surface current passing along a surface, see Fig. 1.2. The linear density of

8
book - 1
this current is defined as

∆I
K = lim i 0 [A/m] . (1.6)
∆x → 0 ∆x

The surface current and its density are an


abstraction used to simplify the mathematical I
description of the current passing a conductor at a i0
K
high frequency. Due to the skin effect, current S
flows through only a very thin layer under the ∆x
material surface. In fact the surface current
defined by (1.6) represents the finite current
passing a cross-section of zero value. This Fig. 1.2
requires infinite material conductivity. Another
widely used abstraction is a current filament representing a conductor of negligible cross-
section (e.g., a line) carrying a finite current.
The total electric current crossing a closed surface is related to the charge accumulated
inside the volume surrounded by this surface by the continuity equation in an integral form

∫∫ J ⋅ dS + jωQ = 0 , (1.7)
S

or in a differential form

div J + jω ρ = 0 . (1.8)

where Q is the total charge accumulated in volume V with boundary S. The current density is
related by Ohm’s law in the differential form to an electric field

J=σE , (1. 9)

where σ is conductivity in S/m. It should be noted that in spite of the movement of the free
electrons, a conductor passed by an electric current stays electrically neutral, as the charge of
the electrons is compensated by the positive charge of the charged atomic lattice.
The electric field is described by the vector of electric field intensity E, the unit of
which is V/m. The magnetic field is described by the vector of magnetic field intensity H, the
unit of which is A/m. These vectors are related to the induction vectors by material relations

B=µH , (1.10)

D=εE , (1.11)

where ε = ε 0ε r and µ = µ 0 µ r are permittivity and permeability, respectively, and


ε 0 = 10−9 / 36π F/m and µ 0 = 4π 10−7 H/m are the permittivity and permeability of a vacuum,
respectively. Vectors E and H are related by the set of Maxwell’s equations. Their
differential form reads

rot H = (σ + jωε )E + J S , (1.12)

9
book - 1
rot E = − jωµ H , (1.13)

div(ε E ) = ρ0 , (1.14)

div(µ H ) = 0 , (1.15)

where JS is a current supplied by an external independent source, ρ0 is the volume density of a


free charge supplied by an external independent source. The differential form of Maxwell’s
equations is valid only at those points where the field quantities
are continuous and are continuously differentiated functions of n
position. They are not valid, for example, on a boundary between 1 ε1 µ1 σ1
two different materials where the material parameters change
step-wise, Fig. 1.3. For this reason we have to append ε2 µ2 σ2
2
corresponding boundary conditions to these equations. We
suppose that the boundary between two materials contains a free Fig. 1.3
electric charge with density σ0, and electric current K passing
along this boundary. The boundary conditions in vector form can be expressed

n ⋅ (ε1E1 − ε 2E 2 ) = σ 0 ,

n × (E1 − E 2 ) = 0 ,

n ⋅ (µ1H1 − µ 2 H 2 ) = 0 , (1.16)

n × (H1 − H 2 ) = K ,

n ⋅ (J1 − J 2 ) = 0 ,

n is the unit vector normal to the boundary. A scalar form using the normal and tangential
components of vectors is

ε1En1 − ε 2 En 2 = σ , (1.17)

Et1 = Et 2 , (1.18)

µ1H n1 = µ 2 H n 2 , (1.19)

H t1 − H t 2 = K , (1.20)

J n1 = J n 2 . (1.21)

Specially on the surface of an ideal conductor with conductivity σ → ∞ , and since the
electric and magnetic fields are zero inside this conductor, we have

ε1En1 = σ 0 , Et1 = 0 , H n1 = 0 , H t1 = K . (1.22)

The first Maxwell equation (1.12) has three terms on its right hand side. Term jωE

10
book - 1
represents the displacement current, term σE represents the conducting current which causes
conducting losses in the material, see (1.9), and JS is the current supplied by an internal
source. Equation (1.12) can be simplified by introducing a complex permittivity

 σ 
rot H = jωε 0ε r E + σE + J S = jωε 0  ε r − j  E + J S = jωε 0ε r (1 − j tg δ e ) E + J S =
 ωε 0 
jωε 0ε c E + J S

where complex permittivity εc is defined

σ
εc = εr − j = ε r (1 − j tg δ e ) = ε '− jε ' ' , (1.23)
ωε 0

and term

σ
tg δ e = , (1.24)
ωε 0ε r

is called the loss factor, or thea loss tangent, and δe is the loss angle. This loss factor is
frequently used to define the losses in a material in spite of the fact that it is frequency
dependent. The reason is that this loss factor, similarly as the imaginary part of permittivity
ε’’, contains in practice not only the conducting losses, but also polarization and another kinds
of losses. Similarly we can introduce complex permeability µc representing all kinds of
magnetic losses

µ c = µ r (1 − j tg δ m ) = µ '− jµ ' ' . (1.25)

In material relations (1.9) – (1.11) conductivity σ , permittivity ε and permeability µ


represent the electric and magnetic properties of a material. In a linear material these
parameters do not depend on field quantities, while in a nonlinear material they depend on E
or H. These parameters can depend on space coordinates, which is the case of a non-
homogeneous material. In a homogeneous material σ, ε, and µ do not depend on
coordinates. An isotropic material has parameters that are constant in all directions, ε, µ, σ
are scalar quantities. Non-isotropic materials possess different behaviour in different
directions. Their permittivity, permeability, or conductivity are tensor quantities that can be
expressed by matrices. An example is provided by magnetized ferrite or magnetized plasma.
E.g., equation (1.11) for magnetized plasma can be rewritten into

D=ε E , (1.26)

which gives three particular scalar equations

Dx = ε xx E x + ε xy E y + ε xz E z ,

D y = ε yx E x + ε yy E y + ε yz E z , (1.27)

Dz = ε zx E x + ε zy E y + ε zz E z .

11
book - 1
Vector D has a different direction from vector E. Permittivity is then not a scalar quantity, but
a tensor quantity

ε xx ε xy ε xz 
 
ε = ε yx ε yy ε yz  . (1.28)
ε zx ε zy ε zz 

Similarly we can express the permeability and conductivity of an anisotropic material.


The density of power transmitted by an electromagnetic wave is described by the
complex Poynting vector. This vector is defined as

1
S= E × H * = S av + jQ , (1.29)
2

where Sav is the average value of the transmitted power, which represents the density of active
power

S av =
1
2
[
Re E × H * . ] (1.30)

Q is the density of reactive power

Q=
1
2
[
Im E × H * . ] (1.31)

Poynting’s theorem represents the balance of power in an electromagnetic system in volume


V. It can be read by dividing power into active and reactive power

PS = PJ + PR , (1.32)

(
QS = 2 ω Wmav − We av + QR , ) (1.33)

where PS and QS are the average values of the active and reactive power supplied by an
external source. The active power supplied by an external source is

PS =
1
2 ∫∫∫
{
Re E ⋅ J *S dV , } (1.34)
V

JS is the current supplied by this source. The active power is partly lost in materials, power PJ,
and partly radiated outside of our volume, power PR. These quantities are

1 2
PJ = ∫∫∫
2 V
σ E dV , (1.35)

{(
PR = ∫∫ S av ⋅ dS = ∫∫ Re E × H * ⋅ dS , )} (1.36)
S S

12
book - 1
where S is the surface surrounding investigated volume V. Similarly, for reactive power we
get

QS = −
1
∫∫∫
2 V
{
Im E ⋅ J *S dV .} (1.37)

This power covers the difference between the average values of the energy of an electric field,
Weav, and the energy of a magnetic field, Wmav, and is partly radiated outside, QR. These values
are

QR =
1
2 ∫∫
{ }
Im E × H * ⋅ dS , (1.38)
S

µ H 2  ε E 2 
Wmav = ∫∫∫   dV , Weav = ∫∫∫   dV . (1.39)
 4   4 
V   V  

1.2 Wave equation


The solution of an electromagnetic field by Maxwell’s equations (1.12) and (1.13)
involves solving six scalar equations for three components of the electric field and for three
components of the magnetic field. We can reduce this number of equations by extracting one
of the two vectors from (1.12) and (1.13), as vectors E and H are mutually dependent. Let us
start with equation (1.12). Let us apply operator rot to this equation

rot rot H = (σ + jωε ) rot E + rot J S .

Inserting for doubly applied operator rotation rotrotH = grad div H − ∆H (13.81) we get

grad div H − ∆H = (σ + jωε ) rot E + rot J S .

rotE can be expressed from Maxwell’s second equation (1.13) and Maxwell’s fourth equation
(1.15) states that divH=0. Now we have

− ∆H = − j ω µ (σ + jωε )H + rot J S .

The term containing material parameters and a frequency in front of H on the right hand side
is usually denoted

k 2 = − j ω µ (σ + jωε ) = ω 2 µε − jωµσ , (1.40)

where k is a complex propagation constant. Using this simplification we get the wave
equation for the vector of magnetic field intensity in the form

∆H + k 2 H = −rot J S . (1.41)

13
book - 1
The derivation of the equation for E is similar. Let us apply operator rotation to (1.13),
and we get

rot rot E = − jωµ rot H .

Using (13.81), expressing divE from Maxwell’s third equation (1.14), and using the
propagation constant k (1.40) we get

ρ0
∆E + k 2 E = grad + jωµ J S . (1.42)
ε

Wave equations (1.41) and (1.42) are basic equations which describe an electromagnetic field
in a space with general sources supplying current JS and charge ρ0. These equations are,
however, not suitable for solving the field in areas with sources, as their right hand sides are
rather complicated functions of source quantities JS and ρ0. Equations for potentials are
preferably applied to solve these problems, which, e.g., involve radiation of antennas.

1.3 Potentials
Stationary or static fields are described by simplified Maxwell’s equations

rot E = 0 , (1.43)

div B = 0 . (1.44)

The potentials for these fields can then be simply defined. The scalar potential is defined by

E = − grad ϕ , (1.45)

and the vector potential by

B = rot A . (1.46)

These potentials are solutions of Poisson’s equations

ρ0
∆ϕ = − , (1.47)
ε

∆A = − µJ S . (1.48)

Equation (1.48) holds, supposing that the Lorentz calibration condition in the form

div A = 0 (1.49)

is laid to the vector potential. The solution of these equations can be obtained by applying the
method of superposition

14
book - 1
1 ρ0
4πε ∫∫∫
ϕ= dV , (1.50)
r

µ JS
A=
4π ∫∫∫ r
dV . (1.51)

In a non-stationary field we have again equation (1.44) and the vector potential is,
therefore, defined by (1.46). An electric field is a different story. It is now described instead
of (1.43) by Maxwell’s equation (1.13), which can be rewritten into

rot E = − jωB = − jω rot A ,

rot (E + jωA ) = 0 .

This last equation tells us that the rotation of the vector in brackets equals zero, consequently
this vector can be expressed by a scalar potential. The electric field is now expressed by both
the scalar potential and the vector potential

E = −grad ϕ − jω A . (1.52)

Let us now derive equations analogous to (1.47) and (1.48) which describe the
distribution of the vector and scalar potentials of a time varying electromagnetic field. First
we insert (1.46) and (1.52) into Maxwell’s first equation (1.12) and we get

rot rot A = ( jωµε + µσ )(− grad ϕ − jωA ) + µ J S ,

( )
grad divA + ∆A = −( jωµε + µσ ) grad ϕ + A ω 2 µε − jωµσ + µ J S ,

∆A + k 2 A = grad[divA + ( jωµε + µσ )ϕ ] − µ J S .

This equation can be simplified supposing that the argument of the gradient is zero, which is
the Lorentz calibration condition for a time varying electromagnetic field

div A = −( jωµε + µσ )ϕ . (1.53)

Finally the wave equation for the vector potential reads

∆A + k 2 A = − µJ S . (1.54)

The equation for the scalar potential is derived in a similar way. We insert (1.52) into
Maxwell’s third equation (1.14)

ρ0
− div grad ϕ − jω div A = .
ε

div A is expressed by calibration condition (1.53), and we have

15
book - 1
ρ0
∆ϕ − jω ( jωµε + µσ )ϕ = − ,
ε

which gives the final form of the wave equation for the scalar potential

ρ0
∆ϕ + k 2ϕ = − . (1.55)
ε

Equations (1.54) and (1.55) are more suitable for solving an electromagnetic field, as the
terms on their right hand sides are simple functions of source quantities JS and ρ0. They are
frequently used to analyze and design antennas. Note that Poisson’s equations (1.47) and
(1.48) represent limiting cases of general wave equations (1.54) and (1.55), assuming that the
frequency is decreased to zero, which is equivalent to decreasing propagation constant k. Due
to this, the solution of wave equations can be expected in a form corresponding to (1.50) and
(1.51). Let us first solve equation (1.55) for the scalar potential in the time domain. A lossless
medium is assumed to simplify understanding. Now the equivalent to (1.55) in the time
domain is

∂ 2ϕ ρ0
∆ϕ + µε =− .
∂t 2 ε

The variations of an electromagnetic field excited at distance R from a source are delayed by
the time which is necessary for the wave to travel at speed v along this distance R = r − r , ,
Fig. 1.4

R
∆t = . (1.56)
v

Consequently we get ϕ
R
 R
ρ0 t −  dV
1 ρ 0 (t − ∆t ) 1  v
ϕ (t ) = ∫∫∫ dV = ∫∫∫ dV ρ0
4πε V R 4πε V R

r
Applying phasors we get, assuming that k=ω/v r’

1 ρ 0 e jω (t − R / v )
ϕ e jωt = ∫∫∫ dV .
4πε V
R
0
The final form of the solution is Fig. 1.4

1 ρ 0 e − jkR
ϕ=
4πε ∫∫∫ R
dV . (1.57)
V

Similarly for the vector potential we have

16
book - 1
µ J S e − jkR
A=
4π ∫∫∫ R
dV . (1.58)
V

Again we see that (1.57) and (1.58) pass to (1.50) and (1.51) when the frequency is reduced to
zero, and, consequently, the propagation constant goes to zero, which causes a transition
e − jkr → 1 .
Equations (1.57) and (1.58) are frequently used to calculate the electromagnetic field
excited by sources, e.g., by antennas in space or probes in waveguides. The electric field and
the magnetic field are then calculated from the potentials, using (1.46) and (1.52). Prior to
performing this calculation we usually have to determine the distribution of the electric
charge and the electric current. This task is often not simple, see section 10.

2. ELECTROMAGNETIC WAVES IN FREE SPACE

2.1 Solution of the wave equation


Let us solve wave equation (1.42) for an electric field. We will look for a solution
which describes a plane electromagnetic wave propagating in an infinite space filled by a
homogeneous material. Wave fronts, which are the surfaces of a constant field phase, are - in
the case of a plane wave - planes perpendicular to the direction of propagation. Such a wave
can be excited by a source located at infinity, which excites a field with a constant amplitude
and phase within the whole infinite plane. So this wave is only an abstraction. We will study
its behaviour and propagation, as this provides a basis for understanding all phenomena
connected with the propagation of electromagnetic waves. Most real waves can be expressed
as the superposition of a series of plane electromagnetic waves.
Let us assume a free space filled by a homogeneous material with parameters ε, µ, and
σ. The wave equation is solved as a homogeneous equation in the form (1.42), i. e., without
any source. This means that the solution describes the possible particular waves, called modes
or eigen waves, which can propagate in our space. We rewrite equation (1.42) into a form
describing the i-th coordinate of the electric field

∂ 2 E i ∂ 2 Ei ∂ 2 E i
+ + + k 2 Ei = 0 , i=x, y, z (2.1)
∂x 2 ∂y 2 ∂z 2

This equation can be solved by the method of separation of variables. Ei is expected in the
form

Ei = X ( x )Y ( y ) Z ( z ) . (2.2)

Introducing (2.2) into (2.1) we get

X '' Y '' Z ''


+ + + k2 = 0 , (2.3)
X Y Z

where

17
book - 1
X '' Y '' Z ''
= − k x2 , = − k y2 , = − k z2 (2.4)
X Y Z

and the propagation constant (1.40) can be written in the form

k 2 = ω 2 µε − jωµσ = k x2 + k y2 + k z2 . (2.5)

Equation (2.3) is then separated into three equations. The equation for the x-coordinate is

X '' + k x2 X = 0 , (2.6)

with its solution in the form of the two propagating plane waves

X = A e jk x x + B e -jk x x . (2.7)

Combining electric field Ei (2.2) we get

( )(
Ei = A e jk x x + B e -jk x x C e
jk y y
+ De
- jk y y
)(E e jk z z
)
+ F e -jk z z . (2.8)

Constants A to F can be determined from the boundary conditions. Multiplying the brackets in
j (± k x ± k y ± k z )
(2.8) we get eight terms of the form e x y z . We confine the solution of the wave
equation to one of the terms representing the particular plane wave with vector complex
amplitude E0

(
− j k x x+k y y +k z z )
E = E 0 e − jk ⋅r = E 0 e , (2.9)

where the scalar product k ⋅ r represents the projection of vector r into the direction of k.
These vectors are

r = x x0 + y y 0 + z z 0 ,

k = k x x 0 + k y y 0 + k z z 0 , k 2 = k x2 + k y2 + k z2 = ω 2 µε − jωµσ .

(2.9) describes the plane electromagnetic wave propagating in the direction determined by
vector k.

2.2 Propagation of a plane electromagnetic wave


A plane electromagnetic wave propagating in a general direction in an unbounded
space filled by a homogeneous material is described by the electric field intensity (2.9)

E = E0 e − jk ⋅r = E0 e − j (β − jα )⋅r = E0 e − α ⋅r e − jβ ⋅r = E0 e − α ⋅r e − jβ ⋅r e jφ 0 , (2.10)

where the generally complex propagation vector can be rewritten into a real part and an
imaginary part

k = β − jα . (2.11)

18
book - 2
The instantaneous value of an electric field is

{ }
E(r, t ) = Im E e jωt = E0 e − α ⋅r sin (ωt − β ⋅ r + φ0 ) . (2.12)

Term E 0 e − α⋅r represents the dependence of the wave


amplitude on position r. From the definition of the constant phase
propagation constant (2.5) it follows that both its real part
β and its imaginary part α are positive numbers.
Exponential function e −α⋅r then decreases with increasing ϕ β
value of its exponent. The amplitude therefore decreases r cos ϕ
and the wave is attenuated in the case of non-zero α. Fig. 2.1
Vector α is therefore known as an attenuation constant
(vector), as it describes the measure of the wave attenuation. The wave described by (2.10)
with a negative exponent therefore propagates in the positive direction – in the direction
defined by the propagation vector. The argument of the sinus function ω t − β ⋅ r + φ
0
determines the dependence of the field phase on time and coordinates. For this reason β is
called a phase constant (vector).
The surfaces of a constant phase – wave fronts – are planes determined by the
condition β ⋅ r = β r cos ϕ = const. We can see from Fig. 2.1 that the surface of a constant
phase is a plane perpendicular to phase vector β. The surfaces of a constant amplitude are
determined by α ⋅ r = const. , and they represent planes perpendicular to attenuation vector α.
The name plane wave is derived from the shape of these surfaces. A uniform wave has
surfaces of constant amplitude and of constant phase that are identical, as vectors α and β are
parallel. A non-uniform wave has surfaces of the constant phase different from surfaces of
the constant amplitude, as vectors α and β are not parallel.
The phase velocity is the velocity of the propagation of planes of a constant phase. Let
us follow the propagation of a plane with a constant phase φ which in a time increment ∆t
moves by a distance ∆r

φ = ωt − βr = ω (t + ∆t ) − β (r + ∆r ) ⇒ 0 = ω∆t − β∆r ,

from this we can define the phase velocity

∆r ω
v= = . (2.13)
∆t β

Group velocity represents the velocity with which the wave transmits energy, or the velocity
of the propagation of planes of a constant amplitude. We will derive the group velocity as the
velocity of the propagation of the amplitude of the superposition of two waves with
frequencies ω and ω+dω and with corresponding phase constants β and β+dβ and with
amplitudes equal to one. The superposition of these waves is

E ( z, t ) = E1 ( z, t ) + E 2 ( z, t ) = sin (ωt − βz ) + sin[(ω + dω )t − (β + dβ )z ] =


 dω t − dβ z   dω   dβ   .
= 2 cos  sin  ω + t −  β + z
 2   2   2  

19
book - 2
The amplitude of the resulting field is described by function cosine. In a time increment ∆t it
moves by a distance ∆z

φ = dωt − dβz = dω (t + ∆t ) − dβ ( z + ∆z ) ⇒ 0 = dω∆t − dβ∆z

from this we can define the group velocity

∆z dω 1
vg = = = . (2.14)
∆t dβ dβ

This velocity must in any case be lower than the velocity of light in a vacuum c.
As we will see later, the phase and the group velocities are not functions of frequency
in a lossless material – an ideal dielectric. This is not the case of a lossy material with nonzero
conductivity, where waves with different frequencies propagate with different velocities. Each
signal transmitting any information is represented by the spectrum of frequencies, and
consequently each component of this signal propagates with a different velocity. The result is
that the signal is distorted by passing the lossy material. This distortion is called dispersion
and the material is called dispersive. So only the ideal dielectric is a non-dispersive material
in which a passing signal is not distorted.
The wavelength is defined as the least distance, measured in the direction of
propagation, of two points with the same phase

sin (ωt − βz ) = sin[ωt − β ( z − λ )] ⇒ βλ = 2π ,

from this we can get

2π 2π v
λ= = = , (2.15)
β ω f
v

where f =ω/2π is the frequency.


In the preceding section we solved the wave equation for an electric field. The relation
between the electric field and the magnetic field can be derived from Maxwell’s equations.
We insert the solutions of the wave equation in the form

E = E0 e − jk ⋅r , H = H 0 e − jk ⋅ r ,

into Maxwell’s second equation (1.13). The rotation of vector E is (….)

( ) ( )
rot E = rot E0 e − jk ⋅r = e − jk ⋅r rot E0 + grad e − jk ⋅r × E0 = − jk e − jk ⋅r × E0 = − jk × E0 e − jk ⋅r ,

as the rotation of constant vector E0 is zero. Inserting into (1.13) we get

− jωµ H 0 e − jk ⋅r = − jk × E0 e − jk ⋅r

jωµ H 0 = jk × E0 . (2.16)

20
book - 2
Inserting into Maxwell’s first equation (1.12) we get

(σ + jωε )E0 = − jk × H 0 . (2.17)

It is evident from (2.16) and (2.17) that vectors E, H and k represent a right handed rotating
set of three vectors, Fig. 2.2. Vectors E and H are perpendicular to each other and both are
perpendicular to the direction of propagation determined by vector
E
k. For this reason a plane electromagnetic wave propagating in an
unbounded space filled by a homogeneous material is called a
transversal electromagnetic wave, abbreviated as TEM. We k
rewrite vector k as k=kn, where vector n is a unit vector H
determining the propagation direction. Now using the form of k
(2.5) we get from (2.17)
Fig. 2.2
− jkn × H 0 jωµ
E0 = = H ×n .
σ + jωε σ + jωε 0

The second root in this equation has the unit [Ω] and therefore it is known as the wave
impedance of the space

k ωµ jωµ
Zw = = = , (2.18)
σ + jωε k σ + jωε

it is generally a complex number which defines the ratio of E over H including the phase shift
between an electric field and a magnetic field. Consequently, the relations between an electric
field and a magnetic field are

E0 = Z w H 0 × n , (2.19)

1
H0 = n × E0 (2.20)
Zw

Let us now simplify the description of a plane electromagnetic wave by considering


that it propagates in the z direction. So we have k = k z 0 and r=z. Vectors E and H can then
be written as

E = Ex x0 + E y y 0 , H = H x x0 + H y y 0 .

Inserting these vectors into (2.19) we get


E x x 0 + E y y 0 = −Z w H x y 0 + Z w H y x 0 ,

using this equation we can define the wave impedance as

Ex Ey
Zw = =− . (2.21)
Hy Hx

21
book - 2
The complex propagation constant k was defined by (2.5). The phase constant and the
attenuation constant can be calculated by inserting the values of the material parameters and
frequency into (2.5). Alternatively α and β can be obtained from (2.5) in the form

1 
2
 σ 
β = ω µε 1+   + 1 , (2.22)
2  ωε  
 

1 
2
σ 
α = ω µε  1 +   − 1 . (2.23)
2  ωε  
 

The power transmitted by a plane electromagnetic wave is defined by the average


value of Poynting’s vector

S av =
1
{ }
1
{ 1
}
Re E × H * = Re E 0 e −αz e − jβz × H *0 e −αz e jβz = e − 2αz Re  0
(
 E × z 0 × E *0 ) =
2 2 2  Z w* 

 1 − 2αz  E 0 
2
 1
1
2  Z w
[( )
= e − 2αz Re  * E 0 ⋅ E *0 z 0 − (E 0 ⋅ z 0 )E*0 ]= e
 2
Re  *  z 0
 Z w 

2
E0
S av = e − 2αz cos ϕ z z 0 , (2.24)
2 Zw

where ϕz is the argument of wave impedance Z w = Z w e jϕ z (2.18). The power is transmitted


in the z direction, i.e., in the direction of the wave propagation.
Let us now discuss the propagation of plane electromagnetic waves in materials with
limit parameters. We will at the same time show the distribution of the electromagnetic field
of this wave in dependence on the time and space coordinates.
An ideal dielectric is a material with zero conductivity. The propagation constant and
the wave impedance are real in this material

k 2 = ω 2 µε = β 2 ⇒ β = ω µε , α =0 , (2.25)

µ µ0 120 π
Zw = = = . (2.26)
ε ε 0ε r εr

This means that the wave propagates with a constant amplitude, i.e., without losses, and such
a material is therefore called a lossless material. The electric field and the magnetic field are
in phase, as Zw is a real number. The electric field and the magnetic field and their
instantaneous values are, assuming E parallel with the x axis and H parallel with the y axis
and zero starting phase,
E0 − jβz
E = E0 e − jβz x 0 , H = H 0 e − jβz y 0 = e y0 ,
Zw

22
book - 2
E0
E( z , t ) = E 0 sin (ωt − βz ) x 0 , H(z, t ) = sin (ωt − βz ) y 0 .
Zw

These functions are plotted in Fig. 2.3 in dependence on time t for constant z and in
dependence on the z coordinate for constant t.
x z=0 m x ωt=π

E E

H t H z

y y

Fig. 2.3

A good conductor is a material in which we have ωε<<σ. The formulas for the
propagation constant (2.5) and the wave impedance (2.18) can under this condition be
simplified to

ωµσ
k = − jωµσ ⇒ k = β − jα = − jωµσ = (1 − j ) , (2.27)
2

ωµσ
β =α = , (2.28)
2

π
jωµ ωµ ωµ j 4 π
Zw = = (1 + j ) = e , ϕz = . (2.29)
σ 2σ σ 4

In a well conducting material the electric and magnetic fields have a mutual phase shift equal
to π/4, i.e., 45°. Generally the phase shift between E and H lies between 0° for a lossless
material and 45° for a good conductor.
Phase constant β and attenuation constant α have high values (2.28), and the wave is
rapidly attenuated. A material is sometimes characterized by penetration depth δ. This
quantity determines the length at which the wave amplitude decreases to 1/e (e=2.718281 is
the basis of the natural logarithm), i.e., to a value of 36.8 % of its starting value. The value of
the penetration depth can be derived

1 2
e −αδ = e −1 ⇒ δ = = . (2.30)
α ωµσ

____________________________
Example 2.1: Determine the penetration depth of a plane electromagnetic wave into copper
for frequencies of 50 Hz, 10 kHz, and 100 MHz.
Applying formula (2.30) we get
δ = 9.35 mm for 50 Hz
23
book - 2
δ = 0.66 mm
for 10 kHz
δ = 6.6 µmfor 100 MHz
At high frequencies the field does not penetrate at all into the conducting material.
____________________________

The electric field and the magnetic field and their instantaneous values are, in the case
of a lossy material

E0 −αz − jβz − jϕ z
E = E0 e −αz e − jβz x 0 , H= e e e y0 ,
Zw

E0 −αz
E( z , t ) = E 0 e −αz sin (ωt − βz ) x 0 , H(z, t ) = e sin (ωt − βz − ϕ z ) y 0 .
Zw

These functions are plotted in Fig. 2.4 in dependence on time t for constant z and in
dependence on the z coordinate for constant t, and a general lossy material is assumed. The
dependence on the z coordinate shows the attenuation of the wave according to function exp(-
αz).
x z=0 m x ωt=π
ϕz/ω
exp(-αz)
E E

H t H z

y y

Fig. 2.4

A dielectric material with non-zero conductivity σ which fulfils the condition

σ<<ωε (2.31)

can be called a lossy dielectric. The plane electromagnetic wave propagating in such a
material can be treated as a wave propagating in the dielectric only in the case of propagation
over a short distance, for which its attenuation can be neglected. A wave propagating over a
long distance can be substantially attenuated, even in the case of a low attenuation constant.
Under condition (2.31) we can accept the following simplification for the propagation
constant

σ  1 σ 
k = ω 2 µε − jωµσ = ω µε 1 − j ≈ ω µε 1 − j  ,
ωε  2 ωε 

1 µ
k = ω µε − j σ , (2.32)
2 ε

24
book - 2
σ µ 60πσ
β = ω µε , α= = . (2.33)
2 ε εr

We see that the phase constant is the same as the phase constant of the wave propagating in a
dielectric, but we have non-zero attenuation. Other parameters Zw, v, λ are determined using
the same formulas as in the case of a wave propagating in a dielectric.

________________________________
Example 2.2: Calculate the instantaneous value of voltage received by a frame antenna
located according to Fig. 2.5, due to the propagation of an electromagnetic wave

 z  z
E = 5 . 10 − 2 sin 10 8 t −  x 0 , H = 1.33 . 10 − 4 sin 10 8 t −  y 0
 3  3

We will calculate the received voltage in two ways. Firstly, according to the definition
of voltage. A closed loop is represented by an antenna perimeter, Fig. 2.5.


( )
 π 
u (t ) = ∫ E ⋅ dl = ∫ E ⋅ dl1 + ∫ E ⋅ dl 2 = −π E1 + π E2 = π 5.10− 2 − sin 108 t + sin 108 t − 
3 
c 1 2  

We can rewrite this result to the final form x c

 π π
u (t ) = −0.157 sin 10 8 t +  .
 3 E E1 E2
k
H dl1 dl2
The second way must result in the same
formula. We apply Faraday’s induction law
π z
y

u (t ) = −
dt
Fig. 2.5
The magnetic flux passing the antenna area is

π π
 z
φ = µ ∫∫ H ⋅ dS = µπ ∫ H dz = µπ 1.33 .10 −4
∫ sin10
8
t − dz =
S 0 0
3
  z 
(
= 157 .10 −11 cos10 8 t −  − cos 10 8 t 
3
)
  

dφ   π   π
u (t ) = −
dt
( )
= 0.157 sin 10 8 t −  − sin 10 8 t  = −0.157 sin 10 8 t +  .
3 3
   

The results are identical. Tilting the antenna into plane y-z, voltage u=0, as vectors dS and H
are perpendicular. In this way we can determine the direction of the wave propagation, i.e. the

25
book - 2
direction in which the transmitter is located. In this example we have the antenna dimensions
2π 2π
comparable with the wavelength, which is λ = = = 6π .
β 13
____________________________________

Example 2.3: Calculate the effective value of the voltage induced in a frame antenna, see Fig.
2.6, due to the propagating plane electromagnetic wave defined in Example 2.2.
In this case the antenna dimensions are much lower than the wavelength. We can
therefore simplify the calculation of the magnetic flux. This can be done by supposing
x
e − jβz ≈ e − jβ ( z + a ) , i.e. e − jβa ≈ 1
, a
i.e. βa << 1 .
E
k a
The magnetic flux is H
a=0.1 m
φ = B S cosψ ,
z
y
where ψ is the angle between vectors B and
dS. In our case this angle is 0°. The induced Fig. 2.6
voltage is

dφ dH U ωµ 0 S H
u=− = −µ0 S , U = − j ω µ0 S H , U ef = = = 1.18 µV .
dt dt 2 2

This value was obtained using the wave parameters from the previous example.
_______________________________________

Example 2.4: A plane electromagnetic wave propagates in the direction of the positive z axis
in air. It is incident perpendicularly to an ideally conducting plane located at z=0. Determine
the field distribution, which is the superposition of the incident and reflected waves supposing
that an electric field reaches at point z = -1 m and at time t = 0 s its maximum value
E ( z = −1, t = 0) = E max = 100 V/m . The frequency is 100 MHz. Calculate the power
transmitted by this wave.
An incident wave propagating in the positive z direction is

E i = E0i e − jkz e jϕ x 0 , H i = H 0i e − jkz e jϕ y 0 .

A reflected wave which propagates in the negative z direction is

E r = E0r e jkz e jϕ x 0 , H r = H 0r e jkz e jϕ y 0 .

The resulting field is the superposition of the incident and reflected waves

E = E i + E r = E0i e − jkz e jϕ x 0 + E0r e jkz e jϕ x 0

H = H i + H r = H 0i e − jkz e jϕ y 0 + H 0 r e jkz e jϕ y 0

26
book - 2
The relation between the amplitudes of the incident and reflected waves can be determined
from the boundary conditions. The electric field is parallel to the conducting plane, and due to
(1.22) its value must therefore be zero

E( z = 0 ) = E i ( z = 0 ) + E r ( z = 0 ) ⇒ E 0 i = − E 0 r .

The relation between the amplitudes of a magnetic


Ei kr
field follows from the orientation of the vectors of the
plane wave. The vectors of both the incident wave and Hi
ki Er
the reflected wave are mutually perpendicular, Fig. Hr
2.7. Consequently we have
Fig. 2.7
H 0i = H 0 r .

Now we can write the field distributions of both phasors and instantaneous values

( )
E = E0i e − jkz − e jkz e jϕ x 0 = −2 jE 0i sin (kz ) e jϕ x 0 = 2 E0i sin (kz ) e j (ϕ −π / 2 ) x 0

E(z , t ) = −2 E0i sin (kz ) sin (ωt ) x 0

( )
H = H 0i e − jkz + e jkz e jϕ y 0 = 2 H 0i cos(kz ) e jϕ y 0

H ( z , t ) = 2 H 0i cos(kz ) sin (ωt ) y 0

We now have a new kind of E (z )


electromagnetic wave. The
dependence on the z coordinate
and time is separated. This
indicates that it is not a
propagating wave but a standing
wave. The distribution of the
electric field and the magnetic −λ -3 λ/4 -λ/2 −λ/4
field of a standing wave is plotted H (z )
in Fig. 2.8. The electric field is
shifted by 90 deg corresponding to
the magnetic field. It is evident
from Fig. 2.8 that using an excited
standing wave we can measure the
wavelength, as the distance
between two adjacent minima or −λ -3 λ/4 −λ/2 −λ/4
maxima is equal to one half of the
wavelength. Fig. 2.8
The power transmitted by
the standing wave is determined
by Poynting’s vector

( )
S av = Re(E × H ) = Re 2 E0i sin (kz ) e j (ϕ −π / 2 ) x 0 × 2 H 0i cos(kz ) e jϕ y 0 = 0

27
book - 2
The power transmitted by the standing wave is zero, which is why it is called standing. This is
a natural result, as the incident wave transmits some power and no power is lost in the ideally
conducting wall. Consequently, the whole power is reflected back in the form of a reflected
wave. The total power, which is the vector sum of these two, must be zero.
The numerical results are:

ω 2π
k= = m −1 , Z w = 120π Ω ,
c 3

at t = 0 s and z = -1 m we have

 2π  π
E max = 100 = 2 E 0i sin  −  sin (ϕ ) ⇒ E 0i = 57.8 V/m , ϕ = -
 3  2

Consequently we have

 π
E(z , t ) = 115.6 sin (kz )sin  ω t −  x 0
 2

 π
H ( z , t ) = 30.6 cos(kz )sin  ω t −  y 0
 2

_________________________________________

2.3 Wave polarization


We showed that vectors E and H are mutually perpendicular and lie in a transversal
plane perpendicular to the direction in which the wave propagates. Their actual position in
this plane can, however, be quite general, and it can vary with time. The wave polarization
determines the way in which the end point of vector E moves in the transversal plane. Let us
find an equation which gives a general description of the behaviour of vector E in the plane
perpendicular to the direction of wave propagation. Let us suppose that this direction is
identical with the z axis and that we have a lossless material. Vector E is

E = E xm sin (ωt − kz + φ x ) x 0 + E ym sin (ωt − kz + φ y )y 0 .

Let us simplify this formula by choosing the position z = φ x k . Then we have

E x = E xm sin (ωt ) , (2.34)

E y = E ym sin (ωt + φ ) , (2.35)

where φ = φ y − φ x . We eliminate time from (2.34) and (2.35). From (2.34) we get

2
E  E 
sin (ωt ) = x , cos(ωt ) = 1 −  x  ,
E xm  E xm 

28
book - 2
and (2.35) gives

Ey
= sin (ωt + φ ) = sin (ωt ) cos φ + cos(ωt )sin φ .
E ym

Inserting for sin (ωt ) and cos(ωt ) we get

2
Ey E  E 
= x cos φ + 1 −  x  sin φ
E ym E xm  E xm 

This formula gives the equation

2 2
 Ey  Ey  E 
  − 2 Ex cos φ +  x  = sin 2 φ . (2.36)
E  E xm E ym
 ym   E xm 

Assuming general values Exm, Eym and φ this equation


represents the equation of an ellipse, Fig. 2.9, in an Ey
arbitrary position in the coordinate system Ex and Ey.
Consequently the end point of vector E moves along an
ellipse, and the wave is an elliptically polarized wave.
The position and the shape of this ellipse depend on Ex
values Exm, Eym and φ.
A special case is φ = nπ , in which we have
Fig. 2.9
cos φ = ±1, and sin φ = 0 . (2.36) then gives the equation

E ym
E y = ±Ex , (2.37)
E xm

Ey Ey Ey Ey
Eym=0 Exm=0 n odd n even

Ex Ex Ex Ex

Fig. 2.10

which is the equation of a line, Fig. 2.10, and our wave is a linearly polarized wave. Another
π
special case is φ = (2 n + 1) , in which cos φ = 0 and sin φ = ±1 . (2.36) then gives the
2
equation

2 2
 Ex   Ey 
  +   =1 , (2.38)
 
 E xm   E ym 

29
book - 2
of an ellipse in the basic position, Fig.
Ey
2.11. This wave has an elliptic Ey
Exm>Eym Exm<Eym
polarization. Assuming further equality
of amplitudes Exm=Eym this elliptically
polarized wave becomes a wave with a
circular polarization, Fig. 2.12. Ex Ex
In the case of a wave with an
elliptical or circular polarization we can
distinguish two types of polarization Fig. 2.11 Ey Exm=Eym
determined by the direction of vector E
rotation. If the angle difference φ is within the interval (0,π),
vector E rotates to the left, when observing the wave in the
direction of propagation, and the wave has a left-handed Ex
polarization, Fig. 2.13a. If the angle difference φ is within the
interval (π,2π), vector E rotates to the right, when observing the
wave in the direction of propagation, and the wave has a right- Fig. 2.12
handed polarization, Fig. 2.13b.
Circularly polarized waves
are widely used in technical left hand polarization right hand polarization
applications, e.g., transmission of y y
a signal in communications, etc. It
can be shown that the
superposition of two circularly ⊗ ⊗
polarized waves, one rotating to x z x z
the left, the second rotating to the
right, gives a wave with an φ ∈ (0, π ) φ ∈ (π ,2π )
arbitrary polarization. This is a b
important, e.g., when treating Fig. 2.13
propagation of waves in non-
isotropic materials, e.g., in a magnetized ferrite. The circularly polarized wave can be
represented by an instantaneous value

 π
E x = E 0 sin (ωt − kz ) , E y = E 0 sin  ωt − kz ±  = ± E 0 cos(ωt − kz ) ,
 2

E( z , t ) = E0 [sin (ωt − kz ) x 0 ± cos(ωt − kz ) y 0 ] ,

and consequently by

E = E0 (x 0 ± j y 0 ) e − jkz .

A survey of basic types of polarization is given in Tab. 2.1.

________________________________
Example 2.5: Decompose the generally elliptical polarized wave into the superposition of the
two circularly polarized waves, one with left-handed polarization, the second with right-
handed polarization.
We start solving a reciprocal problem in which we combine two circularly polarized
waves with opposite orientation of the electric field vector rotation. The waves with left-
handed and right-handed polarization are
30
book - 2
φ (deg) 0 45 90 135 180 225 270 315
Exm=0

Exm< Eym

Exm= Eym

Exm> Eym

Eym=0

left-handed right-handed

Tab. 2.1

E L = A (x 0 + jy 0 ) e − jkz , E R = B (x 0 − jy 0 ) e − jkz .

The superposition of these waves is

E = E L + E R = [( A + B )x 0 + (B − A) jy 0 ]e − jkz .

In the case A=B the resulting wave is linearly polarized, for A>B it is elliptically polarized
right-handed, for A<B elliptically polarized left-handed. Let us reverse this procedure. We
have a wave with general elliptical polarization

E = (Cx 0 + jDy 0 ) e − jkz .

Comparing this formula with the previous formula we get the amplitudes of the particular
waves with circular polarization

C−D C+D
A= , B= .
2 2
_________________________________________

Example 2.6: Calculate the power transmitted by a wave with generally elliptical
polarization.
A wave with elliptical polarization can be written in the form

E = Ax 0 e − jkz + By 0 e − jkz e − jϕ .

Poynting’s vector is

31
book - 2
S av =
1
2
(
Re E × H * =
1
)
2Z w
[ (
Re E × z 0 × E * = )]
1
2Z w
[( ) ( )]
Re Ax 0 e − jkz + By 0 e − jkz e − jϕ × A* y 0 e jkz − B * x 0 e jkz e jϕ =

1
2Z w
Re A 2 + B 2 z 0[ ]
_________________________________________

2.4 Cylindrical and spherical waves


We will solve the wave equation for a vector potential in a cylindrical coordinate
system in the free space filled by a lossless material without a source. We will assume that the
field does not depend on α and z. As a result, we obtain a wave which propagates from the z
axis to infinity with surfaces of a constant amplitude and a constant phase in the shape of
cylinders. Such a wave is an abstraction as a plane wave. The excitation of a cylindrical wave
supposes an infinitely long line source with a passing current of a constant amplitude and a
constant phase along this source.
The homogeneous wave equation for the z component of a vector potential A = A(r)z0
is (1.54)

∆A + k 2 A = 0 .

As the vector potential does not depend on α and z, we have

∂2 A 1 ∂A
2
+ + k2A = 0 . (2.39)
∂r r ∂r

This is a Bessel-type equation with a solution in the form of the sum of Bessel functions of
zero order of the first and second kind, see mathematical appendix,

A = C1 J 0 (kr ) + C 2 Y0 (kr ) = C1, H 01 (kr ) + C 2, H 02 (kr ) , (2.40)

where Hankel functions of the zero order and of the first and second kind are defined by
Bessel functions, see mathematical appendix (13.44) and (13.45). It is evident from their
asymptotical expression (13.46) and (13.47) that function H 02 represents a wave propagating
as kr increases toward infinity, whereas H 01 represents a wave propagating from infinity
towards the z axis. Assuming only a wave propagating from the z axis in an area of high kr,
i.e., far distant from the axis, we have finally the cylindrical wave described by the form

e − jkr
A=C . (2.41)
r

Why does the amplitude of this wave decrease as function 1 / r , as we have no


losses? The decrease in the amplitude is caused by the spread of power to space. Let us
calculate the density of the power transmitted by this wave

32
book - 2
D2 1
S av = r0 ,
2Z w r

where D is a constant and Zw is the wave impedance. The power transmitted through the
surface of a cylinder with radius r and length l is

D2
P = S av 2πrl = 2πl . (2.42)
2Z w

This power is constant, so we have no losses and it confirms why the field amplitude
decreases as 1 / r .
For high kr and a small space angle we can approximate this wave by a plane wave.
Let us now solve the wave equation in a spherical coordinate system, assuming that
the vector potential depends only on distance from the origin r. We will solve a homogeneous
equation, as we have a source-less space. The solution corresponds to a spherical wave. Such
a wave can be excited by an elemental omni-directional source placed at the origin. We then
solve the equation

1 1  2 ∂A  2
2 ∂r
r +k A=0 . (2.43)
r  ∂r 

The solution of this equation corresponding to a wave propagating from the origin towards
infinity is

C 2
A= H 1/ 2 (kr ) , (2.44)
r

using the formula for a Hankel function of order ½ and of the second kind (13.48), we get

e − jkr
A=C . (2.45)
r

The amplitude of this spherical wave decreases due to the spread of power as 1/r. This wave,
for high kr and within a small space angle, can be approximated by a plane wave.

____________________________
Example 2.7: An ideal omnidirectional antenna transmits a spherical wave. What power must
the antenna transmit to get transmitted power density Sav=1 mW/m2 at distance r1=1 km. What
power density corresponds to this at distance r2=1 m? Calculate the corresponding electric
field amplitude at r2=1 m. Assume a lossless material.
The total transmitted power is

P (r1 = 1000 ) = ∫∫ S av dS = S av 4πr12 = 12560 W .

The power density at r2 is

33
book - 2
P
S av (r2 = 1) = 2
= 1000 W/m 2
4πr2

As the power density is

2
1 E
S av = cos(ϕ z )
2 Zw

we get

2 S av Z w
E = = 868.1 V/m
cos(ϕ z )
_________________________________________

2.5 Problems
2.1 Write the solution of the wave equation describing a plane electromagnetic wave
propagating in the free space with parameters εr=8, µr=1, σ=40 S/m. The frequency is
40 GHz.
k = ω 2 µε − jωµσ = (3140 − j 2040 ) m −1
E = E0 e −2040 z e − j 3140 z
E(z , t ) = E0 e −2040 z sin (25.5.1010 t − 3140 z + ϕ )

2.2 The plane electromagnetic wave propagates in a free space filled by a material with
parameters εr=4, µr=1, σ=0. The frequency is 10 MHz. The electric field intensity has the
direction of the x axis and has a positive maximum equal to 10 V/m at z=0 and t=0. Calculate
the electric and magnetic field and their instantaneous values and Poynting’s vector at z=6 m
and t=10-8 s.
E x = 10 e jπ / 2 e − j 0.42 z = 10 e − j 0.94 V/m
 π
E x ( z , t ) = 10 sin  3.14 ⋅ 107 t − 0.42 z +  = −5.94 V/m
 2
jπ / 2 − j 0.42 z − j 0.94
H y = 0.053 e e = 0.053 e A/m
 π
H y ( z , t ) = 0.053 sin  3.14 ⋅ 107 t − 0.42 z +  = −0.0315 A/m
 2
2
S av = 0.26 z 0 W/m

2.3 A plane electromagnetic wave propagates in a free space filled by a material with
parameters εr=80, µr=1, σ=0.002 S/m. The frequency is 500 kHz. The electric field intensity
has the direction of the x axis and has a positive maximum equal to 50 V/m at z=0 and t=0.
Calculate the electric and magnetic field and their instantaneous values and Poynting’s vector.
E x = 50 e −0.039 z e jπ / 2 e − j 0.101z
 π
E x ( z , t ) = 50 e − 0.039 z sin  ω 7t − 0.101z + 
 2

34
book - 2
H y = 1.37 e −0.039 z e j (π / 2 − 0.366) e − j 0.101z
 π 
H y ( z , t ) = 1.37 e − 0.039 z sin  ω 7t − 0.1012 z + − 0.366 
 2 
−0.078 z 2
S av = 32 e z 0 W/m

2.4 Two plane electromagnetic waves propagate in the same direction in a lossy material.
The first wave has the frequency 1 GHz and the corresponding attenuation constant is β1=78
m-1. The second wave has the frequency 3 GHz and the attenuation constant is β2=83 m-1. The
amplitudes of these waves are equal at z=0. Determine the distance z in which E1=10E2.
ln 10
z= = 0.46 mm
β 2 − β1

2.5 A plane electromagnetic wave propagates in a free space filled by a material with
parameters εr=2, µr=1, σ=0.01 S/m. The frequency is 9 MHz. Calculate the electric and
magnetic field amplitudes at the point at which Sav=10 W/m2.
S av 2 Z w
E = = 48.8 V/m
cos ϕ z
E
H = = 0.58 A/m
Zw

2.6 To measure a receiving antenna we need a plane electromagnetic wave. To measure with
a sufficiently low error we can accept a change in field amplitude of 1% within a region 1 m
in length. The wave is excited by an ideal omnidirectional antenna transmitting a cylindrical
wave. Calculate the distance from the transmitting antenna at which we have to measure.
r=99 m

3. WAVES ON A PLANE BOUNDARY

3.1 Perpendicular incidence of a plane wave to a plane boundary


Let us solve the problem of the perpendicular incidence of a plane electromagnetic
wave on a plane boundary between two different materials, Fig. 3.1. An incident wave carries
some power. Part of this power is reflected back at the boundary in the form of a reflected
wave. In the first material, this wave is superimposed on the incident wave. Part of the power
is transmitted to the second material in the form of a transmitted wave. The incident wave is
described by the quantities marked by index i, the reflected wave is marked by index r, and
the wave transmitted to the second material is marked by index t. We assume that the
orientation of the vectors is according to Fig. 3.1. Our task is to determine the relations
between the amplitudes of particular waves. The propagation vectors of these waves are

k i = k1z 0 ,

k t = k2z 0 , (3.1)

k r = − k1z 0 ,

35
book - 2
where k1 and k2 are the propagation
ε1, µ1, σ1 ε2, µ2, σ2
constants of a plane wave in the first
and second materials, respectively. The Ei Et
total field in the first material is the ki kt
superposition of the incident and Hi
reflected waves Ht

0 z
Er
kr

Hr

Fig. 3.1

E1 = E i + E r = x 0 Eix e − jk1z + x 0 E rx e jk1z ,


k
(
H1 = Η i + H r = 1 z 0 × x 0 Eix e − jk1z − z 0 × x 0 E rx e jk1z =
ωµ 1
)
.
=
1
Z1
(
y 0 Eix e − jk1z − E rx e jk1z )
In the second material we have only the transmitted wave

E 2 = E t = x 0 Etx e -jk2 z ,

k2 1
H2 = Ht = y 0 Etx e -jk2 z = y 0 Etx e -jk2 z
ωµ 2 Z2

Both the electric field and the magnetic field are parallel to the boundary placed at z =
0. These fields must fulfill the boundary conditions for the tangential components (1.18) and
(1.121). So we have at z = 0

Eix + E rx = Etx , (3.2)

k1 k2
(Eix − E rx ) = Etx . (3.3)
µ1 µ2

This set of equations has the solution

µ 2 k1 − µ1k 2
E rx = Eix , (3.4)
µ 2 k1 + µ1k 2

36
book - 3
2µ 2 k1
Etx = Eix . (3.5)
µ 2 k1 + µ1k 2

Now we can define the reflection coefficient as

E rx µ 2 k1 − µ1k 2 Z 2 − Z1
R= = = , (3.6)
Eix µ 2 k1 + µ1k 2 Z 2 + Z1

and the transmission coefficient as

Etx 2µ 2 k1 2Z 2
T= = = , (3.7)
Eix µ 2 k1 + µ1k 2 Z 2 + Z1

where Z1 and Z2 are the wave impedances of the two materials (2.18). Values R and T are
generally complex. Inserting R and T from (3.6) and (3.7) into the boundary condition (3.2)
we get the relation between R and T in the form

T = 1+ R . (3.8)

Similarly as in (3.5) and (3.6), we can define the reflection and transmission coefficients for a
magnetic field. The reader can compare the relations for the reflection and transmission
coefficients for a plane electromagnetic wave, which is perpendicularly incident on the plane
boundary of two different
materials, with the incidence Z − Z1 2Z 2
of a wave in the Z1 Z2 R= 2 T=
Z 2 + Z1 Z 2 + Z1
transmission lines, see Fig.
3.2.
For a lossless Fig. 3.2
dielectric, the wave
impedance is defined by (2.26), and we have

Z 2 − Z1 ε r1 − ε r 2 n1 − n2
R= = = , (3.9)
Z 2 + Z1 ε r1 + ε r 2 n1 + n2

2Z 2 2 ε r1 2n1
T= = = , (3.10)
Z 2 + Z1 ε r1 + ε r 2 n1 + n2

where n is the refractive index

n = εr . (3.11)

When ε1<ε2 we have Z1>Z2 and consequently R<0, and the orientation of the electric field
vectors is according to Fig. 3.3a. The electric field is reflected out of phase. When ε1>ε2 we
have Z1<Z2 and consequently R>0 and the orientation of the electric field vectors is according
to Fig. 3.3b. Then electric field is reflected in phase.

37
book - 3
Let us now calculate the power Ei Et Ei Et
transmitted by the field in the first ki kt ki kt
material and the power carried by the
transmitted wave in the second material, Er
assuming that there are no losses in the kr
two materials. In the first material we Er kr
get
a b

Fig. 3.3

S1av =
1
2
( 1
) [( )(
Re E1 × H1* = Re x 0 Eix e − jk1z + x 0 E rx e jk1z × y 0 H iy* e jk1z − y 0 H ry
2
* − jk1z
e = )]
1
[
= Re Eix H iy* − E rx H ry
2
* * − 2 jk1z
− Eix H ry e + E rx H iy* e 2 jk1z z 0 = ] (3.12)
1
( 1
) [
= Re Eix H iy* z 0 + Re − R Eix R * H iy* − Eix H ry
2 2
* − 2 jk1z
e + E rx H iy* e 2 jk1z z 0 = ]
(
= S i av 1 − R
2
)
This is a logical result. The total power transmitted in the first material is equal to the power
transmitted by the incident wave reduced by the power transmitted by the reflected wave. The
power transmitted in the second material must be equal to the power transmitted in the first
material, as no power can be lost in the boundary itself.
Let us now assume that the second material is a perfect conductor with σ2 infinite
and consequently Z2=0. There is a zero field in such a material, i.e., no field penetrates into
this material. We get R = -1 and T = 0. The field in the first material is the superposition of the
incident and reflected waves. This was already solved in Example 2.4. The total field has the
character of a standing wave described, see Example 2.4, by the phasors of the electric field
and the magnetic field

E = −2 jE 0i sin (kz ) x 0 , (3.13)

H = 2 H 0i cos(kz ) y 0 , (3.14)

The instantaneous values of these fields are, see Example 2.4,

E( z , t ) = −2 E0i sin (kz ) cos(ω t ) x 0 , (3.15)

H ( z , t ) = 2 H 0i cos(kz )sin (ω t ) y 0 . (3.16)

This wave does not transmit any power. The distribution of its amplitudes is plotted in Fig.
2.8. As was mentioned in Example 2.4, the excitation of this standing wave is frequently used
to measure the wavelength, or consequently the frequency.
Let us now turn our attention again to the perpendicular incidence of a plane
electromagnetic wave to the plane boundary between two dielectric materials. Let us
analyze in greater detail the field in the first material, which is the superposition of the
incident and reflected waves, Fig. 3.1. We suppose that ε1>ε2 and consequently R>0. The x-
component of an electric field is

38
book - 3
[( )
E1x = Eix + E rx = E0 e − jk1z + R E0 e jk1z = E0 R e − jk1z + e jk1z + (1 − R ) e − jk1z , ]
[ ]
E1x = E0 2 R cos(k1 z ) + (1 − R ) e − jk1z , (3.17)

Similarly for a magnetic field we get

[ ]
H 1 y = H 0 − 2 jR sin (k1 z ) + (1 − R ) e − jk1z . (3.18)

In the case ε1<ε2 the formulae for an electric field and for a magnetic field are exchanged. The
first terms in field distributions (3.17) and (3.18) represent a standing wave with amplitude
2RE0 and 2RH0, whereas the second terms represent a wave traveling toward the boundary
with amplitude E0(1-R) and H0(1-R) and continuing in the second material as the transmitted
wave. Only the traveling part transmits power, it therefore verifies (3.12), and we get

S1av =
1
2
( 2
E0 H 0 1 − R . ) (3.19)

The distribution of the amplitude of an


electric field of this standing wave is λ/2 E
plotted in Fig. 3.4. We can identify the
minima and maxima of the amplitude, Emax=E0(1+R)
with their mutual distance equal to λ/2.
The same distribution of a voltage wave
and a current wave on a transmission line
can be measured, assuming that this wave
is incident to the junction between two
lines with different wave impedances, Emin=E0(1-R)
Fig. 3.2.
Standing waves are described by a z
standing wave ratio p, abbreviated as
SWR, defined by Fig. 3.4

E max 1 + R
p= = . (3.20)
E min 1 − R

This SWR is very often used in practical applications, as it can be measured simply.
Consequently, the measurement of a reflection coefficient or of an impedance is transposed to
the measurement of SWR.
As a last case, we will study the perpendicular incidence of a plane wave to a well
conducting material with conductivity σ. The case of the ideal conductor has been already
solved in Example 2.4, where the standing wave with p = ∞ was introduced. We assume that
the first material is air. So the wave impedance and the propagation constant for the two
materials are

µ0 ωµ 2σ ωµ 2
k1 = ω ε 0 µ 0 , Z1 = , k 2 = (1 − j ) , Z 2 = (1 + j )
ε0 2 2σ

39
book - 3
The transmission coefficient is

2Z 2 2
T= = . (3.21)
Z 2 + Z1 σ
1+ (1 − j )
2ωε 2

The transmitted wave can be expressed

2 Eix
Et = x 0 e − jk2 z , (3.22)
σ
1+ (1 − j )
2ωε 2
ε0, µ0 ε2, µ2, σ2
This intensity enables us to calculate the
total current passing a strip of the second
material of width equal to l, Fig. 3.5, as it Ei ki Et J2 kt
determines the current density (1.9). We l
get
Hi Ht
∞ ∞
I 2 = l ∫ J 2 x dz = lσ ∫ Etx dz = z
0 0

2σZ 2 l y
− jk2 z
= lσEtxm ∫ e dz = H tym =
0
jk 2 Fig. 3.5
= l H tym

The result is very simple

I 2 = l H tym = l K t , (3.23)

the total current is proportional to the value of the magnetic field just below the boundary,
which can be substituted by current density K. The power lost in the strip, Fig. 3.5, of length
d is

P=
1
2
( ) 1
2
(1
Re U I 2* = d Re Etxm I 2* = d l
2
ωµ 2

)
H tym
2
=
1
2
S ρ hf H tym
2
, (3.24)

where S=ld is the surface of the strip, and ρhf is the high frequency resistance of the surface of
a well conducting material

ωµ 2
ρ hf = . (3.25)

Formula (3.24) assumes homogeneous distribution of the magnetic field along the boundary.
The more general form is

40
book - 3
1 2
P=
2
ρ hf ∫∫ H t dS . (3.26)
S

So to calculate the power lost in a material we have to know only the distribution of the
tangent component of the magnetic field just at the surface of this material.

______________________________
Example 3.1: A plane electromagnetic wave with amplitude Eim=300 V/m is incident from
the air perpendicular to the surface of a conducting material with parameters εr=80, µ=µ0, σ=5
S/m, and the frequency is 500 kHz. Calculate:
a) The phasors and the instantaneous values of the reflected and transmitted waves.
b) The power passing the boundary.
c) The penetration depth and the wavelengths in the air and in the material.
d) The distance at which the field amplitude decreases to 1/100 of its amplitude at the
boundary.
The wave impedances and propagation constants are

jωµ 0
Z1 = 120π , Z2 = ≅ 0.888 e jπ / 4 ,
jωε + σ

k1 = ω ε 0 µ 0 = 0.0105 m −1 , k 2 = − jωµ 0σ = (1 − j )π m −1 .

The reflection and transmission coefficients are

Z 2 − Z1 2 Z2
R= ≅ −1 , T= = 0.0047 e jπ / 4 .
Z 2 + Z1 Z 2 + Z1

The phasors and instantaneous values of the electric and magnetic fields are

E i = x 0 300 e − j 0.0105 z , H i = y 0 0.796 e − j 0.0105 z ,

E r = x 0 300 e jπ e j 0.0105 z , H r = y 0 0.796 e j 0.0105 z ,

E t = x 0 1.410 e −πz e − jπz e jπ / 4 , H t = y 0 1.596 e −πz e − jπz ,

E i ( z , t ) = x 0 300 cos(ω t − 0.0105 z ) , H i ( z , t ) = y 0 0.796 cos(ω t − 0.0105 z ) ,

E r ( z , t ) = x 0 300 cos(ω t + 0.0105 z + π ) , H r ( z , t ) = y 0 0.796 cos(ω t + 0.0105 z ) ,

E t ( z, t ) = x 0 1.41e −πz cos(ω t − πz + π / 4) , H t ( z, t ) = y 0 1.596 e −πz cos(ω t − πz ) .

The particular power densities are

2
S iav
1
(
= Re E i × H *i = z 0
2
)
1 Ei
2 Z1
= z 0 119.4 W/ m 2 ,

41
book - 3
S rav =
1
2
( )
Re E r × H *r = − z 0 119.4 W/ m 2 ,

1  Et 
2 2
S tav
1
( *
= Re E t × H t = z 0 Re
2
)
2  Z2 
= z0
1 Et
2 Z2
cos(ϕ z ) = z 0 0.8 W/ m 2 .
 

The penetration depth is

1
δ = = 0.318 m .
β2

The distance z100 at which E(z100)=E(z=0)/100 is calculated

ln (100 )
e − β 2 z100 = 0.01 ⇒ z= = 1.488 m .
β2
______________________________________

I II II
T12 a1
a1 a1 a
a1 a 2 = a1e − jk2d 2
R21 b2  2 k2
b1
k1 k2 k3 k4 ... R12 a1  b = b e − jk2d 2
1 2 b2
d2 d3 b d2
b1 T21 b2
2

I II 
a b c
Fig. 3.6

3.2 Perpendicular incidence of a plane wave to a layered medium


Let us assume a plane electromagnetic wave incident from a half space filled by
material perpendicular to the system of layers of different materials, Fig. 3.6a. Our task is
to describe the field distribution in all layers and the field penetrating through this structure.
This problem can be effectively solved using transmission matrices. We derive separately the
transmission matrix describing the behaviour of the waves on the boundary between two
adjacent materials, Fig. 3.6b, and the transmission matrix of a single layer, Fig. 3.6c.
An electromagnetic wave with the complex amplitude of an electric field a1 is incident
from the left to the boundary between two different materials I and II, Fig. 3.6b. This wave is
partially reflected and partially transmitted through the boundary. Similarly, a wave with
amplitude b2 propagating from the right is reflected and transmitted. T12 is the transmission
coefficient from material I to material II. T21 is the transmission coefficient from material II to
material I. R12 represents the reflection from the left, R21 is the reflection coefficient from the
side of material II. The wave traveling from the boundary to the left has amplitude b1, Fig.
3.6b

b1 = R12 a1 + T12 b 2 . (3.27)

42
book - 3
Amplitude a2 of the wave propagating from the boundary to the right is, Fig. 3.6b

a 2 = R21 b 2 + T12 a1 . (3.28)

From (3.27) and (3.28) we get, using R12 = -R21 and T12T21 – R12R21 = 1,

a1  1  1 R12  a 2 
b  =  1  b 2 
. (3.29)
 1  T12  R12

So the transmission matrix describing the boundary between the two materials is

 1 R12 
[A I,II ] = T1 R 1 
. (3.30)
12  12

Now we derive the transmission matrix of only a single layer, Fig. 3.6c. In this layer
we have one wave propagating to the left with amplitude b2 and one wave propagating to the
right with amplitude a1. These two waves passing the layer of thickness d2 can be expressed

a 2 = a1e − jk2 d 2 , b1 = b 2 e − jk2 d 2 ,

we consequently get the transmission matrix

a1  e jk 2 d 2 0  a 2 
b  =    , (3.31)
 1   0 e − jk 2 d 2  b 2 

So the transmission matrix describing the behaviour of the waves in the single layer is

 jk2 d 2 
[A 2 ] = e 0
 . (3.32)
 0 e − jk2 d 2 

The total transmission matrix of the layered structure, Fig. 3.6a, is the product of
particular matrices of type (3.30) and (3.31)

a1  1  1 R12  e jk 2 d 2 0  1  1 R23  e jk 3 d 3 0  a N 


b  = T R 
1   0
  1 
  . . .  . (3.33)
 1  12  12 e − jk 2 d 2  T23  R23  0 e − jk 3 d 3  b N 

Structures of this kind, Fig. 3.6a, are frequently used namely in optical wavelength
technology as an antireflection coating which can provide even frequency selective behaviour.
These structures can serve as band pass filters, band stop filters, etc. They can be designed in
a similar way as microwave filters composed of sections of transmission lines of different
wave impedances.

_______________________________
Example 3.2: Design an inter-layer placed between two different dielectric materials to get
zero reflection, a so called antireflection layer.

43
book - 3
The structure is shown in Fig. 3.7. There is a
dielectric layer of thickness d placed between two dielectric a1 a3
layers. Our task is to design the thickness and the ε1 ε2 ε3
permittivity of this inter-layer to get zero reflection, i.e., to b1
get b1 = 0. The transmission matrix of our structure is d b3=0
according to (3.33) the product of three matrices

 a1  1  1 R12  e jk 2 d 0  1  1 R23  a 3 
b  = R 
1   0 − jk2 d  R 1   0 
Fig. 3.7
 1  T12  12 e  T23  23

From this set of equations we get

a1 =
1
T12T23
(
e jk2d + R12 R23 e − jk 2d a 3 , )

b1 =
1
T12T23
(
R12 e jk2d + R23 e − jk2d a 3 . )
The reflection coefficient is

b1 R e jk 2d + R23 e − jk2d R12 + R23 e −2 jk2d


R= = 12 = .
a1 (
e jk2d + R12 R23 e − jk2d )
1 + R12 R23 e −2 jk2d

Inserting for particular reflection coefficients R12 and R23 from (3.9) we get

(n1 − n2 )(n2 + n3 ) + (n2 − n3 )(n1 + n2 ) e −2 jk2d


R= .
(n1 + n2 )(n2 + n3 ) + (n1 − n2 )(n2 − n3 ) e −2 jk2d
For the next procedure we assume two possible simplifications relating refractive indices
n1>n2>n3 or n1<n2<n3. Firstly, we have to get the real and minimum value of the reflection
coefficient. This supposes

e −2 jk2 d = −1 ⇒ 2k 2 d = π + 2mπ ,

where m is an integer. As k2 = 2π/λ2, we get the value of the antireflection layer thickness

λ2 λ2
d= +m . (3.34)
4 2

The minimum inter-layer thickness is equal to a quarter of the wavelength of a wave in the
second material. Now from the condition R = 0 we get

(n1 − n2 )(n2 + n3 ) − (n2 − n3 )(n1 + n2 ) = 0 ,


which gives us the condition of permittivity

44
book - 3
ε r 2 = ε r1 ε r 3 or n2 = n1 n3 . (3.35)

Results (3.34) and (3.35) are parameters of the well known quarter-wavelength transformer
frequently used in microwave engineering.
__________________________________

3.3 Oblique incidence of a plane electromagnetic wave to a plane boundary


In the case of the oblique incidence of a plane electromagnetic wave to a plane
boundary between two different materials
and we have to treat two cases x Hr kr
separately. We have to differentiate the two Ei
cases of polarization of an incident wave. Er
We assume a wave with a linear Hi ϕi ki
polarization. In the first case, vector E is kix
kiz n
perpendicular to the plane of incidence,
which is defined by normal vector n ϕi ϕr
together with the propagation vector of an
incident wave ki, Fig. 3.7. This case is
marked as the incidence of a wave with y z
horizontal or perpendicular polarization.
This notification follows from the theory of
wave propagation above the earth ground ϕt
and, e.g., their incidence to the ionosphere. Ht Et
The second case is called the incidence of a
wave with vertical or parallel kt
polarization, as vector E is parallel to the
plane of incidence.
Horizontal polarization. In this Fig. 3.8
case we assume the orientation of all
vectors as defined in Fig. 3.8. Our task is to determine the relation between the amplitudes of
the electric fields of all three waves. To do this we have to force an electric field and a
magnetic field to fulfill the boundary conditions for tangential components at plane x = 0
(1.18) and (1.20), assuming that no current passes along the boundary. Let us first express the
propagation vectors, Fig. 3.8, in their components

k i = −k ix x 0 + k iz z 0 =
, (3.36)
= −k1 cos(ϕ i ) x 0 + k1 sin (ϕ i ) z 0

k r = k1 cos(ϕ r ) x 0 + k1 sin (ϕ r ) z 0 , (3.37)

k t = −k 2 cos(ϕ t ) x 0 + k 2 sin (ϕ t ) z 0 , (3.38)

So the electric and magnetic fields of the incident wave can be expressed in the form

Ei = y 0 Ei 0 e − jk i ⋅r = y 0 Ei 0 e jk1 cos(ϕi )x e − jk1 sin (ϕi )z , (3.39)

45
book - 3
k i × Ei Ei 0
Hi = = (− cos(ϕi ) z 0 − sin (ϕi ) x 0 ) e jk1 cos(ϕi )x e − jk1 sin (ϕi )z , (3.40)
ωµ1 Z1

For the reflected and transmitted waves we have

E r = −y 0 E r 0 e − jk r ⋅r = −y 0 E r 0 e − jk1 cos(ϕ r )x e − jk1 sin (ϕ r )z , (3.41)

E t = y 0 Et 0 e − jk t ⋅r = y 0 Et 0 e jk2 cos(ϕt )x e − jk2 sin (ϕt )z , (3.42)

k r × Er Er 0
Hr = = (− cos(ϕ r ) z 0 + sin (ϕ r ) x 0 ) e − jk1 cos(ϕr )x e − jk1 sin (ϕr )z , (3.43)
ωµ 1 Z1

k t × Et Et 0
Ht = = (− cos(ϕ t ) z 0 − sin (ϕ t ) x 0 )e jk2 cos(ϕt )x e − jk2 sin (ϕt )z , (3.44)
ωµ 2 Z2

The boundary conditions at x = 0 are

Eiy (0, z ) − E ry (0, z ) = Ety (0, z ) , (3.45)

H iz (0, z ) + H rz (0, z ) = H tz (0, z ) (3.46)

Inserting into (3.45) the expressions for the electric field from (3.39) and similar expressions
for Er and Et (3.41) and (3.42) we get

Ei 0 e − jk1 sin (ϕi )z − E r 0 e − jk1 sin (ϕ r )z = Et 0 e − jk2 sin (ϕt )z . (3.47)

This condition must be fulfilled for each coordinate z. This means that the exponents in all
terms must be equal. This results in

k1 sin (ϕ i ) = k1 sin (ϕ r ) = k 2 sin (ϕ t )

This equation defines Snell’s first and second laws

ϕi = ϕ r , (3.48)

k1 sin (ϕ i ) = k 2 sin (ϕ t ) . (3.49)

Snell’s first law tells us that the angle of incidence equals the angle of reflection. Snell’s
second law can be used to calculate the angle of transmission. Boundary condition (3.47) now
reduces to

Ei 0 − E r 0 = Et 0 . (3.50)

Similarly, inserting (3.40) and analogous formulas for Hr and Ht (3.43) and (3.44) into (3.46),
and applying Snell’s laws, we get

46
book - 3
Ei 0 E E
− cos(ϕ i ) − cos(ϕ i ) r 0 = − cos(ϕ t ) t 0 . (3.51)
Z1 Z1 Z2

Reflection coefficient R⊥ and transmission coefficient T⊥ for the oblique incidence of the
plane wave with horizontal polarization can be now derived from the set of equations (3.50)
and (3.51)

E r 0 Z1 cos(ϕ t ) − Z 2 cos(ϕ i ) µ1k 2 cos(ϕ t ) − µ 2 k1 cos(ϕ i )


R⊥ = = = , (3.52)
Ei 0 Z1 cos(ϕ t ) + Z 2 cos(ϕ i ) µ1k 2 cos(ϕ t ) + µ 2 k1 cos(ϕ i )

Et 0 2Z 2 cos(ϕ i ) 2µ 2 k1 cos(ϕ i )
T⊥ = = 1 − R⊥ = = . (3.53)
Ei 0 Z1 cos(ϕ t ) + Z 2 cos(ϕ i ) µ1k 2 cos(ϕ t ) + µ 2 k1 cos(ϕ i )

Note that the derived formulas hold only for Ei x Er


the vector orientation shown in Fig. 3.7. kr
Vertical polarization. In this case we
assume the orientation of all vectors as Hi ki Hr
defined in Fig. 3.9. The procedure for
n
calculating the reflection and transmission
coefficients is similar as the procedure used ϕi ϕr
in the case of horizontal polarization. We
exchange the form of electric field vectors by
the form of magnetic field vectors. The y z
propagation vectors are defined by (3.36) to
(3.38). The field vectors, Fig. 3.9, are Et
ϕt
Ht

kt

Fig. 3.9

E i = Ei 0 (cos(ϕ i ) z 0 + sin (ϕ i ) x 0 ) e jk1 cos(ϕi )x e − jk1 sin (ϕi )z , (3.54)

Ei 0 jk1 cos(ϕi )x − jk1 sin (ϕi )z


Hi = y0 e e , (3.55)
Z1

E r = E r 0 (− cos(ϕ r ) z 0 + sin (ϕ r ) x 0 ) e − jk1 cos(ϕ r )x e − jk1 sin (ϕ r )z , (3.56)

Er 0 − jk1 cos(ϕ r )x − jk1 sin (ϕ r )z


Hr = y0 e e , (3.57)
Z1

E t = Et 0 (cos(ϕ t ) z 0 + sin (ϕ t ) x 0 ) e jk2 cos(ϕt )x e − jk2 sin (ϕt )z , (3.58)

47
book - 3
Et 0 jk2 cos(ϕt )x − jk2 sin (ϕt )z
Ht = y0 e e . (3.59)
Z2

Field distributions (3.54) to (3.59) are inserted into boundary conditions (1.18) and (1.20)
assuming K = 0. We get using Snell’s laws the set of equations

cos(ϕ i )(Ei 0 − E r 0 ) = cos(ϕ t ) Et 0 , (3.60)

(E i 0 + E r 0 ) Et 0
= . (3.61)
Z1 Z2

The reflection and transmission coefficients can be calculated from these two equations

Z1 cos(ϕ i ) − Z 2 cos(ϕ t ) ε 2 k1 cos(ϕ i ) − ε 1k 2 cos(ϕ t )


R = = , (3.62)
Z1 cos(ϕ i ) + Z 2 cos(ϕ t ) ε 2 k1 cos(ϕ i ) + ε1k 2 cos(ϕ t )

2Z 2 cos(ϕ i ) 2ε 1k 2 cos(ϕ i )
T = = . (3.63)
Z1 cos(ϕ i ) + Z 2 cos(ϕ t ) ε 2 k1 cos(ϕ i ) + ε1k 2 cos(ϕ t )

Angle ϕt can be expressed in formulas (3.52), (3.53) and (3.62), (3.63) by the angle of
incidence

2
 k sin (ϕ t ) 
k 2 cos(ϕ t ) = k 2 1 − sin (ϕ t ) = k 2
2
1 −  1  = k 22 − k12 sin 2 (ϕ i ) . (3.64)
 k2 

The reflection and transmission coefficients in the case of both vertical and horizontal
polarization depend not only on the parameters of the material but also on the angle of
incidence. In the case of materials with nonzero conductivity these coefficients are complex
numbers. In many cases the two materials are dielectrics. Then using (3.64) and µ1 = µ2 = µ0
we can express the reflection coefficients in the form

ε1 ε
− cos(ϕ i ) + 1 − 1 sin 2 (ϕ i )
ε2 ε2
R⊥ = , (3.65)
ε1 ε1
cos(ϕ i ) + 1 − sin (ϕ i )
2
ε2 ε2

ε2 ε
cos(ϕ i ) − 1 − 1 sin 2 (ϕ i )
ε1 ε2
R = . (3.66)
ε2 ε1
cos(ϕ i ) + 1 − sin (ϕ i )
2
ε1 ε2

Reflection coefficients (3.65) and (3.66) are plotted in Fig. 3.10 assuming that ε1/ε2 = 3, and
ε1/ε2 = 1/3. In the following text we will discuss some special cases of the oblique incidence
of a plane electromagnetic wave to a plane boundary between two different materials.

48
book - 3
1 1

R
R ε1/ε2=1/3 ε1/ε2=3
0.8 0.8
ϕc
0.6 - 0.6 +
R⊥ R⊥
0.4 0.4
R  - R 
0.2 ϕB 0.2 ϕB
+ - +
0 0
0 30 60 90 0 30 60 90
ϕi (°) ϕi (°)

Fig. 3.10

______________________________ the plane of


Example 3.3: A plane linearly polarized electromagnetic incidence
wave with amplitude Ei0 = 10 V/m is incident to a plane
boundary between two dielectric materials with permittivities Ei║
εr1=1 and εr2=8. Its frequency is 30 MHz. The angle of Ei0
incidence is ϕi = 30°. The electric field contained with the
plane of incidence angle α0 = 45°. Calculate the electric and α0 Hi0
magnetic fields and the transmitted power of the reflected
wave and the transmitted wave. Calculate power transmitted
through the boundary. Ei┴ kr
An incident wave must be decomposed to two waves,
the first one polarized perpendicular to the plane of Fig. 3.11
incidence, while the second wave has a parallel polarization,
see Fig. 3.11, where the wave is observed in its direction of propagation. The two particular
waves have amplitudes

Ei ⊥ = Ei 0 sin (α 0 ) = 7.07 V/m

Ei = Ei 0 cos(α 0 ) = 7.07 V/m

The angle of transmission is

 k1 
ϕ t = arcsin  sin (ϕ i ) = 0.177 ⇒ ϕ t = 10.2°
 k2 

The amplitudes of the electric field of the particular wave with perpendicular and parallel
polarization are

µ1k 2 cos(ϕ t ) − µ 2 k1 cos(ϕ i )


E r ⊥ = Ei ⊥ = −0.53 Ei ⊥ = −3.75 V/m
µ1k 2 cos(ϕ t ) + µ 2 k1 cos(ϕ i )

49
book - 4
2 µ 2 k1 cos(ϕ i )
Et ⊥ = Ei ⊥ = 0.47 Ei ⊥ = 3.32 V/m
µ1k 2 cos(ϕ t ) + µ 2 k1 cos(ϕ i )

ε 2 k1 cos(ϕ i ) − ε1k 2 cos(ϕ t )


Er = Ei = 0.43 Ei = 3.05 V/m
ε 2 k1 cos(ϕ i ) + ε1k 2 cos(ϕ t )

2ε 1k 2 cos(ϕ i )
Et = Ei = 0.5 Ei = 3.54 V/m
ε 2 k1 cos(ϕ i ) + ε1k 2 cos(ϕ t )

Now we can draw diagrams reflected wave transmitted wave


showing the orientation of the plane of plane of
electric field vectors in a
incidence incidence
reflected wave and in a
transmitted wave, Fig. 3.12.
The amplitudes of the reflected Er║ Et║
and transmitted waves and the Et0
angles of their declination from Er0
the plane of incidence, Fig. αt Ht0
αr
3.12, are kr Et┴

Er┴ kt
E r 0 = E r2⊥ + E r2 = 4.8 V/m
Hr0
(a) (b)
Er ⊥
α r = arctg = −51° Fig. 3.12
Er

Et 0 = Et2⊥ + Et2 = 4.9 V/m

Et ⊥
α t = arctg = 43°
Et

To calculate the amplitudes of a magnetic field we have to calculate the wave impedances

µ0 µ0
Z1 = = 120π Ω , Z2 = = 133 Ω ,
ε0 ε 0ε r 2

H i 0 = 0.0265 A/m , H r 0 = 0.0127 A/m , H t 0 = 0.0367 A/m

The densities of the active power transmitted by particular waves are

1
S iav = Ei 0 H i 0 = 0.1325 W/m 2 ,
2

1
S rav = E r 0 H r 0 = 0.0304 W/m 2 ,
2

50
book - 4
1
S tav = Et 0 H t 0 = 0.09 W/m 2 .
2

To calculate the power transmitted through the boundary we have to decompose these powers
to components parallel and perpendicular to the boundary, Fig. 3.13. The power transmitted
through the boundary is

S xav = S iav cos(ϕ i ) − S rav cos(ϕ i ) = S tav cos(ϕ t ) = 0.0886 W/m 2 ,

The power transmitted through the boundary must be of


the same value when calculated from both sides. The
power transmitted along the boundary is
ϕi
S z1 = S iav sin (ϕ i ) + S rav sin (ϕ i ) = 0.08145 W/m 2 , Siav Srav

S z 2 = S tav sin (ϕ t ) = 0.0156 W/m 2 . Stav

The power transmitted in the direction parallel to the ϕt


boundary has different values in the two materials.
____________________________________
Fig. 3.13

Total transmission. Fig. 3.10 shows that the reflection coefficient for the vertical
polarization equals zero at a certain angle. This angle is called Brewster’s angle ϕB. This
means that there is no reflection at Brewster’s angle, and the whole power is transmitted to
the second material. Putting (3.66) equal to zero, we get

ε2 ε2
sin (ϕ B ) = , or tg (ϕ B ) = . (3.67)
ε1 + ε 2 ε1

This angle is sometimes called Brewster’s polarization angle. This name is given because an
incident wave with an elliptic polarization when it is incident under this angle makes a
reflection of only the linearly polarized wave, as there is no reflection of the component with
an electric field parallel to the plane of incidence. This effect is used in polarizing elements.

Total reflection. The plot in Fig. 3.10 drawn for the case ε1>ε2 shows that the
reflection coefficients for the waves of the two kinds of polarization are equal to one starting
from a certain angle. This angle is called the critical angle ϕc. The value of this angle follows
from Snell’s law (3.49). In the case ε1<ε2 we have so called refraction to a normal, Fig.
3.14a, in which ϕi>ϕt. In the case ε1>ε2 we have so called refraction from a normal, Fig.
3.14b, at which ϕi<ϕt. Now angle ϕt can be equal to 90° assuming that ϕi = ϕc. From (3.49)
we get for ε1>ε2

51
book - 4
ϕi ϕi
ε1 ε1

ε2 ε2
ϕt
ϕt
a b
Fig. 3.14

ε2
sin (ϕ c ) = . (3.68)
ε1

At angles ϕi > ϕc the refraction angle has a complex value and R = 1 , Fig. 3.10. This means
that the whole power is reflected back. This of course does not mean that there is no field in
the second material.
Let us now investigate the field distribution created after total reflection on the
boundary between two dielectric materials with permittivities ε1>ε2 and assuming ϕi > ϕc. We
assume only a wave with a horizontal polarization to simplify the derived formulas. The
results are also valid in the case of a wave with a vertical polarization. Angle ϕt has a complex
value, and from (3.49) we have

ε1
sin (ϕ t ) = sin (ϕ i ) > 1 , cos(ϕ t ) = 1 − sin 2 (ϕ t ) = j cos(ϕ t ) .
ε2

Now the field in the second material is, Fig. 3.8 and (3.42),

k2 x cos (ϕ t )
E t = T⊥ Ei 0 y 0 e − jk2 [-x cos(ϕ t )+ z sin (ϕ t )] = T⊥ Ei 0 y 0 e e − jk2 z sin (ϕ t ) . (3.69)

This formula describes a so-called surface wave. The amplitude of this wave decreases
exponentially in the direction into the second material and propagates in the direction of axis z
along the boundary with the propagation constant

k 2 z = k 2 sin (ϕ t ) = k1 sin (ϕ i ) . (3.70)

The phase velocity of this wave is

ω ω ω
v pz = = = < v p2 . (3.71)
k2z k 2 sin (ϕ t ) k1 sin (ϕ i )

52
book - 4
Due to this the surface wave is called a
slow wave in the second material. This is x
the case of a non-uniform vpz>v1
electromagnetic wave, as planes of a fast wave
constant amplitude are parallel with the
boundary, and at the same time planes of
a constant phase are perpendicular to the

boundary. The field distribution is plotted λx =
in Fig. 3.15. k1 cos(ϕ i )
In the case of total reflection we
have R =1 and consequently Ey
R = exp( jψ ) The field in the first
material is the superposition of the vpz<v2
incident and reflected waves. These slow wave
waves are (3.39) and (3.41)

E i = y 0 Ei 0 e − jk1[− x cos(ϕi )+ z sin (ϕi )] ,


Fig. 3.15

E r = −y 0 E r 0 e − jk1[x cos(ϕi )+ z sin (ϕi )] ,

their superposition is

[ ]
E1 = E i + E r = y 0 Ei 0 e jk1x cos (ϕi ) − R⊥ e − jk1x cos(ϕi ) e − jk1z sin (ϕi ) =
[ ]
= y 0 E i 0 e j [k1x cos(ϕi )−ψ / 2] − e − j [k1x cos(ϕi )−ψ / 2] e jψ / 2 e − jk1z sin (ϕi ) = . (3.72)
= y 2 jE sin[k x cos(ϕ ) − ψ / 2]e jψ / 2 e − jk1z sin (ϕi )
0 i0 1 i

The magnetic field can be derived in a similar way using distributions (3.40) and (3.43)

Ei 0
H1 = H i + H r = 2 {− x 0 sin (ϕ i )sin[k1 x cos(ϕ i ) − ψ / 2] −
Z1
− z 0 cos(ϕ i ) cos[k1 x cos(ϕ i ) − ψ / 2]}e jψ / 2 e − jk1z sin (ϕi ) . (3.73)

(3.72) and (3.73) describe a wave which propagates in the z direction and has the character of
a standing wave in the x direction. The distribution of the electric field is plotted in Fig. 3.15.
The phase velocity of this wave

ω ω ω
v pz = = > v p1 = . (3.74)
k1z k1 sin (ϕ i ) k1

For this reason, the surface wave in the first material is known as a fast wave.
Later we will explain the principle of a dielectric waveguide on the basis of the total
reflection and of this surface wave.

53
book - 4
The oblique incidence of a plane wave to the surface of a lossy material. Let us
calculate the distribution of the electric field in the second lossy material, assuming horizontal
polarization. The transmitted wave is (3.42)

E t = y 0T⊥ Ei 0 e − jk2 [− x cos(ϕt )+ z sin (ϕt )] .

The second material is a lossy material with nonzero conductivity, consequently its phase
constant is a complex number k2 = β2 – jα2. From Snell’s law (3.49) it follows that sin(ϕt) is a
complex number and cos(ϕt) = a + jb. Inserting these values into the formula for an electric
field we get

E t = y 0T⊥ Ei 0 e x (α 2a − β 2b )e − j [− x ( β 2a +α 2b )+ k1z sin (ϕt )] . (3.75)

This formula describes a non-uniform


electromagnetic wave. The planes of a
ϕi
constant amplitude are parallel with
the boundary, Fig. 3.16, whereas the ε0, µ0
planes of a constant phase are
determined by the equation ε2, µ2, σ2
plane of a constant phase
x  β 2 a + α 2b  − k1z sin ϕ t  = const . ϕT
plane of a constant
amplitude
This gives the angle under which the
wave propagates in the second
material, Fig. 3.16,
Fig. 3.16

k1 sin (ϕ i )
tg (ϕ T ) = − . (3.76)
β 2 a + α 2b

Increasing the conductivity of the second material causes a rise of α2 and β2, which means
that angle ϕT decreases to zero. Finally a wave in a well conducting material propagates
perpendicular to the boundary, independently of the angle of incidence, and is a uniform wave
which is of course attenuated very fast.

3.4 Problems
3.1 A plane electromagnetic wave is incident from the air to the plane surface of a dielectric.
A reflection causes a standing wave in the air with standing wave ratio p=2.7. Calculate the
permittivity εr of the dielectric and reflection coefficient R.
There are two solutions: R=0.46, εr=0.138
R=-0.46, εr=7.2

3.2 What permittivity has a dielectric, the surface of which reflects at most 1% of the energy
of a wave incident perpendicular.
0.67<εr<1.5

54
book - 4
3.3 A plane electromagnetic wave is incident from the air to the plane surface of sea water
with parameters εr=81 and σ=5 S/m. The frequency is 10 MHz, and the amplitude of the wave
is Eim=100 V/m. Calculate then electric field intensity at depth 1 m under the surface.
Et (1) = 0.00175 V / m

3.4 Design an antireflection layer on the surface of a silicon photodiode for light with
wavelength 0.8 µm. The permittivity of silicon is 12.
d=(0.11+m0.22) µm
εr2=3.46

3.5 Calculate the complex amplitude of a wave passing the layered structure from Fig. 3.6.
The amplitude of the incident wave is Ei=10 V/m, the parameters of the structure are: εr1=1,
εr2=4, εr3=6, d=8 cm, and the frequency is 10 GHz.
a3=6e-j2π/3 V/m

3.6 Calculate the average value of Poynting’s vectors representing the power transmitted by
the surface wave described by (3.69) and (3.72).
2
k 2 T⊥ Ei20
S 2 zav = sin (ϕ t )
2ωµ 0
2 Ei20  ψ 
S1zav = sin (ϕ i )sin 2 k1 x cos(ϕ i ) − r 
Z1  2 
There is no active power transmitted in the x direction.

3.7 Calculate the slant of an output glass window with permittivity εr=2.13, as a wave with a
parallel polarization passes through it without
losses, Fig. 3.17.
The angle must be equal to Brewster’s angle.
ϕ = ϕB = 55.6°
ϕ εr

Fig. 3.17

3.8 A plane electromagnetic wave is incident to the plane boundary between two dielectrics
with permittivities εr1=2.53, εr2=1. Calculate the minimum angle of incidence at which total
reflection takes place.
ϕic = 33°

3.9 A plane electromagnetic wave with perpendicular polarization is incident from the air to
the plane surface of a material with parameters εr2=1.5, σ2=2.10-5 S/m. The frequency is f = 2
MHz, and the angle of incidence is ϕi = 60°. Calculate the direction of propagation of a wave
transmitted into the second material, its attenuation and phase constants and phase velocity.
ψT = 44°
vp2 = 2.42.108 m/s
β 2 = 0.0519 m-1
α 2 = 0.0382 m-1

55
book - 4
3.10 Calculate the angle of propagation of the wave transmitted into copper after the
incidence of a plane electromagnetic wave from the air under angle ϕi = 45°. The frequency is
2 MHz the, and conductivity of copper is σ = 5.8.107 S/m.
ψT = 0°

3.11 Calculate the phase velocity of the surface wave which is excited by the total reflection
of a plane electromagnetic wave that is incident from a dielectric with permittivity εr1=4 to its
boundary with the air. The angle of incidence is ϕi = 45°, and the frequency is 10 MHz.
vpz = 2.16.108 m/s

4. SOLUTION OF MAXWELL EQUATIONS AT VERY HIGH


FREQUENCIES
We will assume the propagation of an electromagnetic wave in a generally
nonhomogeneous lossless medium. The medium will be described by the distribution of the
refractive index n( x, y, z ) = n(r ) = ε r (r ) . We assume a very high frequency, so that the
wavelength is very short, much shorter than the distances at which the refractive index varies
significantly. We assume the distribution of the electric field of a wave propagating in our
medium in a form analogous to a plane electromagnetic wave

E = E m (r ) e − jk0φ (r ) , (4.1)

where vector function Em(r) determines the distribution of the field amplitude and function
φ(r) determines the distribution of the phase, and k0 is the phase constant in the vacuum
k 0 = ω µ 0ε 0 . The equation Em(r) = const defines the planes of a constant amplitude, and the
equation φ(r) = const defines planes of a constant phase, known as wave-fronts. These two
sets of planes are generally different. Nevertheless, we will show that a propagating wave has
locally the same properties as a plane TEM wave propagating in a homogeneous space.
Let us insert (4.1) and a similar formula for the magnetic field into Maxwell’s
equations (1.12), (1.13) and (1.14)

( )
rot H m e − jk0φ = jωε E m e − jk0φ , (4.2)

( )
rot E m e − jk0φ = − jωµ H m e − jk0φ , (4.3)

( )
div ε E m e − jk0φ = 0 . (4.4)

Using formula (13.78) from the mathematical appendix describing the rotation applied to the
product of two functions, we rewrite (4.2)

e − jk0φ rot (H m ) − jk 0 e − jk0φ gradφ × H m = jωε E m e − jk0φ .

Reducing the exponential function we get

56
book - 5
rot (H m ) − jk 0 gradφ × H m = jωε E m , (4.5)

and similarly from equation (4.3)

rot (E m ) − jk 0 gradφ × E m = − jωµ H m . (4.6)

Using formula (13.70) for the divergence of the product of two functions, we rewrite (4.4) to

( )
E m ⋅ grad ε e − jk0φ + ε e − jk0φ div(E m ) = 0

[ ]
E m ⋅ e − jk0φ grad(ε ) − jk 0 e − jk0φ ε gradφ = −ε e − jk0φ div(E m ) .

Finally we have

E m ⋅ [grad(ε ) − jk 0 ε gradφ ] = −ε div(E m ) . (4.7)

Rewriting (4.5), (4.6) and (4.7) we get

ε0 1
gradφ × H m + ε r Em = rot (H m ) , (4.8)
µ0 jk 0

ε0 1
gradφ × E m − Hm = rot (E m ) , (4.9)
µ0 jk 0

1  gradε 
gradφ ⋅ E m =  Em + div(E m ) . (4.10)
jk 0  ε 

Now we use the assumption that we treat the field at very high frequencies. This means that
1
f → ∞ , and consequently k 0 → ∞ and → 0 . We can now set the right hand sides of
jk 0
(4.8), (4.9) and (4.10) equal to zero.

ε0
gradφ × H m + ε r Em = 0 , (4.11)
µ0

ε0
gradφ × E m − Hm = 0 , (4.12)
µ0

gradφ ⋅ E m = 0 . (4.13)

As φ(r) = const determines the wave-fronts, and vector grad φ is perpendicular to these wave-
fronts we have a very important result. According to (4.13) vectors gradφ and Em are
perpendicular. This means that vector Em is tangent to the wave-front, consequently it is
perpendicular to local direction of the wave propagation.
Inserting now from (4.12) for Hm to (4.11) we have
57
book - 5
ε0 ε
gradφ × (gradφ × E m ) = − 0 ε r E m .
µ0 µ0

Rewriting the double vector product (13.11) we have

gradφ (gradφ ⋅ E m ) − E m (gradφ ⋅ gradφ ) = −ε r E m .

Owing to (4.13) we have

2
gradφ = ε r = n2 , (4.14)

this equation couples the wave-fronts with the distribution of the relative permittivity. We are
thus able to determine the distribution of phase φ solving the partial differential equation

2 2 2
 ∂φ   ∂φ   ∂φ  2
  +   +   = n . (4.15)
 ∂x   ∂y   ∂z 

The mutual relations between the electric and magnetic fields follow from (4.11) and
(4.12)

µ 0ε 0
Em = - gradφ × H m , (4.16)
ε

µ 0ε 0
Hm = gradφ × E m . (4.17)
µ

Using (4.14), vector gradφ can be expressed as

grad φ = grad φ l 0 = nl 0 , (4.18)

where unit vector l0 determines the local direction of the wave Em gradφ = nl0
propagation. (4.16) and (4.17) tell us that vectors Em and Hm are
mutually perpendicular and are perpendicular to the direction of
the wave propagation at each point, Fig. 4.1. This means that a
wave propagating at very high frequencies in a generally r
nonhomogeneous medium behaves locally as a TEM wave
propagating in a free space. Hm
The power transmitted by this wave is defined by
Poynting’s vector. Locally this vector again corresponds to the
Poynting vector of a plane TEM electromagnetic wave Fig. 4.1

1 1 ε
Re[E m × H m ] =
2
S av = Em l0 . (4.19)
2 2 µ

The power is transmitted in the l0 direction, i.e., perpendicular to the wave-fronts.

58
book - 5
The row of vectors l0 represents a ray. Let us derive the equation describing the ray.
We will start with equation (4.18), where,
according to Fig. 4.2, l0 = dr/ds. Deriving
(4.18) over s we have l0

d  dr  d ray
 n(r )  = grad[φ (r )]
ds  ds  ds φ s
ds
dr
Let us now pay attention to the right hand
side part of this equation φ+dφ
r
d dx ∂ ∂φ
grad φ = ∑ i ∑ ∂x x j 0 , r+dr
ds i ds ∂xi j j
0
where xi is the i-th coordinate, i.e., x, y or z. Fig. 4.2
Changing the order of the summation we get

dx ∂ ∂φ ∂ ∂φ dxi ∂ dr
∑ dsi ∂x ∑ ∂x x j0 = ∑ x j0 ∑ = ∑ x j0 grad[φ (r )] ⋅ =
i i j j j ∂x j i ∂xi ds j ∂x j ds

= ∑ x j0 n(r ) l 0 ⋅ l 0 = grad[n(r )]
j ∂x j
Consequently we get

d  dr 
 n(r )  = grad[n(r )] , (4.20)
ds  ds 

which is the equation describing the trajectory of the ray. The changes in the direction of the
rays are caused by changes in the refraction index. Solving (4.20) we are able to find the rays
and to represent the wave by rays. In the case of a homogeneous material we have n = const,
consequently grad(n) = 0 and

d  dr 
 n(r )  = 0 => r = sa + b
ds  ds 
dS2
where a and b are constant vectors. The ray P1
represents a straight line.
φ2
According to (4.19) the power transmitted by
the wave propagates along rays. From this we can a tube of
estimate the value of the field amplitude. Let us take constant
an area dS1 on the wave-front φ1. The wave transmits power
power P1 through this area, Fig. 4.3. The rays
starting at the perimeter of area dS1 circumscribe a
tube of constant power and mark on wave-front φ2 P1
φ1
area dS2 through which the same power P1 passes.
This power is dS1
Fig. 4.3

59
book - 5
1 ε 2 1 ε 2
P1 = E m1 dS1 = E m 2 dS 2 .
2 µ 2 µ

From this equation we get

dS1
E m 2 = E m1 . (4.21)
dS 2

The amplitude of the electric field is higher where the area is smaller, which corresponds to
the higher density of the rays.
This technique using rays to describe waves is known as the geometric optic. The
waves are described by a set of rays which propagate independently. Their trajectories are
described by (4.20). If they fall on any boundary, they behave according to Snell’s laws.

________________________________
Example 4.1: Determine the trajectory of the ray which propagates at the plane y = 0 in a
dx
dielectric layer, see Fig. 4.4. The boundary conditions are x(z=0) = 0, = tg (ϑ0 ) ≈ ϑ0 . The
dz
refractive index depends on the x component

( )
n( x ) = n0 1 − k x 2 , where constant k << 1 .

This is known as the parabolic distribution of z


the refractive index. Assume that the ray
changes its direction very little in the relation ϑ0
to the z axis, so we can put s = z. Such a ray
is known as a par-axial ray.
The trajectory of the ray is described x
by (4.20). This equation can be generally Fig. 4.4
solved only numerically. Assuming a par-
axial ray we have s ≈ z , and (4.20) now reads

d  dr  dn ( x )
 n( x )  = x0 , r = xx0 + zz0 .
dz  dz  dx

d   dx  dn( x )
 n( x ) x 0 + z 0  = x0
dz   dz  dx

dn
as = 0 we get
dz

d  dx  dn ( x ) d2x d2x 2 k n0 x
n( x )  = , => n( x ) = −2 k n0 x => =−

dz  dz  dx dz 2 dz 2
(
n0 1 − kx 2 )
Now we neglect kx2 comparing to 1 and we get

60
book - 5
d2x
= −2 k x .
dz 2

The solution of this equation is

x = A sin (qz ) + B cos(qz ) , q = 2k .

Applying the boundary conditions we get

x(0) = 0 => B=0

dx dx ϑ0
= A q cos(qz ) , = ϑ0 => A= ,
dz dz z =0 2k

ϑ0 z0 zp
x=
2k
(
sin 2k z . ) z
ϑ0 xmax
A part of the ray trajectory is shown in x
Fig. 4.5. The parameters of the Fig. 4.5
trajectory are

π zp ϑ0
zp = , z0 = , x max =
2k 2 2k

The ray is coupled to the area around the z axis, and it bends and returns back. For x < 0 the
ray has a symmetric shape.
__________________________________

5. GUIDED WAVES
Up to now we have studied electromagnetic waves propagating in a free space filled
by a homogeneous or non-homogeneous material. This is very important for the theory of
wave propagation in the atmosphere, and is used in the design of communication systems,
particularly the channels represented by transmitting and receiving antennas, and the space
between them. On the other hand, there are waves the existence of which is based on the
presence of a boundary between different materials. These boundaries can be of various
shapes, depending on the required behaviour and the application, and they form a
transmission line. We have studied the surface wave excited due to the incidence of a plane
electromagnetic wave on a plane boundary of two dielectric materials under an angle greater
than the critical angle. This surface wave is one from the examples of guided waves.
A wide variety of transmission lines are used. Their geometry depends on many
factors. The most important are the technology of the system or the circuit in which the line is
applied, the frequency band and the transmitted power. Transmission lines can be generally
categorized into three groups according to the form of the transmitted wave. In the first group,
there are transmission lines which are able to transmit a TEM wave. Later we will see that
such a transmission line must be able to transmit DC current, so it must consist of at least two

61
book - 5
separate conductors. Examples of these transmission lines are shown in Fig. 5.1a. The two
transmission lines shown in Fig. 5.1b consist of two conductors, and are thus able to conduct
DC current. These are planar lines with conducting strips located on a dielectric substrate. The
electric field and the magnetic field have to fulfill boundary conditions on the surfaces of this
substrate. Due to this, the longitudinal components of the fields are always present. The
propagating wave is not a TEM wave. At low frequencies this wave can be treated as a TEM
wave. Consequently it is known as a quasi TEM wave. The lines shown in Fig. 5.1c do not
guide a TEM wave. The first two lines in Fig. 5.1c are known as waveguides as they guide a
wave along their hollow center.

two-conductor
transmission line
co-axial
transmission line
parallel-plate
waveguide

Fig. 5.1a Transmission lines with a TEM wave

strip-line

microstrip line coplanar waveguide


dielectric
Fig. 5.1b Transmission lines with a quasi-TEM wave

waveguide with a rectangular waveguide with a circular optical fibre


cross-section cross-section

Fig. 5.1c Transmission lines without a TEM wave


To simplify the analysis of transmission lines we will assume lines without losses, i.e.,
dielectric materials with zero conductivity, and metals with infinite conductivity. The lines

62
book - 6
will be assumed to be infinitely long and longitudinally homogeneous. The particular
materials will be homogeneous. The lines will be directed in the positive z-axis direction,
which is thus equal to the direction of the wave propagation.
The field distribution of a wave propagating along the transmission line in the z-axis
direction will be calculated by solving the homogeneous wave equation (2.1). Therefore we
will get only the waves, that can propagate along the line. These waves are known as the
eigenmodes, and represent particular solutions of the wave equation. Let us apply the method
of separation of variables for solving the wave equation. The longitudinal component of
electric field Ez is assumed in the form

E z ( x, y, z ) = E 0 z ( x, y )P( z ) . (5.1)

This form of the field distribution is inserted into the wave equation (2.1) and is rewritten to

∆ T E0 z 1 d 2 P
+ + k2 = 0 (5.2)
E0 z P dz 2

where ∆ T represents Laplace’s operator calculated using derivatives over the transversal
coordinates x and y. The first and second terms on the left hand side of (5.2) must to be equal
to constants kp and kz, which represent the transversal propagation constant and the
longitudinal propagation constant. In this way we decompose equation (5.2) into two
equations for E0z and P

∆ T E0 z + k 2p E0 z = 0 , (5.3)

P ' '+ k z2 P = 0 . (5.4)

The propagation constants kp and kz are coupled by

k 2 = k p2 + k z2 . (5.5)

Equation (5.4) has the solution describing the phase variation of the wave propagating along
the line in the form

P ( z ) = e − jk z z . (5.6)

We omitted here the wave propagating in the negative z direction. All components of the
electric and magnetic fields have the same character as Ez has. Consequently the derivatives
of these components over z can be expressed simply. So for the i-th component of the electric
field we have

dE i
= − jk z E i . (5.7)
dz

The character of the field described by (5.1) enables us to divide the modes
propagating along the transmission line into two groups, and the modes in each group can be
treated separately. These modes are the transversal electric (TE) modes and the transversal
magnetic (TM) modes. The TE modes have Ez=0, and their field contains components: Ex,

63
book - 6
Ey, Hx, Hy, Hz, while in complement the TM modes have Hz=0, and their field contains
components: Ex, Ey, Ez, Hx, Hy. The field distribution and the propagation constants of these
modes are determined by solving the wave equation (5.3) for the longitudinal components of
the electric and magnetic fields, and the transversal components are derived from the
longitudinal components.
We start from Maxwell’s first and second equations (1.12) and (1.13), putting σ=0 and
Js=0. These two vector equations can be expressed in the scalar form

∂H z
+ jk z H y = jω ε E x , (5.8)
∂y

∂H z
− jk z H x − = jω ε E y , (5.9)
∂x

∂H y ∂H x
+ = jω ε E z , (5.10)
∂x ∂y

∂E z
+ jk z E y = − jω µ H x , (5.11)
∂y

∂E z
− jk z E x − = − jω µ H y , (5.12)
∂x

∂E y ∂E x
+ = − jω µ H z . (5.13)
∂x ∂y

From (5.12) we express Hy and insert it into (5.8). Now (5.8) contains only Ex, Ez, and Hz and
in this way we get the dependence Ex on the longitudinal components of the electric and
magnetic fields. Similarly we get the other transversal components. So we have

jω µ  ∂H z k ∂E z 
Ex = −  + z  , (5.14)
k p  ∂y ω µ ∂x 
2 

jω µ  ∂H z k ∂E z 
Ey = − − + z  , (5.15)
k p  ∂x ω µ ∂y 
2 

jω ε  ∂E z k z ∂H z 
Hx =  −  , (5.16)
k p2  ∂y ω ε ∂x 

jω ε  ∂E z k z ∂H z 
Hy = −  +  . (5.17)
k p2  ∂x ω ε ∂y 

Now putting Ez=0 we will get the transversal components Ex, Ey, Hx, Hy as functions of
Hz and we have TE modes. Conversely, putting Hz=0 we will get the transversal components
Ex, Ey, Hx, Hy as functions of Ez and we have TM modes.

64
book - 6
Let us now turn our attention to (5.5). Using k = ω µ ε we get the longitudinal
propagation constant

k z = ω 2 µ ε − k p2 . (5.18)

Three cases can be now distinguished. At high frequencies we have

ω 2 µ ε > k p2

and kz is a real number, so the corresponding mode represents a wave propagating along the
transmission line. At low frequencies we have

ω 2 µ ε < k 2p

and kz is an imaginary number, so the corresponding mode represents an evanescent wave


which does not propagate along the transmission line and its amplitude exponentially
decreases. The boundary point determines the so called cut-off frequency of the mode

kp
fc = . (5.19)
2π µ ε

This means that the transmission line behaves the mode does not the mode propagates
as a high-pass filter, as it transmits the mode propagate
starting from the cut-off frequency, Fig. 5.2.
Note that TEM modes have, as will be shown,
0 fc f
kp=0 and, consequently, fc=0, and they can
propagate from zero frequency. Fig. 5.2
Each transmission line has its dominant
mode. This is the mode with the lowest
possible cut-off frequency, i.e., with the lowest kp. It is desired to operate the transmission line
in the frequency band of the single mode operation. This is the frequency band at which
only the dominant mode propagates. This band is limited from above by the cut-off frequency
of the nearest higher mode. When the two modes can propagate simultaneously, the
transmitted signal is coupled to these two modes. As they have different propagation
constants (5.18), they have different phase velocities and the field of these modes arrives at
the output port with a different delay. This results in distortion of the transmitted signal.
The particular transmission lines and the modes propagating along them will be
studied in the following sections.

6. TEM WAVES ON A TRANSMISSION LINE

6.1 Parameters of a TEM wave


Let us study a transmission line with a general cross-section, homogeneous along its
infinite length, located in the unbounded homogeneous space parallel to the z-axis. Let the
line have perfect conductors with infinite conductivity. We will study the propagation of a

65
book - 6
TEM (transversal electromagnetic) wave along this transmission line. The TEM wave has
field components only in the transversal plane and Ez=0, Hz=0. On an infinitely long
homogeneous line we assume dependence on the longitudinal coordinate in the form e− jk z z .
Thus we have

E(x, y, z ) = E T ( x, y ) e − jk z z , H ( x, y, z ) = H T ( x, y ) e − jk z z . (6.1)

The transversal field components have the form ET = E x x 0 + E y y 0 , H T = H x x 0 + H y y 0 .


Inserting these forms into wave equations (1.41) and (1.43) in the space without sources we
get

∆ T E T + k p2 ET = 0

∆ T H T + k 2p H T = 0

where ∆ T represents Laplace’s operator applied according to transversal coordinates x and y,


and k 2p = k 2 − k z2 = ω 2 µε − k z2 is a transversal propagation constant. As both the electric and
magnetic fields have zero longitudinal components, these fields have lines of E and H lying
only in the transversal planes and therefore they must have a constant phase in these planes.
This determines kp = 0 ant thus kz = k. As a result we have

∆T HT = 0 , (6.2)

∆ T ET = 0 , (6.3)

Equations (6.2) and (6.3) are Laplace equations. The electric and magnetic field transversal
components are solutions of the Laplace equation, and this electromagnetic field therefore has
the character of a stationary field, of course except for its wave character in the longitudinal
direction. As the field distribution in the transversal plane is the same as the distribution of the
stationary field, the line must be able to conduct a DC current. Such a transmission line must
consist of at least two conductors to transmit the DC current. Generally, a TEM wave can
propagate only along a transmission line which consists of at least two conductors.
As the field has the character of a stationary field we can unambiguously define the
voltage and the current at any point on the transmission line. The voltage is defined by the
integral along an arbitrary path c1 lying in the transversal plane

u ( z , t ) = ∫ E T ⋅ ds . (6.4)
c1 z = const .

This voltage depends only on the z-coordinate and time t. The electric current is determined
by the integral along the closed path c2 which surrounds one of the conductors (Ampere’s
law), and lies in the transversal plane

i ( z , t ) = ∫ H T ⋅ dl . (6.5)
c2 z = const .

66
book - 6
This current depends only on the z-coordinate and time t.
Now we can express the electric field and the magnetic field using the voltage and the
current

E T ( x, y , z , t ) = e T ( x, y ) u ( z , t ) , (6.6)

H T ( x, y , z , t ) = h T ( x, y ) i ( z , t ) . (6.7)

Vector function eT(x,y) represents the distribution of the electric field when DC voltage 1 V is
connected to the line terminals. Inserting (6.6) into (6.4) we get

∫ e ( x , y ) ⋅ ds = 1 .
c1
T (6.8)

Vector function hT(x,y) represents the distribution of the magnetic field when DC current 1 A
passes the transmission line. This function is normalized to one, as follows from the insertion
of (6.7) to (6.5),

c2
∫ h T ( x , y ) ⋅ dl = 1 . (6.9)

The problem of calculating the field distribution is thus divided into two parts. The
calculation of functions eT(x,y) and hT(x,y) can be done according to techniques known from
the electrostatic field and the stationary magnetic field, solving (6.2) and (6.3). The reader is
therefore able to solve this part of the problem. The calculation of voltage u(z,t) and current
i(z,t) is known from circuit theory. This is described by the well known telegraph equations.
Assuming a steady harmonic state, these equations are

∂U
= − ( R + jω L ) I , (6.10)
∂z

∂I
= −(G + jωC )U . (6.11)
∂z

R, L are line series resistance and inductivity per unit length, G and C are line parallel
conductance and capacity per unit length. These primary parameters of a transmission line
can be calculated by methods known from electromagnetic field theory. We can eliminate
either the voltage or the current from the couple of these equations to get one equation of the
second order

∂ 2U
− (R + jωL )(G + jωC )U = 0 , (6.12)
∂z 2

∂2I
− (R + jωL )(G + jωC )I = 0 . (6.13)
∂z 2

These equations are equivalent to the wave equation. The propagation of the TEM wave along
the line is characterized by secondary parameters. These are propagation constant γ,
characteristic impedance ZC, phase velocity v, and wavelength λ.

67
book - 6
We start by writing the solution of the telegraph equation for voltage U (6.12)

∂ 2U
2
− γ 2U = 0 , (6.14)
∂z

where the propagation constant is written from (6.12)

γ = (R + jωL )(G + jωC ) . (6.15)

Here we take the notation of the propagation constant, which is known from circuit theory.
The lossless transmission line has R=0 and G=0 and the propagation constant will be

γ = − ω 2 LC = jω LC = jβ , (6.16)

where β is the phase constant in the sense of (2.11) valid for a plane electromagnetic wave
propagating in a homogeneous space filled by a lossless material. Note that in most of the
formulas valid for the TEM wave propagating along the transmission line we can substitute L
for µ and C for ε , and we get formulas valid for the plane electromagnetic wave propagating
in free space.
The solution of (6.14) and the corresponding equation valid for the electric current are,
assuming a single wave propagating on an infinitely long line in the positive z direction,

U = U 0 e −γz , (6.17)

I = I 0 e −γz , (6.18)

where U0 and I0 are wave amplitudes. Inserting into (6.12) for the voltage and the current we
get

− γ U 0 e −γz = −(R + jωL ) I 0 e −γz ,

and from this formula we get, using (6.15), the characteristic impedance of the transmission
line, which is defined as the ratio of the voltage amplitude and the current amplitude,

U 0 R + jωL R + jωL
ZC = = = . (6.19)
I0 γ G + jωC

A lossless line has

L
ZC = . (6.20)
C

The voltage and the current are in phase on a lossless line, as the characteristic impedance is a
real number. We can again compare (6.20) with the characteristic impedance of a lossless
material for a plane electromagnetic wave (2.26).

68
book - 6
The phase velocity is defined and can be determined in the same way as the phase
velocity of a plane electromagnetic wave in free space (2.13). Assuming a lossless line we can
get

ω 1
v= = . (6.21)
β LC

This velocity determines the velocity of the propagation of a constant phase along the line. In
the case of a lossless line it also determines the velocity of the energy transmitted along the
line, i.e., the group velocity.
The wavelength has again the same meaning as the wavelength of a plane
electromagnetic wave propagating in an unbounded space (2.15)

v 2π
λ= = . (6.22)
f β

___________________________
Example 6.1: Calculate the secondary parameters of the co-axial transmission line shown in
Fig. 6.1, assuming lossless materials. The line
parameters are 2r1=0.46 mm, 2r2=1.5 mm,
εr=2. Calculate the distribution of the electric -τ
field and the magnetic field in the cross-
section of this transmission line.
From the theory of electromagnetic
field we know the line capacity per unit
length
τ
r3
2πε
C= , r2
r2
ln
r1 εr
r1
and the line inductivity

µ 0 r2
L= ln .
2π r1
Fig. 6.1
The characteristic impedance is, according to
(6.20),

µ 0 r2
ln
L 2π r1 60 r
ZC = = = ln 2 = 50.13 Ω
C 2πε 0ε r ε r r1
r2
ln
r1

It is evident that the given parameters correspond to the standard 50 Ω co-axial cable. The
phase constant (6.19) is

69
book - 6
µ 0 r2 2πε
β = ω LC = ω ln = ω µε = k 0 ε r ,
2π r1 r2
ln
r1

where k0 is the phase constant of the wave propagating in a vacuum. The wavelength (6.22) is

2π 2π λ0
λ= = = ,
β k0 ε r εr

where λ0 is the wavelength of a plane electromagnetic wave in a vacuum.


The distribution of the electric field and the magnetic field in the cross-section of this
transmission line is known from electromagnetic field theory

τ U I
Er = = , Hα = .
2πε r r ln (r2 r1 ) 2πr

Note that, using the formula for characteristic impedance ZC of this transmission line, ratio
E r H α gives a value equal to the characteristic impedance for a plane electromagnetic wave
propagating in the space filled by a dielectric with permittivity εr.
________________________________

The results of the above example can be generalized. The TEM wave propagates along
a transmission line with a homogeneous dielectric material between the conductors in the
same way as a plane electromagnetic wave propagating in the unbounded space filled by the
same homogeneous material. This refers to the propagation constant, the wavelength, and the
phase velocity. The characteristic impedance is different from the characteristic impedance of
the material for a plane wave. The characteristic impedance of a transmission line is defined
as the ratio of the amplitudes of the voltage and of the current. So it must depend not only on
the material parameters but also on the line transversal dimensions. The characteristic
impedance of the material for a plane wave is defined as the ratio of the field amplitudes, so it
depends only on the material parameters.

______________________________
Example 6.2: Calculate the secondary parameters of the two-conductor transmission line
(twin-lead transmission line) from Fig. 6.2, assuming lossless materials.
In electromagnetic field theory we calculated the line capacity and inductivity per unit
length

πε
C= .
a
ln
r

µ0 a r r
L= ln .
π r a

The characteristic impedance (6.20) is


Fig. 6.2

70
book - 6
µ0 a
ln
L π r = 120 ln a .
ZC = =
C πε εr r
a
ln
r

The phase constant and the wavelength are

λ0
β = ω LC = ω µ 0ε 0ε r = k 0 ε r , λ= .
εr
_____________________________

The results of the above example validate the remarks made to Example 6.1.
The TEM wave is not the only wave that can propagate along the above mentioned
lines. So called waveguide modes, treated in Chapter 7, can propagate along these lines at
sufficiently high frequencies above their cut-off frequency. These modes are undesired, and
the lines must be designed to prevent the propagation of waveguide modes.

6.2 Transformation of the impedance along the line


6.2.1 An infinitely long line
The general solution of the telegraph equations (6.10) and (6.11) on an infinitely long
transmission line consists of two waves, one propagating to the right (denoted by subscript +),
and the second propagating to the left (denoted by subscript -)

U = U + e −γz + U − eγz , (6.23)

U + −γz U − γz
I = I + e −γz + I − eγz = e − e , (6.24)
ZC ZC

assuming the current amplitudes I + = U + Z C , I − = − U − Z C , as the current flowing to the


left has the opposite orientation. The impedance at any point along the line is

U U + e −γz + U − e γz
Z (z ) = = Z C , (6.25)
I U + e −γz − U − eγz

On an infinitely long line with the only one source connected at z = −∞ , we have only one
wave propagating to the right, and the above formulas read
U = U + e −γz , (6.26)

I = I + e −γz , (6.27)

The impedance at any point along the line is

71
book - 6
U+
Z (z ) = = ZC . (6.28)
I+

The impedance at any point on an infinitely long transmission line is thus equal to the
characteristic impedance. This gives us the recipe for realizing an infinitely long line. A line
of finite length is terminated by an impedance equal to ZC. This ensures no reflection at the
line termination, and the line behaves as an infinitely long line.

6.2.2 A line of finite length


Let us take a transmission line of finite length l, terminated at the end by impedance
ZL. The voltage U 2 = U (l ) and current I 2 = I (l ) at the end are known. Their ratio is

U2
ZL = . (6.29)
I2

To simplify this problem we use a new coordinate s=l-z l-z 0


s
measured from the line end, Fig. 6.3. As ds=-dz, we can
rewrite equations (6.12) and (6.13) to the form
0 z l z
∂ 2U (s ) U2, I2
2
− γ 2U (s ) = 0 , (6.30)
∂s Fig. 6.3

∂ 2 I (s )
2
− γ 2 I (s ) = 0 . (6.31)
∂s

The solution of these equations is

U = U + e γs + U − e −γs , (6.32)

I=
1
ZC
(
U + e γs − U − e −γs . ) (6.33)

For z=l we now have s=0 and

1
U2 = U+ +U− , I2 = (U + − U − ) .
ZC

From these formulas we get the amplitudes of the wave traveling to the right and the wave
traveling to the left as functions of the voltage and the current at the line end

1
U+ = (U 2 + Z C I 2 ) , (6.34)
2

1
U− = (U 2 − Z C I 2 ) . (6.35)
2

72
book - 6
Inserting (6.34) and (6.35) into (6.32) and (6.33) we get the voltage and the current at any
point on the transmission line

U (s ) = U 2 cosh (γs ) + Z C I 2 sinh (γs ) , (6.36)

U2
I (s ) = I 2 cosh (γs ) + sinh (γs ) . (6.37)
ZC

The impedance at any point along the transmission line is

U U 2 cosh (γ s ) + Z C I 2 sinh (γ s ) Z + Z C tgh (γ s )


Z= = = ZC L , (6.38)
I U2 Z C + Z L tgh (γ s )
I 2 cosh (γ s ) + sinh (γ s )
ZC

Formula (6.38) is the basic formula used in the analysis and design of high frequency circuits.
It determines how the terminating impedance is transformed to any point along the line.
The propagation constant of a lossless line is determined by (6.16), which is a purely
imaginary number. The characteristic impedance is a real number (6.20). Equations (6.36) and
(6.37) can now read (13.35) and (13.36)

U (s ) = U 2 cos(β s ) + jZ C I 2 sin (β s ) , (6.39)

U2
I (s ) = I 2 cos(β s ) + j sin (β s ) . (6.40)
ZC

The impedance at any point along the line is

U Z + jZ C tg (β s )
Z= = ZC L . (6.41)
I Z C + jZ L tg (β s )

The two formulas (6.38) and (6.39) confirm the fact stated at the end of the preceding
paragraph. Terminating the line by an impedance equal to characteristic impedance ZC, we get
the impedance at any point equal to the characteristic impedance. The line behaves as an
infinitely long line.

6.2.3 A line terminated by a short cut or by an open end


A line terminated by a short cut or by an open end is exposed to total reflection, and
thus a reflected wave has the same amplitude as the wave incident to the termination. As a
result we get a standing wave. This effect is analogous to the perpendicular incidence of a
plane electromagnetic wave to the surface of a well conducting material or a material with
infinite permeability – a perfect magnetic material.
An open end termination represents an infinite impedance, and the current passing
through this impedance I 2 = 0 . Formulas (6.39) and (6.40) are now, using β = 2π / λ (2.15),

 2π 
U = U 2 cos(β s ) = U 2 cos s , (6.42)
 λ 

73
book - 6
U2 U  2π 
I= j sin (β s ) = j 2 sin  s . (6.43)
ZC ZC  λ 

These voltage and current u (s )


distributions are plotted in Fig.
6.4. Formulas (6.42) and (6.43)
describe a standing wave on a
transmission line and are
analogous to formulas describing
the standing wave after the
perpendicular incidence of a s
plane electromagnetic wave to the λ 3 λ/4 λ/2 λ/4
surface of a perfect magnetic i (s )
material. The active power
transmitted by this wave is zero,
which gives the wave its name.
Using a probe sliding along the
line we can detect the distribution
of the voltage determined by
(6.42) with stable maxima and s λ 3 λ/4 λ/2 λ/4
minima. The distance between the
two adjacent maxima or minima Fig. 6.4
is equal to a half wavelength. In
this way we can measure the
wavelength.
The impedance at any point on
the line is
X(s)
U  s
Z= = − jZ C cot g 2π  .
I  λ
(6.44)

A lossless line terminated by an open 3λ/4 λ/4


end behaves as a reactance with a
λ λ/2
value from minus to plus infinity.
This reactance X(s) is plotted in Fig.
6.5. Fig. 6.5 shows the character of
the impedance. It changes from the
capacitive character to an impedance
of a series resonance circuit, to an
inductive character, and the
impedance of a parallel resonant
circuit. Very close to the line end, Fig. 6.5
where s<<λ, the impedance is

L 1 L 1 1
Z ≈−j =−j = , (6.45)
C βs C ω LC s jωCs

74
book - 6
and the line behaves here as a capacitor with a capacity proportional to the line capacity per
unit length.
The short cut termination represents zero impedance, and the voltage at the line end
is equal to zero. Formulas (6.39) and (6.40) are now

 s
U = j Z C I 2 sin (β s ) = j Z C I 2 sin  2π  (6.46)
 λ

 s u (s )
I = I 2 cos(β s ) = I 2 cos 2π  .
 λ
(6.47)

These voltage and current


distributions are plotted in Fig.
6.6. Formulas (6.46) and (6.47) s
describe a standing wave which λ 3 λ/4 λ/4
does not transport any active i (s )
power. The reader can compare
these formulas with (3.13) and
(3.14), which describe the
distribution of a standing wave
created by the perpendicular
incidence of a plane
electromagnetic wave on the s λ 3 λ/4 λ/2 λ/4
surface of a perfect conductor.
The impedance at any
point on the line is Fig. 6.6

U  s
Z= = jZ C tg 2π  .
I  λ
(6.48)
X(s)
This function is plotted in Fig. 6.7.
Choosing a suitable line length, we
can realize a line stub with an
arbitrary input reactance value. Fig.
6.7 shows the character of this
impedance. Very close to the end of a λ λ/2
line (s<<λ) we have the impedance 3λ/4 λ/4

 s s
Z = jZ C tg 2π  ≈ jZ C 2π =
 λ λ
,
L
= j ω LC s = jωLs
C
(6.49)

and the line terminated by a short cut Fig. 6.7


behaves as an inductor with its
inductivity proportional to the line inductivity per unit length.
75
book - 6
A line stub with a short cut is applied more frequently than a stub with an open end.
The reason is that the former radiates less energy than the latter. The radiation of energy
causes a non zero real part of the impedance, as it represents losses.

6.3 Smith chart


When designing microwave circuits the designer is interested in the values of voltages
U, currents I, impedances Z and reflection coefficient ρ. It would be a tedious work to
calculate these quantities, specially for a lossy transmission line, and transform them along a
transmission line without a computer and modern CAD tools. There are some graphical tools
that can be used to simplify this task. One of them was introduced by Smith in 1939. It is a
chart that enables us to make a simple transformation of the impedance along a transmission
line and to recalculate the impedance to a reflection coefficient, and vice versa. It is a very
useful tool for designing microwave circuits. It is even used in acoustics. As we work with
values of voltage and current, this tool is applicable in the case of transmission lines with a
TEM wave, where they are uniquely defined. The Smith chart can even be applied in the case
of lines with no TEM wave, e.g., waveguides. Here we can use an appropriate scaling of
propagating waves to some hypothetic TEM waves described by voltages and currents.
Let us take a TEM transmission line, Fig. 6.8, fed from a generator and terminated by
a load, with length l
d
and characteristic
impedance ZC. The generator load
ZC
position along this
line is determined by z
the z coordinate z1
measured from the
z2
terminals of the
generator. The l
propagation constant
is assumed in the Fig. 6.8
notation used in the
theory of electric circuits (6.15)

γ = jk = α + jβ , (6.50)

where α is the attenuation constant and β is the phase constant. At point z we have an
impedance Z = U(z)/I(z), this impedance causes a reflection with a reflection coefficient

Z (z ) − Z C
ρ (z ) = . (6.51)
Z (z ) + Z C

The reflection coefficient is determined in the same way as the reflection coefficient for the
TEM wave incident perpendicular to the boundary between two materials (3.6). From (6.51)
we get

1 + ρ (z )
Z (z ) = Z C . (6.52)
1 − ρ (z )

The standing wave ratio is defined in the same way as (3.20)

76
book - 61
1 + ρ (z )
p( z ) = . (6.53)
1 − ρ (z )

The transformation of the impedance along the transmission line is described by


(6.38). Using (6.38) we can recalculate the impedance from point z1 to z2 and back

U ( z1 ) Z ( z 2 ) + Z C tgh (γ d )
Z ( z1 ) = = ZC , (6.54)
I (z1 ) Z C + Z ( z 2 ) tgh (γ d )

U (z 2 ) Z ( z1 ) − Z C tgh (γ d )
Z (z 2 ) = = ZC , (6.55)
I (z 2 ) Z C − Z ( z1 ) tgh (γd )

where d = z2 – z1, Fig. 6.8. To define the reflection coefficient we have to decompose the
voltage at any point to a wave propagating to the right, i. e., in the positive z direction, and a
wave propagating to the left

U + ( z ) = U 0 e −γz , U − ( z ) = U 0 e γz .

The transformation of the voltage along the line is

U + ( z 2 ) = U + ( z1 ) e −γ ( z2 − z1 ) , U − ( z 2 ) = U − ( z1 ) e γ ( z2 − z1 ) , (6.56)

From (6.56) we have the transformation of the reflection coefficient

U − (z 2 )
ρ (z 2 ) = = ρ ( z1 ) e 2γd , (6.57)
U +
(z 2 )

U − ( z1 )
ρ ( z1 ) = = ρ ( z 2 ) e − 2γd . (6.58)
U +
(z1 )
To get a universal tool for analyzing of microwave circuits we use the normalized
impedances

Z
z= . (6.59)
ZC

Consequently (6.51), (6.52) and (6.54) have the form

z −1
ρ (z ) = , (6.60)
z +1

1+ ρ
z= , (6.61)
1− ρ

77
book - 61
z ( z 2 ) + tgh (γ d )
z ( z1 ) = . (6.62)
1 + z (z 2 ) tgh (γ d )

Now we have everything necessary for deriving the Smith chart. First we show the
way to make a graphic representation in a complex plane of the reflection coefficient and its
transformation. The reflection
coefficient is generally a complex jv ρ
jv 90°
number with an amplitude lower |ρ| |ρ|=1
0.7
than 1. It is plotted into a complex
ϕ 0.2
plane u, jv, Fig. 6.9. The lines of 180° ϕ=0
constant amplitude are circles with 0
u -180° u
their center at the origin, and the
lines of constant phase are radial
lines, Fig. 6.9. The reflection
coefficient is transformed -90°
Fig. 6.9
according to (6.58). For a lossless
line it is γ = jβ. The transformation of the reflection coefficient is controlled by

ρ ( z1 ) = ρ ( z 2 ) e − j 2 βd . (6.63)

Now we have to distinguish between the two shift toward jv 0.125


βλ=const
directions of the shift along a line. In the case of d > the generator
0 it is z2 > z1, z1 is closer to the generator and the
shift is toward the generator. As d > 0 the phase 0 0.25
in (6.63) decreases and the shift toward the 0.5 0 u
generator corresponds to the rotation of the
reflection coefficient in the complex plane to the
right, Fig. 6.10. In the case of d < 0 it is z2 < z1, z1 is
closer to the load and the shift is toward the load. shift toward 0.375
As d < 0 the phase in (6.63) increases and the shift the load
toward the load corresponds to the rotation of the Fig. 6.10
reflection coefficient in the complex plane to the
left, Fig. 6.10. According to (6.63), the measure of the shift is equal to the product βd. The
trip around the circumference of the whole complex plane represents angle 2π and, as β =
2π/λ, this corresponds to

2π d d
2π = 2 βd = 2 d = 4π => = 0 .5 .
λ λ λ

The angle in the complex plane is not defined in degrees, but in the relative distance
measured in the number of wavelengths in the range from 0 to 0.5, see Fig. 6.10. We have
to distinguish between the direction toward the load and the direction toward the generator. In
the case of a lossless line the transformation is performed along a circle, as the modulus of the
reflection coefficient stays unchanged. In the case of a lossy line the modulus of the reflection
coefficient decreases exponentially, and the transformation is therefore performed along a
spiral.
The complex normalized impedance (6.59) and its complex conjugate value are

78
book - 61
1+ ρ
z = r + jx = , (6.64)
1− ρ

1+ ρ*
z * = r − jx = . (6.65)
1− ρ*

By adding and subtracting these two equations we get the real r and imaginary x parts of the
normalized impedance

1 1+ ρ 1+ ρ* 
r=
1
2
(
z + z * =  ) +  ,
2  1 − ρ 1 − ρ * 
(6.66)

1 1+ ρ 1+ ρ* 
jx =
1
2
(
z − z * =  ) −  .
2  1 − ρ 1 − ρ * 
(6.67)

From (6.66) and (6.67) we can derive equations which determine images of the lines of the
constant values of the real part of normalized impedance r and of the imaginary part of
normalized impedance x in the complex plane of the reflection coefficient. The goal is to
convert (6.66) and (6.67) into the equation of a general circle in the complex plane ρ = u + jv

(
ρ ρ * − (m − jn ) ρ − (m + jn ) ρ * + m 2 + n 2 − R 2 = 0 , ) (6.68)

where R is the radius of a circle, m and jn are the coordinates of jv


the center of the circle, see Fig. 6.11. After some manipulations, R
(6.66) can be rewritten into
jn
r r r −1
ρ ρ* − ρ− ρ* + =0 . (6.69)
r +1 r +1 r +1
m u
Comparing (6.69) with (6.68) we get the coordinates of the
center of the circle (6.69) and its radius
Fig. 6.11
r 1
m= , n=0, R= . (6.70)
r +1 r +1

Now we are able to jv


transform the lines jx r=0
of the constant
values of the real r=1
part of the r=2
normalized r= ∞
impedance from the u
r
complex plane z = r
+ jx, where these
lines are lines r=0 r=1 r=2
parallel with the
imaginary axis, into
Fig. 6.12
79
book - 61
the complex plane ρ = u + jv. This transformation is shown in Fig. 6.12. The corresponding
quantities are given in Table 6.1.
r m n R The right hand side half of the complex plane z = r + jx is in this
0 0 0 1 way transformed into the inside of the circle in the complex
1 1/2 0 1/2 plane ρ = u + jv with radius R = 1. The same procedure can be
2 2/3 0 1/3 performed with the imaginary part of the normalized impedance.
∞ 1 0 0 Equation (6.67) can be rewritten into

Table 6.1  1  1
ρ ρ * − 1 − j  ρ − 1 + j  ρ * + 1 = 0 . (6.71)
 x  x

Comparing (6.71) with (6.68) we get the coordinates of the center of the circle (6.71) and its
radius
1 1
m=1, n= , R= . (6.72)
x x

Now we are able to transform the lines of the constant values of the imaginary part of the
normalized impedance from the complex plane z = r + jx, where these lines are parallel with
the real axis, into the complex plane ρ = u + jv. This transformation is shown in Fig. 6.13. The
corresponding quantities are given in Table 6.2.
jx
x=2 x=1

x=1 jv

x=2
x=0
r

x= -1
x= ∞ x=0
x= -2 x= - ∞ u

x m n R
∞ 1 0 0
2 1 1/2 1/2 x= -2
1 1 1 1
0 1 ∞ ∞
-1 1 -1 1
-2 1 -1/2 1/2 x= -1
-∞ 1 0 0
Fig. 6.13 1
Table 6.2 The Smith chart is obtained by combining the chart from
Fig. 6.12, representing the lines of constant r, and the chart from
Fig. 6.13, representing the lines of constant x. We have to take into account that the real part
of the impedance must be positive. This means that the lines of constant x are limited to the
right half of the complex plane z = r + jx, and at the same time to the inside of the circle with
80
book - 61
the unit radius in the plane ρ = u + jv. The final version of the Smith chart is shown in Fig.
6.14.
Each point in the Smith chart corresponds to an impedance and at the same time to the
reflection coefficient (6.60). The scale of the modulus of the reflection coefficient is shown
below the chart. The corresponding angle must be measured using a protractor, or the scale of
the angles is shown in some versions of the chart along its perimeter. The scale of the
standing wave ratio (6.53), shown in some versions of the chart, is a nonlinear rating from 1
to ∞ . The transformation of the reflection coefficient is performed as explained in the text
referring to Fig. 6.10. The scale of the relative distance is shown along the perimeter of the
chart, its range being between 0 and 0.5. There are the two scales of the length. The first
corresponds to the orientation toward the generator, the second toward the load. The scale of
the real part of the normalized impedance is on the horizontal axis of the chart. The scale of
the imaginary part of the normalized impedance is shown along the perimeter of the chart.

z = 0.4+j0.7

|ρ| =0.588

Fig. 6.14

In the following examples we show some basic operations with the Smith chart. To
show the advantages of applying the Smith chart we choose cases where analytical calculation
81
book - 61
can be used.
_____________________________________
Example 6.3: Show the normalized impedance z = 0.4+j07 in the Smith chart.
This impedance is shown in Fig. 6.14 together with the corresponding value of the
reflection coefficient

z −1
ρ (z ) = = −0.143 + j 0.57 = 0.588 e j104° .
z +1
_____________________________________
Example 6.4: The lossless transmission line is terminated by a short. Calculate the
normalized impedance at distance l = 3λ/8 from the short.
As z(z2) = 0, we can apply (6.48) and we get

 2π 3  3 
z ( z1 ) = j tg (βl ) = j tg λ  = j tg π  = − j .
 λ 8  4 

Fig. 6.15

In the Smith chart we use the following technique, Fig. 6.15. The short is a zero
impedance. From this point we move in the direction toward the generator along the circle of
a constant reflection coefficient by the normalized length l/λ = 3/8 = 0.375. At this distance
we read the normalized impedance –j.
_____________________________________

82
book - 61
Example 6.5: The lossless transmission line is terminated by an open end termination.
Calculate the normalized impedance at distance l = λ/4 from the short.
The normalized impedance at any point along the open end terminated transmission
line is given by (6.44), so we get

π 
z ( z1 ) = − j cotg (βl ) = − j cotg  = 0 .
2

The open end transforms as a short. In the Smith chart we perform the transformation
according to Fig. 6.16. The open end termination represents the infinite normalized
impedance. From this point we move in the direction toward the generator along the circle of
a constant reflection coefficient by the normalized length l/λ = 1/8 = 0.25. At this distance we
read the normalized impedance 0.

Fig. 6.16
_____________________________________
Example 6.6: The lossless transmission line is terminated by the normalized impedance zL =
0.8+j. Calculate the normalized impedance at distance l = 0.2λ from the end of the line.
Using (6.62) applied to the lossless line and the product βl we get


βl= 0 .2 λ = 0 .4 π
λ

z L + j tg (β l ) 0.8 + j + j tg (0.4 π )
z ( z1 ) = = = 0.807 − j 1.0061
1 + j z L tg (β l ) 1 + (0.8 + j ) j tg (0.4 π )

83
book - 61
In the Smith chart, Fig. 6.17, we first determine the point corresponding to the loading
impedance zL = (0.8+j). We draw a radial line of the constant value of βl through this point
and read the value of the relative distance 0.15 on the perimeter of the chart. From this point
we move toward the generator by the relative distance 0.2, i.e., to the point 0.15+0.2 = 0.35.
Here we read on the circle representing the same value of the modulus of the reflection
coefficient as at point zL the normalized impedance z(z1) = 0.8 – j, which is approximately the
same value as calculated.

l/λ=0.2

0.8+j

0.8-j

Fig. 6.17
___________________________________
Example 6.7: The lossy transmission line with attenuation α’ = 0.5 dB is terminated by
normalized impedance zL = 0.8+j2.2. Calculate the normalized impedance and the standing
wave ratio at distance l = 20 m from the end of the line. The wavelength is 0.44 m.
From the attenuation in dB we determine attenuation constant α and from the
wavelength phase constant β

( )
α ' = 20 log e −α = −α 20 log(e ) => α = −
α'
20 log(e )
= 0.05756 m-1, β=

λ
= 14.27 m-1

Using (6.62) we get

z L + tgh (γ l )
z ( z1 ) = = 1.02 + j 0.156
1 + z L tgh (γ l )

84
book - 611
In the Smith chart, Fig. 6.18, we first determine the point corresponding to the loading
impedance zL = (0.8+j2.2). We draw the radial line of the constant value of βl through this
point and read the value of the relative distance 0.188 on the perimeter of the chart. From this
point we move toward the generator by the relative distance l/λ = 20/λ = 45.454 so we get to
the point 0.188+45.454 = 45.642. This means that we go round the chart 91 times and stop at
the point 0.142. This is the position of the impedance we are looking for. Now we have to
determine the value of the reflection coefficient. The reflection coefficient transforms
according to (6.58) so we get

ρ1 = ρ 2 e −2αl = ρ 2 0.1

In the Smith chart in Fig. 6.18 we read the normalized impedance z(z1) = 1.02+j0.15, and
using (6.60) and (6.53) we get the standing wave ratio p(z1) = 1.16.
0.142

0.188

zL=0.8+j2.2

|ρ2|
|ρ1|

zL=1.02+j0.15

Fig. 6.18
_______________________________________

The normalized impedance could be calculated simply by using transformation


formulas (644), (6.48) and (6.62) in examples 6.4 to 6.7. The Smith chart however simplifies
the work. In example 6.8 we perform impedance matching, i.e., we make corrections in a
circuit to reduce the reflection at the given frequency to zero. This cannot be done simply, as
it includes the need to solve a transcendent equation. The Smith chart is a very efficient tool
in such a case.
___________________________________

85
book - 611
Example 6.8: The lossless transmission line with
characteristic impedance ZC = 300 Ω is terminated by l2 ZC
impedance ZL = (420-j180) Ω. Match this impedance by
the stub of the same line terminated by a short end
connected in series to the line so as to get a zero ZC ZL
reflection coefficient, Fig. 6.19. The wavelength is 5 m.
First we must determine distance l1 at which we
l1
connect the stub. The loading impedance normalized to
the characteristic impedance zL = ZL/ZC = 1.4-j0.6 must
be transformed within distance l1 to impedance z1 = Fig. 6.19
1+jx. Here we connect the stub with input impedance z2
= 0-jx. The resulting impedance at this point is then z = z1+z2 = 1, and therefore the reflection
coefficient is zero.

l2/λ=0.086
0+j0.6

zL
z1

l1/λ=0.043
Fig. 6.20

The solution of this problem in the Smith chart is simple, see Fig. 6.20. Impedance zL
is transformed along the circle of the constant value of the reflection coefficient to the point at
which the impedance lies on the circle where the real part of the normalized impedance is 1.
This gives z1 = 1-j0.6, Fig. 6.20. Corresponding length l1/λ = 0.043. Using the given value of
λ we have l1 = 0.215 m.
Let us now form the stub. At its input it must have the normalized impedance equal to
0+j0.6. This stub is terminated by a short that is zero impedance. The length of this stub l2 is
determined by transforming zero impedance to the point corresponding to the impedance
0+j0.6. From the Smith chart in Fig. 6.20 it follows that the necessary length is l2/λ = 0.086,

86
book - 611
which gives l2 = 0.43 m. The second solution can be obtained by transforming zL to the point
with impedance z1 = 1+j0.6. Then the stub must have input impedance 0-j0.6. This solution is
not optimal as the corresponding lengths l1 and l2 are longer than those obtained above.
_______________________________________

6.4 Problems
6.1 A longitudinally homogeneous transmission line transmits a TEM wave with frequency
f=1 kHz. The amplitude of the voltage on the input port is Um=60 V. The end terminals of the
line are connected to an impedance equal to the line characteristic impedance. The line
parameters are R=4 Ω/km, G=0.5 µS/km, L=2 mH/km, C=6000 pF/km. Calculate the phasor
of the voltage on the line end.
U2=-24.75+j33.93

6.2 A transmission line is 50 km long, and its parameters are R=6.46 Ω/km, G=1 µS/km, L=4
mH/km, C=12.2 nF/km. The line is loaded by a resistor R L=1000 Ω. The voltage on the load
is U2=10 V. The frequency is 0.8 kHz. Calculate the voltage and the current on the input of
the line.
U1 = 9.033 e j1.748 = −1.5921 + j8.892 V
I1 = 20.77 e j1.848 = −569 + j 20 mA

6.3 Calculate the input impedance of a lossless transmission line of the length l=0.4 m. The
characteristic impedance is 75 Ω, and the frequency is 0.6 GHz. The line has a shortcut at the
end.
Z1=230.8 Ω, L1=0.12 µH

6.4 Calculate the input impedance of a lossless transmission line of length l=0.4 m. The
characteristic impedance is 75 Ω, and the frequency is 0.6 GHz. The line has an open end.
Z1=-24.4 Ω, C1=21.76 pF

6.5 A lossless very short transmission line (l=125 mm) has the input impedance at frequency
f=60 MHz Z1S=j8850 Ω when terminated by a short circuit, and Z1O=-j35.5 Ω when terminated
by an open circuit. Calculate the inductivity per unit length and the capacity per unit length.
L=75 µH/m, C=0.56 pF/m

6.6 Match the transmission line with characteristic impedance


50 Ω terminated by a series combination of the resistor with
R
resistivity R = 100 Ω and a capacitor with capacity C = 1.5 pF, ZC λ
Fig. 6.21, at frequency 5 GHz, which corresponds to the
C
wavelength on the line λ = 42.5 mm.

The short circuited stub of Fig. 6.21


the same transmission line of
length l2 = 4.6 mm must be
connected in series at distance l1 = 3 mm from the load,
Fig.6.19.

87
book - 611
7. WAVEGUIDES WITH METALLIC WALLS
In this section we will study transmission lines with metallic walls. Such transmission
lines are known as waveguides. The field distribution is determined by solving the wave
equation for the longitudinal components of the electric and magnetic fields, as described in
section 5. Our task is to solve equation (5.3) with the proper boundary conditions. The
tangential component of the electric field has to be equal to zero on the metallic walls. This
technique will be directly applied to a parallel plate waveguide and to a waveguide with the
rectangular cross-section. The field distribution is described in a rectangular coordinate
system in the case of these two lines. In the case of a waveguide with a circular cross-section
we have to use a circular coordinate system. The parallel plate waveguide consists of the two
separate conducting plates, so it can conduct DC current and it can guide the TEM mode.
Together with this TEM mode we will describe the modes known as the waveguide modes.
These modes have nonzero cut-off frequency. Waveguides with rectangular and circular
cross-sections consist of only one conductor, so they are not able to conduct the TEM mode.

7.1 Parallel plate waveguide


The cross-section of the parallel plate x
waveguide is shown in Fig. 7.1. It is created by σ =∞
the two parallel infinitely wide conducting plates
which are located at a distance a. The space ε, µ
between these plates is filled by an ideal
dielectric material. a
We will describe separately the field of
the TE and TM modes. The calculation
procedure is simplified by the fact that the field
does not depend on the y coordinate. So we put
σ =∞ y z
∂ ∂y = 0 in (5.14) to (5.17). Fig. 7.1
The TE modes thus have Ey, Hx and Hz
components. Setting Ez=0 in (5.14) to (5.17) we get

jω µ ∂H z
Ey = , (7.1)
k 2p ∂x

k z ∂H z
Hx = − j . (7.2)
k p2 ∂x

Function H0z representing the transversal field distribution analogous to (5.1) depends only on
the x coordinate, so we can put

k p = kx , (7.3)

and the H0z component is determined by solving the wave equation

dH 0 z
+ k x2 H 0 z = 0 . (7.4)
dx

88
book - 7
The solution is in the form

H 0 z = A sin (k x x ) + B cos(k x x ) .

The boundary conditions cannot be fulfilled by this component. We have to use E0y, which is
tangent to both conducting plates. At x=0 we have E0y(0)=0, Fig. 7.1, and, consequently, A=0.
The second boundary is at x=a, where E0y(a)=0 and sin(kxa)=0 and we get the propagation
constant in the x direction kx


k xm = . (7.5)
a

Consequently, the solution of the wave equation (7.4) is

 mπ 
H 0 z = B cos x . (7.6)
 a 

(7.1) and (7.2) now have the form

jω µ  mπ 
E0 y = − B sin  x , (7.7)
k xm  a 

k zm  mπ 
H 0x = j B sin  x . (7.8)
k xm  a 

To remove kxm from the denominators of (7.7) and (7.8) we introduce a new constant
jω µ
B' = − and we get the field distribution of the TE modes
kx

 mπ  − jk zm z
E y = B' sin  x e , (7.9)
 a 

k zm  mπ  − jk zm z
Hx = B' sin  x e , (7.10)
ωµ  a 

k xm  mπ  − jk zm z
Hz = j B' cos xe . (7.11)
ωµ  a 

The TM modes have nonzero field components Ex, Ez and Hy. The transversal
components are from (5.14) to (5.17)

k zm ∂E z
Ex = − j , (7.12)
k p2 ∂x

ω ε ∂E z
Hy = −j . (7.13)
k p2 ∂x

89
book - 7
The equation for determining the transversal distribution of the longitudinal component of the
electric field is

dE 0 z
+ k x2 E 0 z = 0 . (7.14)
dx

The boundary conditions are met directly by Ez, which is tangent to both plates. Applying a
similar procedure as for the TE modes we get the field distribution

 mπ 
H y = C ' cos x  e − jk zm z , (7.15)
 a 

k zm  mπ 
Ex = C ' cos x  e − jk zm z , (7.16)
ωε  a 

jk xm  mπ  − jk zm z
Ez = C ' sin  x e . (7.17)
ωε  a 

The propagation constant in the x-direction has the form (7.5) as in the case of the TE modes,
and (7.3) is again valid.
Modal number m determines the x
form of the field distribution in the
waveguide cross section. This is shown in m=0 m=1 m=2 m=3
Fig. 7.2. m determines how many half-
periods of the sinus function across the a
space between the plates the distribution
has. Fig. 7.2 shows the distribution of Ez for
the TM modes, or Ey for the TE modes.
The longitudinal propagation Ez, Ey
constant kz is given by (5.18). Using (7.3)
and (7.5) we have for the m-th mode Fig. 7.2

2
 mπ 
k zm = ω 2 µε −   . (7.18)
 a 

Now we have from (5.19) the cut-off frequency and the cut-off wavelength of both the TE and
TM modes with modal number m in the parallel plate waveguide

m
f cm = , (7.19)
2a µε

2a
λcm = . (7.20)
m

The parallel plate waveguide behaves like the high-pass filter, as the modes can propagate
starting from their cut-off frequency.
From the longitudinal propagation constant we can determine the parameters
describing the propagation of the modes along the parallel plate waveguide. These are the
90
book - 7
wavelength, the phase velocity, and the group velocity. The phase velocity of the m-th mode
is
ω ω 1
v pm = = = =
k zm  m π 
2
 mπ 
2
ω 2 µε −   µε 1 −  
 2  
 ω µε a 
. (7.21)
v v
= =
2 2
 f cm   λ 
1 −   1 −  
 f  λ
 cm 

Similarly we can get formulas for the group velocity, the wavelength and the longitudinal
propagation constant

v pm λ λ
λ gm = = = , (7.22)
f f 
2
 λ 
2
1 −  cm  1 −  
 f  λ
 cm 

2 2
dω f   λ 
v gm = = v 1 −  cm  = v 1 −   , (7.23)
dk zm  f   λcm 

2 2
f   λ 
k zm = k 1 −  cm  = k 1 −   . (7.24)
 f   λcm 

The plots of these functions vpm/v


are shown in Fig. 7.3. It is m=1 m=2
seen from Fig. 7.3 that as λ gm/λ
the frequency approaches
from above the cut-off
frequency, the 1
corresponding mode stops
propagating as its vgm/v
propagation constant and kzm/k
group velocity tend to zero. m=1 m=2
At the same time its phase
velocity and wavelength fc1 fc2 fc3
tend to infinity. Below the
cut-off frequency these
quantities are imaginary Fig. 7.3
numbers and the mode is
the evanescent mode, the amplitude of which decreases exponentially along the line, and this
mode does not propagate.
The field distributions of the TE and TM modes are different, so their wave
impedances are different. They are defined as the ratio of the transversal components of the
electric and magnetic fields. Using (7.9) and (7.10) we get for the TE modes

91
book - 7
Ey ωµ Z0
Z mTE = − = = . (7.25)
Hx k zm f 
2
1 −  cm 
 f 

For the TM modes we get from (7.15) and (7.16)

2
E k f 
Z mTM = x = zm = Z 0 1 −  cm  . (7.26)
H y ωµ  f 

Z0 is the wave impedance of the free space.


Let us now look for the dominant mode of the parallel plate waveguide. This is the
mode with the lowest possible modal number m. It follows from the distribution of the
electromagnetic field of the TE modes, (7.9), (7.10), (7.11) that these modes cannot have m =
0, as this case gives a zero field. Consequently, the lowest TE mode is the mode TE1. For the
TM modes we can allow m=0, as from (7.15-17) we get

H y = C ' e − jkz , (7.27)

k
Ex = C ' e − jkz , (7.28)
ωε

Ez = 0 . (7.29)

This TM0 mode has only the transversal components of the electric and magnetic fields.
Therefore it is the TEM wave. Its cut-off frequency is zero and this mode can propagate from
zero frequency and kz0 = k. This confirms the fact that the parallel plate waveguide consists of
two separate conductors and is therefore able to conduct the DC current. So the dominant
mode of the parallel plate waveguide is the TEM mode, which can propagate from zero
frequency.
The field distribution of the two lowest TE modes on the parallel plate waveguide is
shown in Fig. 7.4. Fig. 7.5 shows the field distribution of the three lowest TM modes.

TE1 TE2
Fig. 7.4

92
book - 7
E

TM0 = TEM

TM1 TM2
Fig. 7.5

The explanation of wave propagation based on geometrical optics is shown in Fig. 7.6.
The wave is coupled between the plates under angle Θ and bounces up and down due to
reflections on the conducting
plates. The wavelength in the
direction along the line is the
Θ λ
projection of the free space
wavelength, so it must be longer.
We have
λ
λg = , λg
cos Θ
Fig. 7.6
consequently

2
f 
cos Θ = 1 −  cm  . (7.30)
 f 

This angle of establishment determines the conditions of the wave propagation. It corresponds
to the relative frequency distance of the wave from the cut-off frequency. The lower this angle
is, the further the wave is from the cut-off. Θ = 90°corresponds to the cut-off frequency and it
is obvious that the wave does not propagate.

__________________________
Example 7.1: Determine the modes which can propagate in the parallel plate waveguide at
the frequency 10 GHz. The distance between the plates is 40 mm and the space is filled by air.
The mode can propagate if the wavelength at 10 GHz is lower than the cut-off
wavelength of this mode. The wavelength at 10 GHz is

c
λ= = 30 mm
f

93
book - 7
The cut-off wavelength of the m-th mode is determined by (7.20). So we have

m=0 λc0 = ∞ the mode can propagate


m=1 λc1 = 80 mm the mode can propagate
m=2 λc2 = 40 mm the mode can propagate
m=3 λc1 = 26.7 mm the mode cannot propagate
The parallel plate waveguide can transmit the TM0, TM1, TE1, TM2 and TE2 modes at the
frequency 10 GHz.
________________________________

Example 7.2: Determine the distance between the two conducting planes to get the parallel
plate waveguide along which the only TE1 mode propagates at 10 GHz and the mode TE2 is
attenuated by –120 dB/m. Do not consider the TM modes.
The attenuation of the TE2 mode is

 E 0 e − jk z 2d 
20 log  = −120 dB ,

d=1m
 E0 

e − jk z 2d = 10 −6 , -jkz2d = -13.8 , kz2 = -j13.8

As kz2 is an imaginary number the TE2 mode does not propagate and is an evanescent mode.
From the determined value of kz2 we will calculate the transversal propagation constant ky2

mπ 2π
= k 2 − k z22 = ω 2 µε − (− j13.8) = 210
2
k y2 = = ⇒ a = 29 mm
a a

The cut-off wavelength of the TE1 mode is

λc1 = 2a = 58 mm .

The TE1 mode can propagate.


___________________________________

7.2 Waveguide with a rectangular cross-section


The parallel plate waveguide is not suitable for practical applications, as it is open
from the sides and it can radiate energy to the
y
sides. To get a practical line we have to
confine it by two conducting planes from
sides. Thus we get a waveguide of rectangular b
cross-section. This waveguide is completely
shielded. It is shown in Fig. 7.7. The inner
dimensions are a in the direction of the x axis,
and b in the direction of the y axis. We will
assume an infinitely long waveguide with
0 a x
ideally conducting walls filled by a lossless
dielectric.
The process of solving the wave z
equation is shown in chapter 5. In the case of Fig. 7.7

94
book - 7
the TM modes we solve the wave equation (5.3) for E0z and the transversal components of the
electric and magnetic fields are calculated from (5.14-17) putting H0z = 0.
For the TE modes we determine H0z from the equation analogous to (5.3) and put E0z
= 0. The longitudinal component of the magnetic field is

( )(
H 0 z = A e jk x x + B e − jk x x C e
jk y y
+ De
− jk y y
), (7.31)

where the separation constants, which represent the propagation constants in the x and y
directions, are coupled with kp by

k 2p = k x2 + k y2 . (7.32)

Unknown constants A, B, C, D and the propagation constants must be determined applying


the boundary conditions. To meet these boundary conditions we have to know the transversal
components of the electric field. From (5.14) and (5.15) we get

E0 x = −
jωµ ∂H 0 z ω µ k y
2
k p ∂y
= 2
kp
( jk y
A e jk x x + B e − jk x x C e y − D e y )(
− jk y
)

E0 y =
jωµ ∂H 0 z
2
k p ∂x
=−
ω µ kx
2
kp
( jk y
A e jk x x − B e − jk x x C e y + D e y .)(
− jk y
)
The boundary conditions are

y = 0, Ex = 0 => C–D=0 => Ce ( jk y y


−e
− jk y y
) = 2 jCsin (k yy ),
x = 0, Ey = 0 => A–B=0 => ( )
A e jk x x − e − jk x x = 2 jAsin (k x x ) ,


x = a, Ey = 0 => sin (k x a ) = 0 => kx = , (7.33)
a


y = b, Ex = 0 => ( )
sin k y b = 0 => ky =
b
, (7.34)

Now setting a new constant M = 4AC, which determines the field amplitude, we have the field
distribution

 mπ   nπ  − jk zmn z
H z = M cos x  cos ye , (7.35)
 a   b 

jk zmn mπ  mπ   nπ  − jk zmn z
Hx = 2
M sin  x  cos ye , (7.36)
kp a  a   b 

jk zmn nπ  mπ   nπ  − jk zmn z
Hy = − 2
M cos x  sin  ye , (7.37)
kp b  a   b 

95
book - 7
jωµ nπ  mπ   nπ  − jk zmn z
Ex = M cos x  sin  ye , (7.38)
k 2p b  a   b 

jωµ mπ  mπ   nπ  − jk zmn z
Ey = − 2
M sin  x  cos ye , (7.39)
kp a  a   b 

where

2 2
 mπ   nπ 
k 2p =  +  , (7.40)
 a   b 

A similar procedure can be followed in the case of TM modes. Now we solve the
wave equation (5.3). Its solution is

( )(
E 0 z = A e jk x x + B e − jk x x C e
jk y y
+ De
− jk y y
). (7.41)

The longitudinal component of the electric field is tangent to all walls of the waveguide. This
simplifies the solution. The boundary conditions are

x = 0, Ez = 0 => A+B=0
=> ( )
E 0 z = M sin (k x x ) sin k y y
y = 0, Ez = 0 => C+D=0


x = a, Ez = 0 => sin (k x a ) = 0 => kx = , (7.42)
a


y = b, Ez = 0 => ( )
sin k y b = 0 => ky =
b
. (7.43)

The particular components of the electric and magnetic field are derived from (7.41) using
(5.14-7)

 mπ   nπ  − jk zmn z
E z = M sin  x  sin  ye , (7.44)
 a   b 

jk zmn mπ  mπ   nπ  − jk zmn z
Ex = − 2
M cos x  sin  ye , (7.45)
kp a  a   b 

jk zmn nπ  mπ   nπ  − jk zmn z
Ey = − 2
M sin  x  cos ye , (7.46)
kp b  a   b 

jωε nπ  mπ   nπ  − jk zmn z
Hx = 2
M sin  x  cos ye . (7.47)
kp b  a   b 

96
book - 7
jωε mπ  mπ   nπ  − jk zmn z
Hy = − M cos x  sin  ye . (7.48)
k p2 a  a   b 

kp is determined by (7.40) and kzmn by (5.18).


Phase constants kp, kx, ky, kz are equal for the TE and TM modes. This means that all
parameters describing the propagation of these modes along the waveguide with a rectangular
cross-section are equal for the TE and TM modes. Each mode is described by the two modal
numbers m and n. They determine the field distribution in the corresponding direction, as
shown in Fig. 7.2. The longitudinal propagation constant is, as follows from (5.18),

2 2
2  mπ   nπ 
k z = ω µε − k 2p 2
= ω µε −   −  . (7.49)
 a   b 

From the condition of zero value of kz we get the cut-off frequency (5.19) of the mode with
modal numbers m and n

2 2
1  mπ   nπ 
f cmn =   +  . (7.50)
2π µε  a   b 

The cut-off wavelength is

v 2π
λcmn = = . (7.51)
f cmn 2
 mπ   nπ 
2
  + 
 a   b 
The reader can now compare relations (7.49), (7.50) and (7.51) valid for a waveguide
with a rectangular cross-section with those valid for a parallel plate waveguide (7.18), (7.19)
and (7.20). Instead of mπ/a we now have (7.40). It follows that the same relations can be
derived for the wavelength along the waveguide, the phase velocity, the group velocity and
the longitudinal propagation constant as (7.21), (7.22), (7.23) and (7.24). The wave
impedance must be defined separately for the TE and TM modes, as their field distribution is
different. It is defined as the ratio of the transversal components

Ex Ey
Z TE ,TM = =− .
Hy Hx

These quantities are described as in the parallel plate waveguide by (7.25) and (7.26).

____________________________
Example 7.3: Determine the modes which can propagate along a rectangular waveguide with
the internal dimensions a = 40 mm, b = 25 mm at the frequency f = 10 GHz.
The propagating modes must have f < fcmn, which corresponds to the condition k2 > kp2.
This gives us

2 2
2  mπ   nπ 
ω µε > k 2p =  +  ,
 a   b 

97
book - 7
and results in the equation

43.681 > m 2 ⋅ 6.162 + n 2 ⋅ 15.775

This inequality is fulfilled for the modes: TE10, TE01, TE20, TE11, TM11, TE21, TM21

________________________________
Example 7.4: Calculate the phase and group velocities, the wavelength, the longitudinal
propagation constant and the wave impedance of the TE10 and TM12 modes in a rectangular
waveguide filled by air with the internal dimensions a = 22 mm, b = 10 mm. The frequency is
f = 10 GHz.
First we calculate the cut-off frequencies of these two modes

1
f c10 = = 6.82 GHz ,
2a µε

2 2
1  π   2π 
f c12 =   +   = 30.76 GHz .
2π µε a  b 

The TE10 mode can propagate, but the TM12 mode cannot as its cut-off frequency is greater
than 10 GHz. Therefore the required parameters will be computed only for the TE10 mode.
We will first calculate the term

2
f 
1 −  c10  = 0.73
 f 

Consequently we have

c
v p10 = = 4.1 ⋅ 10 8 m/s
0.73

v g10 = c ⋅ 0.73 = 2.19 ⋅ 108 m/s

λ
λ g10 = = 41 mm
0.73

k z10 = ω µε ⋅ 0.73 = 152.57 m-1

TE Z0
Z10 = = 516.2 Ω
0.73
____________________________________

The most important mode is the dominant mode. This is the mode with the
combination of the lowest possible modal numbers m and n and consequently with the lowest
cut-off frequency and the simplest field distribution. From the distribution of the field of the
TM modes (7.44-48) it follows that it is not possible to set m = 0 and n = 0 as the field is zero.

98
book - 7
In the case of the TE modes it is possible, see (7.35-39), to put m = 0 or n = 0, but not
simultaneously. So we have two possible combinations of m and n (1,0) or (0,1). Which cut-
off frequency is lower depends on the relation between the dimensions of the waveguide.
Accepting the standard notation a > b, we get as the dominant mode the mode TE10. From
(7.50) and (7.51) we will get its cut-off wavelength and its cut-off frequency

λc10 = 2a , (7.52)

1
1
f c10 = . (7.53)
2 µε a

Formula (7.52) tells us that the waveguide transmits the TE10 mode starting from the
frequency at which a = λ/2. The field distribution of the dominant TE10 mode follows from
(7.35-39)

jω µ a π 
Ey = − M sin  x  e − jk z z , (7.54)
π a 

π 
H z = M cos x  e − jk z z , (7.55)
a 

jk z a π 
Hx = M sin  x  e − jk z z . (7.56)
π a 

There is only one transversal component of the electric field and one transversal component
of the magnetic field. The field does not depend on the y coordinate. The longitudinal
propagation constant is from (7.49)

2
π 
2
k z = ω µε −   . (7.57)
a

The electric field of this mode has its maximum at the center of the wider dimension,
so it can be simply excited by a probe located at the
position of the field maximum, Fig. 7.8.
Tangent components of the magnetic field create
an electric current passing along the surface of the
waveguide walls in the direction perpendicular to this
field. This means that the Hx component creates currents Ey
flowing along the waveguide. This must be taken into
account when parts of the waveguide are mounted
together. Their flanges must be well fitted to allow these
currents to pass, otherwise the increased resistance coaxial probe
between flanges raises the losses.
Fig. 7.8
____________________________
Example 7.5: Determine the frequency band of the single-mode operation of the rectangular
waveguide with internal dimensions a = 22.86 mm, b = 10.16 mm. The waveguide is filled by
air.
99
book - 7
The frequency band is limited from below by the cut-off frequency of the dominant
mode TE10. The cut-off wavelength of this mode is (7.52)

λc10 = 2a = 45.72 mm .

The cut-off frequency is

c
f c10 = = 6.56 GHz .
λc10

The frequency band of the single-mode propagation is limited from above by the cut-off
frequency of the nearest higher mode. In the rectangular waveguide it could be TE01 or TE20
modes, depending on the ratio between the waveguide dimensions. The cut-off wavelengths
of these two modes are

λc 01 = 2b = 20.32 mm, λc 20 = a = 22.86 mm.

In our case a > 2b and consequently λc20 > λc01. The nearest higher mode is therefore the TE20
mode. The cut-off frequency corresponding to λc20 is

c
f c 20 = = 13.12 GHz.
λc 20

The frequency band of the single-mode operation is from 6.56 to 13.12 GHz.

____________________________________
Example 7.6: Design the dimensions of the rectangular waveguide a, b as it transmits at the
frequency 10 GHz the only dominant mode TE10 and the two nearest higher modes TE01 and
TE20 are equally attenuated.
The two modes TE01 and TE20 will be equally attenuated if their longitudinal
propagation constants are equal

π 2π
kz01 = kz20 => k 2 − k p2 01 = k 2 − k 2p 20 => k p 01 = = k p 20 =
b a

Consequently we have a relation that assures equal attenuation of the two modes

a = 2b

The condition for propagation of the dominant mode TE10 is

c c
f > f c10 = => a> = 15 mm
2a 2f

The condition for the attenuation of the two higher modes is

c c
f < f c 01 = f c 20 = => b< = 15 mm, a = 2b => a < 30 mm
2b 2f

100
book - 7
The waveguide has the dimensions a = 2b and 15 < a < 30 mm.
_____________________________________

The power transmitted by the rectangular waveguide is calculated from the Poynting
vector

2
1  E y E y 
*
1 Ey
S av
1
2
(
* 1
2
) *
(
= Re E × H = − Re E y H x z 0 = − Re
2  Z10
)
TE  0
z =
2 Z TE
z0 .
 10

The total active transmitted power is inserted for Ey (7.54)

ab a
1 2 E m2 b 2  πx 
P = ∫∫ S av ⋅ dS = TE ∫ ∫ Ey dxdy = TE ∫ sin   dx ,
S 2 Z10 00 2 Z10 0 a

E m2 ab
P= TE
. (7.58)
4Z10

The maximum amplitude of the electric field is from (7.54) E m = Mωµ a / π .

_________________________________
Example 7.7: Calculate the maximum power that can be transmitted by the dominant mode
TE10 in a rectangular waveguide with the internal dimensions a = 22.86 mm, b = 10.16 mm at
the frequencies 5 GHz and 10 GHz. The waveguide is filled by air with the electric strength
Ep = 30 kV/cm.
According to Example 7.5 the cut-off frequency of the dominant TE10 mode in this
waveguide is 6.56 GHz. This means that this mode does not propagate at 5 GHz. So we will
calculate the power only at 10 GHz. The wave impedance of this mode is

TE Z0
Z10 = = 500 Ω ,
2
f 
1 −  c10 
 f 

consequently the transmitted power (7.58) allows the maximum value of the electric field to
be equal to the electric strength

E p2 ab
P= TE
= 1.05 MW .
4 Z10

___________________________________
Example 7.8: A waveguide with a rectangular cross-section transmits the TM11 mode at the
frequency 6 GHz. The waveguide is filled by air and its internal dimensions are a = 71 mm
and b = 35.5 mm. Calculate the magnitude of the total current passing the waveguide walls in
the longitudinal direction and the magnitude of the displacement current passing the
waveguide cross-section in the longitudinal direction. The field distribution is described by
the formulas

101
book - 7
E z = j 3000 sin (πx / a )sin (πy / b ) e − j 78.2 z ,

E x = −1060 cos(πx / a ) sin (πy / b ) e − j 78.2 z ,

E y = −2120 sin (πx / a ) cos(πy / b ) e − j 78.2 z ,

H x = 9.04 sin (πx / a ) cos(πy / b ) e − j 78.2 z ,

H y = −4.52 cos(πx / a ) sin (πy / b ) e − j 78.2 z .

The conducting current passing along the surface of the waveguide walls can be calculated
from the magnetic field boundary condition (1.16)
y Kz1
n×H = K .
Hx
The current directions are shown in Fig. 7.9. n Hy
b
In the upper and bottom walls we have Hy Kz2 n Kz2
n Kz1
K z1 = 9.04 sin (πx / a ) A/m. n Hx

In the left and right walls we have a


0 x
K z 2 = 4.52 sin (πy / b ) A/m.
z
Fig. 7.9
The total current passing the upper wall is

a a

I1 = ∫ K z1dx = 9.04 ∫ sin (πx / a )dx = 9.04 = 0.409 A
0 0
a

The total current passing the left wall is

b b

I 2 = ∫ K z 2 dy = 4.52 ∫ sin (πy / b )dy = 4.52 = 0.102 A
0 0
b

The total current is

I = 2 I1 + 2 I 2 = 1.022 A

The density of the displacement current is

J Dz = ωε 0 E z = 1000 sin (πx / a )sin (πy / b )

The total displacement current passing the waveguide cross-section is

102
book - 7
ab ab
4ab
I D = ∫ ∫ J Dz dxdy = ∫ ∫ 1000 sin (πx / a )sin (πy / b )dxdy = 1000 = 1.022 A
00 00 π2

The results show that the current continuity is preserved. The total value of the
conducting current passing the waveguide walls equals the value of the displacement current
passing the waveguide cross-section.
____________________________________

All previous results were derived assuming that all materials are without losses. This is
an ideal case. There are two sources of losses in a real waveguide. These are the finite
conductivity of the waveguide walls and the finite conductivity of the dielectric filling the
waveguide. Other losses are caused by the roughness of the waveguide wall and nonperfect
connections between the flanges. All these losses increase with the frequency. They are about
0.1 dB/m for the frequency band from 8 to 12.4 GHz, which is the so called X band.
It is not possible to design a rectangular waveguide with arbitrary dimensions. The
dimensions of these waveguides are determined by international standards depending on the
frequency band. An example is the waveguide for the X band. Its internal dimensions are a =
22.86 mm, b = 10.16 mm. The waveguides are at present used at lower frequencies only in the
case of the need to transmit high power. Planar transmission lines such as the microstrip line,
Fig. 5.1b, are used for low power applications instead of waveguides. Metallic waveguides
are very massy and their production is very expensive, so they are not suitable for mass
production. At high frequency bands, above 50 GHz, waveguides are still used, as they have
lower losses than planar transmission lines. Planar transmission lines are open transmission
lines and lose energy due to radiation at high frequencies.

7.3 Waveguide with a circular cross-section


A waveguide with a circular cross-section, Fig. 7.10, is used only in some special
applications. An example is a rotating joint which transmits an electromagnetic wave to the
feeder of a rotating radar antenna. The problem of these waveguides is that due to their
geometry they do not keep the plane of the polarization of the transmitted mode when the
waveguide is long. The frequency band of the single mode operation of a circular waveguide
is narrower than the same band of a rectangular waveguide.
In this section we will show only the basic
properties of a waveguide with a circular cross-section,
Fig. 7.10. The internal radius of this waveguide is ro, and
the longitudinal axis is z. We will assume an infinitely
long waveguide with ideally conducting walls filled by a
lossless dielectric. Even in a waveguide of this geometry ro
we can get results similar to those obtained in section 5.
We have to use the cylindrical coordinate system. The
field in a circular waveguide can be separated into TE and
TM modes, which are treated separately. We will see that
unlike in the rectangular waveguide the longitudinal z
propagation constants and the cut-off frequencies are
different for the TE modes and for the TM modes. Fig. 7.10
First we will treat TM modes. The field
distribution of these modes is determined by solving the wave equation (5.3) written for the
longitudinal component of the electric field (5.1). The cylindrical coordinate system must be
used here, see the mathematical appendix.

103
book - 7
E z (r ,α , z ) = E0 z (r ,α ) e − jk z z , (7.59)

where the transversal field distribution is obtained by solving the wave equation (5.3)
expressed in the cylindrical coordinate system (13.86)

∂ 2 E0 z 1 ∂E0 z 1 ∂ 2 E0 z
+ + 2 + k p2 E0 z = 0 . (7.60)
∂r 2 r ∂r r ∂α 2

The propagation constants kp, kz and k are coupled by (5.5). We will solve (7.60) by the
method of separation of variables. Function E0z is assumed in the form

E 0 z (r ,α ) = R(r ) ⋅ Α(α ) . (7.61)

Inserting (7.61) into (7.60) and after some manipulation we will get the equation

R' ' R ' Α' '


r2 +r + + r 2 k 2p = 0 . (7.62)
R R Α

Term Α' ' / Α must be independent of r and α, and must be equal to constant m, which gives
us the equation for function Α

Α' '+ m 2 Α = 0 . (7.63)

The solution of (7.63) is obtained using a proper origin for reading angle α in the simple form

Α = Ccos(mα ) . (7.64)

The electric field must be unique, therefore E 0 z (α ) = E0 z (α + 2mπ ) . It follows from this that
m must be an integer number. Inserting solution (7.64) into (7.62) we will get

d 2 R 1 dR  2 m 2 
+ +  k p − 2  R = 0 . (7.65)
d r 2 r d r  r 

This equation is Bessesl’s equation of the m-th order. The solution is in the form of the sum of
the Bessel functions of the m-th order Jm and Ym, see the mathematical appendix,

( )
R = A J m k p r + BYm k p r . ( ) (7.66)

To determine constants A and B we have to apply the boundary conditions. The tangent
component of the electric field on the surface of the waveguide wall must be zero and the
field must have a finite value at the waveguide center. The second condition gives us B = 0, as
Bessel function Ym has an infinite value at the origin, see the mathematical appendix.
Component Ez is tangent to the waveguide wall, so the first boundary condition reads

R(r = ro ) = 0 => ( )
J m k p ro = 0 . (7.67)

104
book - 7
The product kpro must be equal to the n-th zero point of the Bessel function of the m-th order
αmn. These values are listed in Table 13.1 in the mathematical appendix. Consequently the
transversal propagation constant of the mode with modal numbers m and n is

TM α mn
k pmn = . (7.68)
ro

From (5.18) and (5.19) we get the longitudinal propagation constant and the cut-off
frequency of the TM modes in the circular waveguide

2
α 
TM
k zmn = ω µε −2 2
k pmn = ω µε −  mn 
2
, (7.69)
 ro 

TM 1 α mn
f cmn = . (7.70)
2π µε ro

According to Table 13.1 we have the order of the lowest modes TM01, TM11, TM21,
TM02 and TM31, as the lowest numbers αmn are α01 = 2.40482, α11 = 3.83171, α21 = 5.13562,
α02 = 5.52007, α312 = 6.38016.
The field distribution of particular modes can be determined in a similar way as the
distribution of TM modes in a rectangular waveguide. The cylindrical coordinate system must
be applied. Longitudinal component Ez has the form, using (7.64) and (7.66),

( )
E z = E 0 J m k p r cos(mα ) e − jk zmn z , (7.71)

where E0 is an unknown amplitude. The transversal components are Er, Eα, Hr and Hα. The
field distribution of the lowest modes is shown in Fig. 7.11.

electric field magnetic field

TE11 TM01 TE01 TM11


Fig. 7.11

The field distribution of the TE modes is determined by solving the wave equation
written for the longitudinal component of the magnetic field. Using the same procedure as in
the case of the TM modes we will get the distribution of Hz

( )
H z = H 0 J m k p r cos(mα ) e − jk zmn z , (7.72)

105
book - 7
where H0 is an amplitude. The TE modes have the Eα component of the electric field tangent
to the waveguide wall. This component is proportional to ∂H z / ∂r , so the boundary condition
is
∂H z
∂r r = ro
= 0 => J m' k p ro = 0 .( ) (7.73)

The product kpro must be equal to the n-th zero point of the derivative of the Bessel function
'
of the m-th order α mn . These values are listed in Table 13.2 in the mathematical appendix.
Consequently the transversal propagation constant of the mode with modal numbers m and n
is
'
TE α mn
k pmn = . (7.74)
ro

From (5.18) and (5.19) we get the longitudinal propagation constant and the cut-off
frequency of the TE modes in a circular waveguide

2
α ' 
TE
k zmn = ω µε −2 2
k pmn = ω µε −  mn 
2
, (7.69)
 ro 

'
TE 1 α mn
f cmn = . (7.70)
2π µε ro

According to Table 13.2 we have the order of the lowest modes TE11, TE21, TE01, TE31
' ' ' ' '
and TE41, as the lowest numbers α mn are α 11 = 1.84118, α 21 = 3.05424, α 01 = 3.83170, α 31
'
= 4.20119, α 41 = 5.31755. The transversal propagation constants and the cut-off frequencies
of the TE modes are different from those of the TM modes. The field distribution of the
lowest modes over the waveguide cross-section is shown in Fig. 7.11.
For the circular waveguide the same relations as (7.21), (7.22), (7.23) and (7,24) can
be derived for the wavelength along the waveguide, the phase velocity, the group velocity,
and the longitudinal propagation constant, of course separately for the TE and TM modes.
The wavec impedance is defined as the ratio of the transversal components

Er E
Z TE ,TM = =− α .
Hα Hr

The dominant mode of the circular waveguide is the TE11 mode, as its cut-off
'
frequency has the lowest value. The value of the corresponding parameter α 11 = 1.841. The
cut-off frequency of this mode is

TE 1
1.841
f c11 = . (7.71)
2π µε ro

The cut-off wavelength of the dominant TE11 mode is

106
book - 7
2π ε r ro
λTE
cmn = = 3.4 ε r ro . (7.72)
1.841

___________________________________
Example 7.9: Determine the frequency band of the single mode propagation of a circular
waveguide filled by a dielectric with relative permittivity εr with radius ro = 8.2 mm.
The dominant mode of the circular waveguide is the TE11 mode with the cut-off
wavelength (7.72). The nearest higher mode is the TM01 mode with the cut-off wavelength

v 2π ε r ro
λTc 01M = = = 2.6 ε r ro .
f cTM
01 2.40482

For the waveguide radius ro = 8.2 mm we get the band of the single mode operation

21.32 < λ < 27.88 mm

10.76 < f < 14.07 GHz


____________________________________

7.4 Problems
7.1 Calculate for the modes propagating between the two parallel plates from Example 7
the wavelength, the phase velocity, the group velocity, the longitudinal propagation constant
and the wave impedance. The frequency is 10 GHz.
mode vpm vgm λgm kzm ZTM ZTE
-1
(m/s) (m/s) (mm) (m ) (Ω) (Ω)
8 8
0th 3 10 3.10 30 209 376.6 -
8 8
1st 3.24 10 2.78 10 32.4 193.7 349.1 406.3
2nd 4.55 108 1.98 108 45.5 137.9 248.6 570.6

7.2 Calculate the frequency band of the propagation of only the TE1 mode in a parallel
plate waveguide filled by air. The distance of the plates is 40 mm. Do not consider the TM
modes.
fc1 = 3.75 < f < fc2 = 7.5 GHz

7.3 How does the result of Problem 7.2 change if we replace the air by a dielectric with
permittivity εr = 9?
fc1 = 1.25 < f < fc2 = 2.5 GHz

7.4 Determine the points at the cross-section of the rectangular waveguide at which the
magnetic field of modes TE10, TE01, TE20, TE11 and TM11 is zero.

7.5 Determine the points at the cross-section of the rectangular waveguide at which the
electric field of modes TE10, TE01, TE20, TE11 and TM11 is zero.

7.6 The maximum value of the amplitude of the electric field in the rectangular waveguide
is Em = 600 V/m. The dominant mode TE10 propagates in this waveguide at the frequency 10

107
book - 7
GHz. The waveguide is filled by air and its internal dimensions are a = 22.86 mm and b =
10.16 mm. Calculate the amplitudes of all field components.
Eym = -600 V/m
Hxm = 1.19 A/m
Hzm = 1.04 A/m

7.7 Calculate the electric field intensity at the point with the coordinates x = 5.5 mm and y
= 5.1 mm in the cross section of a rectangular waveguide filled by air. The waveguide
transmits the dominant mode TE10 at the frequency 15 GHz. The transmitted power is 10 W.
The internal dimensions of the waveguide are a = 22.9 mm, b = 10.2 mm.
E y = − j 575 e − j 282 z V/m

7.8 Determine which modes can propagate along a circular waveguide filled by air with
radius ro = 5.2 mm at frequency f = 30 GHz.
TE11, TM01, TE21

7.9 Calculate for the mode TE11 propagating along the waveguide from problem 7.8 at
frequency of 30 GHz the longitudinal propagation constant and the wavelength.
-1
k zTE
11 = 518.9 m

λTE
g11 = 12.1 mm

8. DIELECTRIC WAVEGUIDES
In the past, dielectric waveguides were used in microwave systems mainly as
dielectric antennas. The wave was guided by a dielectric rod, the cross-section of which was
tapered and the wave was gradually radiated out into the space. Dielectric waveguides are
nowadays frequently applied in optical communication systems as optical fibers and in the
form of planar optical waveguides in the circuits of optical integration systems.
The operation of dielectric waveguides is
based on total reflection of the wave on the boundary evanescent wave
between the two dielectric materials treated in
paragraph 3, see Fig. 3.15. The wave is totally vpz<v2
reflected back if it is incident to the boundary of the ε2
slow wave
two dielectric materials from the side of the material
with greater permittivity at an angle greater than the standing wave
critical angle. The field penetrates into the second
Ey
material as an evanescent wave with the amplitude ε1
decreasing exponentially perpendicular to the vpz>v1
boundary. Perpendicular to the boundary the wave in fast wave
the first material has the character of a standing wave.
The wave propagates as a surface wave in the vpz<v2
ε2
direction along the boundary, Fig. 3.15. Let us now slow wave
imagine that we confine the material with greater
permittivity to a layer of finite thickness 2a. We ε1 > ε2
evanescent wave
couple into this layer the wave as it is incident to the
boundary at an angle greater than the critical angle. It Fig. 8.1
is reflected back and is incident to the opposite

108
book - 8
boundary and bounces up and down. According to Fig. 8.1 the wave is in this way coupled to
the layer, but it also penetrates into the surrounding material, where its amplitude decreases
exponentially. The field distribution across the layer depends on the frequency and on the
thickness of the layer.
Thus any dielectric structure with a core made from a material with permittivity
greater than the permittivity of the surrounding material can behave as a dielectric waveguide.
The guiding of the wave is due to the total internal reflection on the surface of the core.

8.1 Dielectric layers


A dielectric layer is the dielectric d
guiding structure which can be treated most -a ε2
simply. Let us assume a dielectric c
y
homogeneous layer that is infinitely long and 0
wide, and its thickness is 2a, see Fig. 8.2. We ε1 z
assume that the field does not depend on the y a

ε2 e
coordinate, therefore = 0 , and that the x
∂y
wave propagates in the z axis direction. Thus Fig. 8.2
we can treat the TE and TM modes
separately, similarly as in the parallel plate
waveguide. According to (5.14-17) the TE modes have non-zero field components Ey, Hx and
Hz, and the TM modes have components Hy,
Ex and Ez. TE modes
Let us first study the TE modes. The H d
mode propagates in the dielectric layer due -a ε2
to the internal total reflections, see Fig. 8.3. c
y
Its electric field has only the Ey component, 0
and the magnetic field must thus have the Hx E ε1 z
and Hz components. The wave equation can a
by now solved for the Ey component of the ε2 e
electric field. The solution obtained x
similarly as in paragraph 5 by the method of
separation of variables can be written in the Fig. 8.3
particular regions c, d, e, Fig. 8.3,

E y1 = [A sin (k x1 x ) + B cos(k x1 x )]e − jk z z , x <a, (8.1)

E y 2 = C e jk x 2 z e − jk z z , x < -a , (8.2)

E y 3 = D e − jk x 2 z e − jk z z , x>a, (8.3)

where the separation constants are coupled by

k12 = ( ε r1 k 0 ) = ( ε r1ω ε 0 µ 0 ) = k x21 + k z2 ,


2 2
(8.4)

k 22 = k 32 = ( ε r 2 k 0 ) = ( ε r 2 ω ε 0 µ 0 ) = k x22 + k z2 .
2 2
(8.5)

109
book - 8
Constants kx1 and kx2 are the propagation constants in the x direction in the dielectric layer and
in the surrounding material, respectively.
We can simplify the treatment of the field by dividing the TE modes into two groups.
In the first group we have symmetric, or even modes, which satisfy the condition

E y ( x ) = E y (− x ) . (8.6)

For these modes we put in (8.1) A = 0 and they are described by the cosine function. In the
second group we have anti-symmetric, or odd modes which satisfy the condition

E y ( x ) = − E y (− x ) . (8.7)

For these modes we put in (8.1) B = 0 and they are described by the sinus function.
The Ey component of the even TE modes in the dielectric layer is

E y1 = B cos(k x1 x ) e − jk z z . (8.8)

The magnetic field is related to the electric field by Maxwell’s second equation (1.13).
Assuming Ex = Ez = 0 we will get

1 ∂E y
Hz = , (8.9)
jωµ ∂x

and the z component of the magnetic field in our three particular areas is

k x1
H z1 = − B sin (k x1 x ) e − jk z z , x <a, (8.10)
jωµ

k x2
H z2 = C e jk x 2 z e − jk z z , x < -a , (8.11)
ωµ

k x2
H z3 = − D e − jk x 2 z e − jk z z , x>a. (8.12)
ωµ

The field must fulfill the boundary conditions on the layer surfaces at x = a and x = -a.
Consequently at x = a and for arbitrary z we have

E y1 (a ) = E y 3 (a ) , H z1 (a ) = H z 3 (a ) ,

B cos(k x1a ) = D e − jk x 2a , (8.13)

k x1 k
− B sin (k x1a ) = − x 2 D e − jk x 2a .
jωµ ωµ

− jk x1 B sin (k x1a ) = k x 2 D e − jk x 2a . (8.14)

110
book - 8
Dividing (8.14) by (8.13) we get

(k x1a ) tg(k x1a ) = j (k x 2 a ) . (8.15)

This equation couples propagation constants kx1 and kx2, but to determine them we need to get
another equation. This equation is derived by inserting from (8.4) for kz into (8.5)

k x21 − k x22 = (ε r1 − ε r 2 )ε 0 µ 0ω 2 . (8.16)

Now to remove the imaginary unit from (8.15) we introduce a new variable

k x' 2 = j k x 2 . (8.17)

Consequently we have the two equations known as dispersion equations

(k x1a ) tg(k x1a ) = (k x' 2 a ) , (8.18)

(k a ) + (k a )
x1
2 '
x2
2
= (ε r1 − ε r 2 )ε 0 µ 0ω 2 a 2 . (8.19)

The Ey component of the odd TE modes is

E y1 = A sin (k x1 x ) e − jk z z . (8.20)

Applying exactly the same procedure for calculating the distribution of the corresponding
magnetic field and fulfilling the boundary conditions we will get the dispersion equation for
the odd TE modes in the form

(
− (k x1a ) cotg(k x1a ) = k x' 2 a . ) (8.21)

The second dispersion equation remains in the Ey


form (8.19).
The electric field distribution of TE
d
modes in the plane transversal to the
propagation direction is described by (8.8) or m=0 m=1 m=2
(8.20) in the layer itself and by (8.2) and (8.3) in
the surrounding material. This distribution is c 2a
plotted for the three lowest TE modes in Fig.
8.4. The modes are denoted by numbers which
determine the number of zeros of the Ey
function across the layer. So we have, Fig. 8.4, TE0 TE1 TE2 e
the series of modes: even TE0, odd TE1, even even odd even
TE2, odd TE3, etc.
Solving equations (8.18) and (8.19) or Fig. 8.4
(8.21) and (8.19) we determine the propagation
constants kx1 and kx2, and the longitudinal propagation constant is then calculated from (8.4)
or (8.5). In this way we can get the dispersion characteristic of the given TE mode
propagating along our dielectric layer. The dispersion characteristic represents the dependence
of the propagation constant on the frequency. Dispersion equations (8.18) and (8.19) cannot
111
book - 8
be solved analytically. The solution must be k‘x2a
performed using a proper numerical
method. This problem can be overcome by M0
a simple graphical technique, which gives a M1
clear physical insight of the effect of
guiding the waves along the dielectric
layer. In the coordinate system M2

( )
(k x1a ) and k x 2 a ,
'
equation (8.19)
TE0 TE1 TE2 TE3

represents a circle with its center at the


origin. The radius of this circle is
π/2 π 3π/2 kx1a
R =ωa (ε r1 − ε r 2 )ε 0 µ 0 .
(8.22)
R
The radius is proportional to the frequency,
the layer thickness and the second root of TE4
the difference between the permittivities.
The solution of the set of equations (8.18) M0non
and (8.19) or (8.21) and (8.19) can be
obtained graphically as an intersection of
the circle (8.19) with the curve represented Fig. 8.5
by (8.18) or (8.21), Fig. 8.5.
From Fig. 8.5 we see that we always have one solution of the dispersion equations,
which is represented by the intersection of the circle (8.19) with the first branch of the tangent
function on the left hand side of (8.18). This means that the TE0 mode can propagate under
any circumstances at any layer at any frequency. This is the dominant mode of this dielectric
waveguide. The nearest higher mode is the TE1 mode. So the upper frequency of the
frequency band of the single mode propagation is determined by the cut-off frequency of the
TE1 mode. The solutions of the dispersion equations lying in the lower half plane of the plane
( ) ( )
(k x1a ) and k x' 2 a determined by k x' 2 a < 0 are not physical, as they correspond to the field
(8.2) and (8.3) which increases exponentially in the surrounding areas d and e.
Consequently, the cut-off frequency of the TE1 mode follows from the condition R = π/2,
which gives the cut-off frequency of this mode

1 1
f c1 = . (8.23)
4a (ε r1 − ε r 2 )ε 0 µ 0
The cut-off frequency of any mode with the modal number m > 0 is determined by the
condition R = mπ/2, so generally we have

m 1
f cm = . (8.24)
4a (ε r1 − ε r 2 )ε 0 µ 0
Fig. 8.5 shows the plots of the dispersion equations (8.18) for modes TE0, TE2 and
TE4, (8.21) for modes TE1 and TE3. There are the three propagating modes TE0, TE1 and TE2
at the given frequency. The corresponding solutions of the dispersion equations are marked in
Fig. 8.5 by M0, M1 and M2. The solution marked M2non is non-physical.

112
book - 8
_______________________________________
Example 8.1: Calculate the longitudinal propagation constant of the TE 0 mode propagating
along a dielectric layer 5 mm in thickness with the relative permittivity εr = 2.6 at the
frequency f = 62.5 GHz. The layer is surrounded by air.
According to (8.22) we have R = 4.135. The corresponding dispersion equations are
plotted in Fig. 8.5. Our solution is marked by M0. From Fig. 8.5 we can read
( )
(k x1a ) = 1.29, k x' 2 a = 3.93 . From these values we have kx1 = 516 m-1, and (8.4) gives us kz =
2042 m-1.
___________________________________ TM modes

The same procedure can be followed ε2 d


for the TM modes. These modes propagate -a H
in the dielectric layer due to the internal total 0 y c
reflections, see Fig. 8.6. Their magnetic field E ε1 z
has only the Hy component and the electric a
field must thus have the Ex and Ez ε2 e
components. The wave equation can now be x
solved for the Hy component of the magnetic
field. The solution obtained similarly as in Fig. 8.6
paragraph 5 by the method of separation of
variables can be written in the particular regions c, d, e, Fig. 8.6,

H y1 = [A sin (k x1 x ) + B cos(k x1 x )]e − jk z z , x <a, (8.25)

H y 2 = C e jk x 2 z e − jk z z , x < -a , (8.26)

H y 3 = D e − jk x 2 z e − jk z z , x>a, (8.27)

where the separation constants are coupled by (8.4) and (8.5). The distribution of the Hy
component of the magnetic field of the TM modes is the same as the distribution of the
electric field shown in Fig. 8.4. The rest of the process is the same as in the case of the TE
modes. From the Hy component of the magnetic field we calculate the distribution of the Ez
electric field by

1 ∂H y
Ez = − . (8.28)
jωε ∂x

We divide the field into even and odd modes and we meet the boundary conditions at x = a,
which are Hy1(a) = Hy3(a) and Ez1(a) = Ez3(a). As a result we obtain the dispersion equations
for the even TM modes

ε2
ε1
(
(k x1a ) tg(k x1a ) = k x' 2 a , ) (8.29)

and for the odd TM modes

113
book - 8

ε2
ε1
( )
(k x1a ) cotg(k x1a ) = k x' 2 a . (8.30)

k‘x2a
The equation of the circle (8.19) remains
unchanged. The cut-off frequencies of the M0
TM modes are then defined by (8.24) as for
the TE modes. The dominant mode is the M1
TM0 mode, which can propagate at any TM0 TM2
frequency. The frequency band of the single TM1
mode operation is confined from above by
the cut-off frequency of the TM1 mode M2
(8.23). The only difference between the TM3
solutions of the dispersion equations is their
shift caused by the term ε2 /ε1, Fig. 8.7. π/2 π 3π/2 kx1a
The dispersion characteristic of the
modes propagating along the dielectric layer TM4
can be obtained by repeating the procedure R
shown in Example 8.1 for the TE modes,
and similarly for the TM modes. The
dispersion characteristic of the TM0 and TE1
modes propagating along the dielectric layer M0non
defined in Example 8.1 is plotted in Fig. 8.8.
Note the cut-off frequency of the TE1 mode.
Fig. 8.7

8.2 Dielectric cylinders


The dielectric cylinder, the cross-
section of which is shown in Fig. 8.9, TM0
represents the dielectric waveguide
widely used in practical applications. TE1
Optical fiber is based on it. In the case
of optical fiber the surrounding material
with relative permittivity εr2 does not fill
the whole space, but creates a cover
layer, see Fig. 5.1c. The cylindrical
dielectric waveguide is able to transmit fc1
the wave due to the internal reflection,
Fig. 8.8
assuming that ε1 > ε2.
The field distribution can be determined
similarly as in the case of a cylindrical
εr2 waveguide, using the wave equation in the
cylindrical coordinate system. However, we have
εr1 to assume that the field penetrates into the
d c r0
surrounding material d. The field cannot be
z divided into TE and TM modes due to the
boundary conditions on the surface of the
dielectric cylinder. All modes propagating along
Fig. 8.9 the dielectric cylinder have nonzero components
114
book - 8
Ez and Hz. Only for the cylindrically symmetric modes with the modal number m = 0 do we
have the modes TE0n and TM0n.
The longitudinal components of the electric and magnetic fields are assumed in the
form

E z1 (r ,α , z ) = E 0 z1 (r ,α ) e − jk z z , r < r0 , (8.31)

E z 2 (r ,α , z ) = E0 z 2 (r ,α ) e − jk z z , r > r0 . (8.32)

H z1 (r ,α , z ) = H 0 z1 (r ,α ) e − jk z z , r < r0 , (8.33)

H z 2 (r , α , z ) = H 0 z 2 (r ,α ) e − jk z z , r > r0 . (8.34)

Functions E0z1, E0z2, H0z1 and H0z2 are solutions of the wave equation (7.60) and a similar
equation for the magnetic field. Applying the same procedure as in the case of the metallic
waveguide we will obtain the solution

( )
E z1 = A J m k p1r e jmα e − jk z z , r < r0 , (8.35)

( )
E z 2 = B K m jk p 2 r e jmα e − jk z z , r > r0 , (8.36)

( )
H z1 = C J m k p1r e jmα e − jk z z , r < r0 , (8.37)

( )
H z 2 = D K m jk p 2 r e jmα e − jk z z , r > r0 . (8.38)

Where Km is the modified Bessel function of the second kind, see the mathematical appendix.
This function describes the evanescent character of the field outside the cylinder. The electric
and magnetic fields (8.35-38) together with Eα1, Eα2, Hα1 and Hα2 have to meet the boundary
conditions on the cylinder surface as they are tangent to this surface. From these boundary
conditions we can determine the constants B, C and D and the dispersion equations
necessary to determine the propagation constants of particular modes.

8.3 Problems
8.1 Determine the frequency band of the single mode operation and the cut-off frequencies of
four modes in a plexiglass slab 5 mm in thickness with relative permittivity equal to 2.6. The
slab is infinitely wide and long and is surrounded by air.
fc1 = 23.7 GHz
fc2 = 47.4 GHz
fc3 = 71.1 GHz
fc4 = 94.8 GHz
the single mode operation band is 0 < f < 23.7 GHz

8.2 Determine the maximum thickness of a dielectric layer with relative permittivity εr = 2.28
which ensures propagation of the TE0 mode alone at the wavelength λ0 = 1.3 µm. The
dielectric layer is placed between two glass layers with the relative permittivity εr = 2.26.
h ≤ 4.6 µm
115
book - 8
8.3 Calculate the phase velocity of the TE0 mode propagating along the dielectric layer from
the example 8.1 at the frequency f = 62.5 GHz.
vpTE0 = 2.53 108 m/s

8.4 Determine the field distribution and dispersion equations of the TE and TM modes
propagating along a dielectric layer of thickness a placed on an ideally conducting plane, Fig.
8.10.
The structure in Fig. 8.10 is one half of the structure shown in Fig. 8.2, with the plane
of symmetry located at x = 0. Only the
even TM modes and the odd TE modes x
can exist on this structure due to its ε2 d
symmetry. Apart from this, everything else a
is the same as was explained for the 0 y ε1 c
dielectric layer. The dominant mode is the σ →∞ z
TM0 mode, and the TE0 mode does not Fig. 8.10
exist.

9. RESONATORS

9.1 Cavity resonators


Microwave resonators are used in microwave circuits similarly as standard resonant
LC circuits at lower frequencies. At high frequencies we have to protect the electromagnetic
field from radiation outside the resonator. For this reason, resonators are very often used as
cavity resonators. These resonators are formed by the volume filled by a dielectric material,
mostly air. This volume is shielded by a suitable metallic coating. The electromagnetic field is
coupled inside, using a suitable probe. We will treat lossless resonators, assuming that the
dielectric material has zero conductivity and on the other hand the metal coating has infinite
conductivity.
Let us first determine the resonant frequency of a general cavity resonator. We use the
power balance for the reactive power described by equation (1.33). Assuming a shielded
lossless structure without any radiation, the power balance for the reactive power is

1
∫∫∫
2 V
( ) 1
(2 2
Im J ⋅ E * dV = ω ∫∫∫ ε E − µ H dV .
2 V
)
Disconnecting the source we have J = 0 and we get the self-oscillation of the field in the
volume. The condition for self-oscillation is determined by the equality of energies We = Wm,
which is expressed by

2 2
ε ∫∫∫ E dV = µ ∫∫∫ H dV .
V V

The electric field can be determined from Maxwell’s first equation (1.12) assuming σ = 0 and
no external current. Then we have

116
book - 9
1
rot (H ) .
2 2
E = 2 2
ω ε

Finally we get the resonant frequency of our cavity

∫∫∫ rot(H ) ∫∫∫ rot(E)


2 2
dV dV
1 1
ω 02 = V
= V
. (9.1)
µε 2
µε 2
∫∫∫ H dV ∫∫∫ E dV
V V

It follows from (9.1) that the resonant frequency of the cavity depends on the material
parameters µ and ε and on the cavity geometry that is hidden in the integrals. This is common
with standard LC resonant circuits. However, (9.1) shows that the resonant frequency
depends on the distribution of the electric and magnetic fields. Consequently as we can
excite an infinite number of particular modes in the cavity, we have an infinite number of
resonant frequencies corresponding to these modes. Some modes are known as degenerated.
These modes have a different field distribution but an equal resonant frequency.
Let us first determine the resonant frequency of
the cavity resonator formed by a segment of any
waveguide terminated at both ends by ideally
conducting planes. The planes are perpendicular to the
longitudinal axis of the waveguide. The length of the
d z
segment is d, Fig. 9.1, so the conducting planes are 0
located at z = 0 and z = d. It is reasonable to assume that d
the field distribution of the mode excited in this cavity
originates from the distribution of the modes existing in Fig. 9.1
the original waveguide forming the cavity. These
modes have transversal and longitudinal propagation constants kpmn and kzmn (5.5) which
depend on two transversal modal numbers m and n. The distribution of the transversal
components of its electric field is ET0(x,y) as the function of the two transversal coordinates x,
y. The electric field in our cavity is now the superposition of the two waves traveling in the
positive z direction and in the negative z direction

(
ET (u, v, z ) = E T 0 (x, y ) A e − jk zmn z + B e jk zmn z . )
This field must fulfill the two boundary conditions at z = 0 and z = d. At z = 0 we have

ET 0 ( A + B ) = 0 => A = -B .

The field distribution is now

( )
ET (u, v, z ) = E T 0 (u, v ) − B e − jk zmn z + B e jk zmn z = ET 0 (u, v )2 jB sin (k zmn z ) . (9.2)

At z = d we have

sin (k z d ) = 0 .

117
book - 9

k zmnp = , p = 1, 2, 3, … . (9.3)
d

(9.3) tells us that we have the set of discrete values of the longitudinal propagation constants
of the modes oscillating in the resonator. These constants depend on modal numbers m, n and
p. Now using (5.5) and (2.25) we have the set of resonant frequencies

2
1 2  pπ 
f 0 mnp = k pmn +  . (9.4)
2π µ 0ε 0 ε r  d 

The particular value of the resonant frequency depends on transversal propagation


constant kpmn, which is determined by the type of waveguide from which the resonator is
composed. In the case of a waveguide with a rectangular cross section with dimensions a
and b, kpmn is determined by (7.40) and the resonant frequency is

2 2 2
1  mπ   nπ   pπ 
f 0 mnp =   +  +  . (9.5)
2π µ 0ε 0 ε r  a   b   d 

This formula is valid for both the TE and TM modes.

_____________________________________
Example 9.1: Design a resonator tuned to the resonant frequency 15 GHz for the mode TE101.
The resonator is formed by the segment of the rectangular waveguide with the dimensions a =
30 mm, b = 15 mm. The resonator is filled with air.
The dominant mode of the rectangular waveguide has the simplest field distribution,
so we design the resonator for this mode. Thus for mode TE101 we have the resonant
frequency (9.5)

c 1 1
f 0101 = 2
+ 2 ,
2 a d

where c is the speed of light in a vacuum. From the above formula we can determine the
necessary length d

1
d= = 10.6 mm.
2
 2 f 0101  1
  − 2
 c  a

The resonator must be 10.6 mm long.


___________________________________

In the case of a cylindrical resonator with radius ro we have to distinguish between


the TE and TM modes, as their transversal propagation constants are different. For TE
modes, kpmn is determined by (7.74), and the resonant frequency is

118
book - 9
2 2
 α mn
' 
f 0 mnp =
1   +  pπ  . (9.6)
2π µ 0 ε 0ε r  r   d 
 o 

For TM modes, kpmn is determined by (7.68), and the resonant frequency is

2 2
1  α mn   pπ 
f 0 mnp =   +   . (9.7)
2π µ 0 ε 0ε r  ro   d 

The dominant mode of the cylindrical waveguide is the TE11 mode, so the resonators are
'
mostly designed to work with this mode, where α 11 = 1.841 .
_____________________________________
Example 9.2: Design a cylindrical resonator tuned to the resonant frequency 14 GHz for the
mode TE111. The resonator is filled with air and its radius is ro = 8.2 mm.
Applying the same procedure as in example 9.1 we get the resonator length

π
d= = 16.67 mm.
2 2
 2πf 0101   α 11 
'
  −
 c   ro 

The resonator must be 16.67 mm long.


___________________________________

A similar procedure can be applied to determine the resonant frequency of a dielectric


resonator of either the circular or rectangular cross section terminated by conducting planes.
The conducting planes must exceed the cross section of the resonator, as the field is not
confined to the dielectric, but it penetrates into the surrounding air. The resonant frequency is
then determined by (9.4), where the proper value of the transversal propagation constant must
be used.
Cavity resonators are used in microwave technology at frequencies above about 5
GHz. At lower frequencies the transversal dimensions of the waveguide, a segment of which
forms the resonator, are too big. For this reason, these resonators become impractical at such
frequencies. We can use either dielectric resonators or ferrite resonators. Another possibility
is to form the resonator from a TEM transmission line. This line can be left either short
circuited or open at its ends. A short circuited termination is more convenient, as it does not
radiate. A TEM transmission line has kpmn = 0, consequently the resonant frequency is
determined directly by the number of periods of the standing wave along the line segment. So
we have from (9.4)

p
f0 p = . (9.8)
2d µ 0 ε 0 ε r

In the case of a real resonator with some losses we define the quality factor of the
resonator. The internal quality factor determined by the losses of the resonator itself is defined
by

119
book - 9
ω 0W
Q= , (9.9)
PL

where ω0 is the resonant frequency, W is the averaged value of the total energy of the
electromagnetic field inside the cavity, PL is the power lost in the resonator. The losses in the
dielectric material filling the cavity can usually be neglected, namely in the case of air.
Consequently the power is lost in the cavity walls. The stored energy is

ε 2

2 ∫∫∫
W = Weav + Wmav = 2Weav = E dV . (9.10)
V

The power lost in the cavity walls can be calculated from the known distribution of the
magnetic field on these walls using (3.26).
Dielectric resonators are used namely in microwave integrated systems. They are
significantly smaller than cavity resonators, which is their main advantage. These resonators
are usually designed as cylinders made of a dielectric material with high permittivity. This
significantly reduces the dimensions. The disadvantage is that the field penetrates into the
surrounding space, which may increase the losses and reduce the quality factor. On the other
hand the electromagnetic field can be simply coupled into such a dielectric resonator by the
proximity effect. The resonator is placed in the vicinity of a microstrip line and the field is
directly coupled into its volume. Due to the field penetration outside the dielectric resonator,
its resonant frequency can in most cases be determined by solving the field distribution
numerically. The resonant frequency can be approximately determined by (9.4), as mentioned
above, when the dielectric resonator is terminated by conducting planes.
Ferrite resonators are produced from mono-crystals of yttrium iron garnet (YIG) in
the shape of small spheres. The resonant frequency of these resonators depends on the
stationary magnetic field H0 applied to this material. It is, as will be shown in chapter 11,

ω0 = γ H0 , (9.11)

where γ is gyromagnetic ratio, see Paragraph 11.1. These resonators can be simply tuned by
changing the applied stationary magnetic field.

9.2 Problems
9.1 Calculate the resonant frequency of the cavity resonator formed by a segment of the
waveguide with a rectangular cross section with the dimensions a = 25 mm, b = 15 mm, d =
10 mm for the modes TE101 and TE111. The cavity is filled with air.
f0101 = 18.14 GHz
f0111 = 19 GHz

9.2 Calculate the necessary length d of a cavity resonator of rectangular shape to get the
resonator tuned to 10 GHz for the mode TE101. The cavity is filled with air. The transversal
dimensions are a = 40 mm, b = 15 mm.
d = 16.2 mm

9.3 Calculate the resonant frequency of a resonator with a circular cross section filled with air
for the mode TE111. The radius of the cavity is ro = 8.2 mm, the length is d = 25 mm.
f0111 = 12.27 GHz

120
book - 9
9.4 Calculate the resonant frequency and the quality factor of a resonator with a rectangular
cross section filled with lossless air for the TE101 mode. The dimensions of the cavity are a =
22.86 mm, b = 10.16 mm, d = 22.86 mm. The wall conductivity is σ = 5.7 . 107 S/m.
f0101 = 9.28 GHz
Q = 7758

10. RADIATION
Radiation is usually understood as the transmission of electromagnetic waves in a free
space from a source of finite dimensions supplying finite power. We assume an infinite space
filled by a homogeneous and isotropic dielectric material without any losses. We have already
studied the propagation of waves in a free space, or along transmission lines. We did not take
into account the source of the electromagnetic field. For example, a plane electromagnetic
wave with surfaces of constant amplitude and phase in the shape of an infinite plane must be
excited by a hypothetical source with infinite dimensions that must radiate infinite power. In
the case of transmission lines, we just studied the field in the view of possible waves or
modes, known as eigen modes, which can propagate along this line. We did not investigate
whether these modes would really propagate, assuming that the line were connected to a real
source supplying finite power.
Radiation is used for the transmission of signals namely in telecommunications,
radars, sensors and navigation. Radiation also occurs as a parasitic effect that represents
losses, cross-talk and parasitic interference. For example, radiation from open transmission
lines and electric circuits.
In the first section we showed that electric current and electric charges are sources of
an electromagnetic field. However, electric current is the flow of the charge. These two
quantities are coupled by a continuity equation (1.8). This means that we can consider only
the electric current as the source. The real source is therefore a conducting body of finite
dimensions, and the electric current passes through its volume or at high frequencies only
along its surface. From the known distribution of this current we can calculate the field
distribution by solving the wave equation. The wave equations written directly for field
vectors (1.41) and (1.42) have rather complicated functions of the source current and charge.
Therefore they are not suitable for solving our problem. To remove this inconvenience we
have introduced vector and scalar potentials, which represent the electric and magnetic fields
by (1.46) and (1.52). Inserting these formulas into Maxwell’s equations we have derived wave
equations for potentials with simple functions of the source current and charge (1.54) and
(1.55). Their solution is in the form of the superposition of spherical waves (1.57) and (1.58).
In this textbook we will calculate the electromagnetic field distribution only in the
cases of simple sources – antennas, where we know in advance the distribution of the electric
current. First we will treat simple elementary electric and magnetic dipoles. An electric dipole
is represented by a very short straight conductor. The magnetic dipole is represented by a
conducting loop of infinitely small dimensions. We will show how to calculate the field
radiated by sources of dimensions comparable with the wavelength. We will define the
parameters of antennas. Finally, we will mention antenna arrays and receiving antennas.

10.1 Elementary electric dipole


An elementary electric dipole consists of two point electric charges q and -q placed at
distance d. These charges harmonically change their values. The dipole moment is

121
book - 9
Id
p = Qd = , (10.1)

z
as the charge is an integral of the electric current. So the dipole A
can be represented by a straight conductor with passing current r
I, Fig. 10.1. The conductor must be much shorter than the dS
wavelength, as the amplitude and phase of the current are ϑ
constant along the conductor. The cross-section of the
d y
conductor is negligible. Then the vector potential is simply
calculated using (1.58) J
x
α
− jkr
µ Je
A = z0
4π ∫∫∫ r
dV ,
Fig. 10.1
V

Current density J has according to Fig. 10.1 the direction determined by unit vector z0, and is
constant within the conductor volume. The integral can be omitted, as J is constant within the
volume of the dipole and we get

µ e − jkr
A = z0 Id . (10.2)
4π r

The magnetic field is calculated from the known distribution of the vector potential using
(1.46) and (1.12), assuming that the space is filled with a non-conducting material. First we
transform the vector potential from the Cartesian coordinate system to the spherical
coordinate system. Using Fig. 10.2 and using the fact that the field does not depend on angle
α we get

A = Ar r0 + Aϑϑ0 = Az cos(ϑ ) r0 + Az sin (ϑ )ϑ0 . (10.3)


z
The magnetic field can be obtained using (1.46), taking into account Aϑ
that Aϕ = 0 and ∂ / ∂α = 0 , and using the expression of the rotation Az
operator in the spherical coordinate system (13.89) ϑ
Ar

(rAϑ ) − ∂Ar  .
1 α0  ∂
H= rot (A ) =  (10.4) ϑ
µ µr  ∂r ∂ϑ 

Inserting (10.3) into (10.4) we get Fig. 10.2

k 2Id  j 1  − jkr
H=  + e sin (ϑ )α 0 . (10.5)
4π  kr (kr )2 

The electric field follows from Maxwell’s first equation (1.12) using (13.89)

1  r0
E=
1
rot (H ) = 

[sin (ϑ ) H α ] − ϑ0 ∂ (rH α ) . (10.6)
jωε jωε  r sin (ϑ ) ∂ϑ r ∂r 

Inserting (10.5) into (10.6) we get the components of the electric field vector
122
book - 10
Idωµ  j 1  − jkr
Er = − j  + e cos(ϑ ) , (10.7)
λ  (kr ) (kr )3 
2

Idωµ 1 j 1  − jkr
Eϑ = j  − − e sin (ϑ ) , (10.8)
2λ  kr (kr ) (kr )3 
2

The vector of the magnetic field (10.5) has only one component

k 2Id  j 1  − jkr
Hα =  + 2
e sin (ϑ ) . (10.9)
4π  kr (kr ) 

The electric field has two components, Er and Eϑ , and the magnetic field has only one
component Hα. This is natural, as the lines of vector H must be of the shape of a circle drawn
around the conductor carrying an electric current.
Let us now discuss the character of formulas (10.7) to (10.9). The first fraction of
these formulas characterizes the source by its current and length, and the frequency and
material filling the space. The second term in rectangular brackets defines the dependence of
the radiated field on the distance in the form of the normalized distance r/λ as k = 2π/λ. The
third term determines the change of the field phase with distance. The last term describes the
dependence of the radiated field on angle ϑ.
8
The dependence on distance contains three 1
powers of product kr or relative distance r/λ: (kr )3
6
1 1 1
and . These functions are
(kr ) (kr )2 (kr )3 4
1 1
plotted in Fig. 10.3. They differ by the speed of
their increase when the argument tends to zero 2 (kr )
(kr )2
and by the speed of their decrease when the
argument increases to infinity. Consequently, we 0
can distinguish three areas around the dipole at 0 1 2 3
which the field has a particular behaviour. The kr
well known near field zone, which is very close Fig. 10.3
to the dipole, is defined by condition r/λ << 1.
Here field components with the term 1/(kr)3
prevail. This term is present in the formulas describing the electric field (10.7) and (10.8).
Assuming r/λ << 1, we can simplify (10.7) and (10.8) to

Idωµ 1
Er = cos(ϑ ) ,
jλ (kr )3
(10.10)
Idωµ 1
Eϑ = sin (ϑ ) .
2 jλ (kr )3

Inserting I = jωq into (10.10) we get the well-known distribution of the electrostatic field
excited by a dipole. For this reason, this zone is called the static zone. The magnetic field is
of negligible value in this zone.

123
book - 10
At slightly longer distances, term 1/(kr)2 starts to have an important value, and the
magnetic field must be taken into consideration. In this area it can be expressed as

k 2Id 1 − jkr Id sin (ϑ )


Hα = e sin (ϑ ) 0 ≈ , (10.11)
4π (kr )2
4πr 2

which, according to the Biot-Savart law, represents the magnetic field excited by the element
of a conductor with electric current I, d in length. This area is known as the inductive zone.
The most important zone is the far-field zone. This zone is defined by r >> λ. Only
terms 1/(kr) are of importance in the formulas describing the field distribution. The field
contains only two components

j µ e − jkr
Eϑ = Id sin (ϑ ) , (10.12)
2λ ε r

j e − jkr
Hα = Id sin (ϑ ) . (10.13)
2λ r

These components of the electric and magnetic fields are in phase, are mutually
perpendicular, and are perpendicular to the direction of propagation, which is determined by
unit vector r0. Their ratio

Eϑ µ
= = Zw .
Hα ε

equals the wave impedance of the free space. Thus the field has the character of a TEM plane
wave propagating in the free space. In fact the field is represented by a spherical wave
modified in its direction of propagation by function sin(ϑ). Only components Eϑ and Hα
participate in the transmission of power from the dipole. For this reason, the area is known as
the radiation zone.
Function sin(ϑ) determines the radiation pattern of the dipole. The radiation pattern
is generally a function dependent on two angles, ϑ and α, and is defined by

Eϑ (r = const ,ϑ ,α )
F (ϑ ,α ) = , (10.14)
Eϑ max

The radiation pattern is usually α = 0 z ϑ ϑ = 90°


plotted as its cuts in
characteristic planes. For our
dipole F (ϑ ,α ) = sin (ϑ ) , and
this radiation pattern is shown x z y
in Fig. 10.4. The radiation
pattern is omnidirectional in the
horizontal plane ϑ = 90°. x
The power radiated by
the dipole to the whole space is Fig. 10.4
calculated using Poynting’s
vector, calculated in the far field zone
124
book - 10
1  j 
S av =
1
2
( 1
) (
Re E × H * = Re Eϑ H α* r0 = Re 
2 2  2λ
) µ I d − jkr
ε r
e sin (ϑ )
− j I d − jkr
2λ r
e sin (ϑ ) r0 =

2
µ I d2
= sin 2 (ϑ ) r0
ε 8λ2 r 2
(10.15)

The total power is calculated integrating (10.15) over the surface of a sphere with radius r
much greater than the wavelength. An element of this surface in the spherical coordinate
system is dS = r2sin(ϑ)dαdϑ

sin 2 (ϑ ) 2
2ππ π
µ I 2d 2 µ I 2d 2 µ I 2d 2 4
∫ ∫ r 2 r sin (ϑ ) dϑdα = ε 4λ2 ∫0
( )
3
P= π sin ϑ d ϑ = π
ε 8λ2 00
ε 4λ 2 3

2
µ I 2 d 2π 40 I 2 d 2π 2 40 2  πd 
P= = = I   . (10.16)
ε 3λ2 ε r λ2 εr  λ 

The dipole radiating this power represents a load for the source feeding it. So we can
substitute the dipole by a radiating resistor with resistance Rr. This is a resistor in which the
same power is consumed. The power is

1
P= Rr I 2 ,
2

so we get

2
2P µ d 2 2π d 
Rr = 2 = = 80 π 2   . (10.17)
I ε 3λ 2
λ

As d << λ, the value of Rr is very small and it is practically impossible to match this
elementary dipole to any generator.

____________________________________
Example 10.1: Calculate the power radiated by a dipole radiating into a space filled with air,
and also the radiation resistor of this dipole. The dipole is d = 2 m in length. The passing
current has amplitude 3 A and frequency 1 MHz.
The radiating power is

2 2

40 2  πd
40 2  πdf 
P= I   = I   = 0.16 W
εr  λ  εr  c 

From the radiated power and the passing current we get

2P
Rr = = 0.035 Ω
I2

125
book - 10
____________________________________
Example 10.2: What power must be radiated by the electric dipole to get the value of an
electric field Em = 25 mV/m at the distance from the antenna r = 100 km determined by angle
ϑ = 90°. The antenna is much shorter than the wavelength, which is 450 m. The space is
filled with air.
We are at a distance r=105 >> λ = 450, so we are in the far field zone. The value of the
electric field is here, assuming ϑ = 90°,

1 µ0 1
Eϑ = E m = Id ,
2λ ε0 r

and from this formula we get the product Id characterizing the antenna

E m 2λr
Id = = 5864 Am .
µ0
ε0

The radiated power is (10.16)

2
 πId 
P = 40   = 69.5 kW.
 λ 
________________________________________
z
dS
I
10.2 Elementary magnetic dipole
An elementary magnetic dipole is represented by a
a y
current loop with radius a and passing current I. The magnetic
dipole is shown in Fig. 10.5. The radius of this loop is much x
lower than the wavelength, and consequently we can assume
that the current has a constant amplitude and phase along the Fig. 10.5
loop. The moment of this dipole is

m = I dS , (10.18)

where dS = π a 2 . The excited field has rotational symmetry, so we can look for the field
distribution only in the plane xz. The current densities in the loop at any position x can be
decomposed into two components. The x components have opposite directions, so the field
excited by them vanishes. The radiated field is thus excited by only the y components of the
current density, Fig. 10.6a. Using (1.58), we get the vector potential describing the field
excited at point P, Fig. 10.6b, in the form

J (Q ) e
− jkrQP
µ
A(P ) = ∫∫∫ d VQ
4π V
rQP

126
book - 10
P
y J z ϑ rQP
Jy = Jcos(α)
Jx = Jsin(α) Q r rcos(ϑ)
Q y
α
a
Jy x α
O
acos(α) rsin(ϑ) X x
Jx

a b
Fig. 10.6

where dVQ = S c a dα , Sc is the cross-section of the loop conductor. The vector potential has
the direction of the exciting current. This is the y component in the plane xz, but in the whole
space it is the α component. Taking into account only the y component Jy = Jcos(α) of the
current density and assuming that the integral of J over cross-section Sc equals current I,
which is constant along the loop, we get

2π − jkrQP 2π − jkr
µ e µ e QP
Aα (P ) = ∫ ∫∫ J cos(α )dS c a dα = a∫ dα ∫∫ J cos(α )dS c =
4π 0 Sc
rQP 4π 0 rQP S c
(10.19)
2π − jkrQP π − jkrQP
µ e µ e
= aI ∫ cos(α )dα = aI∫ cos(α )dα
4π 0
rQP 2π 0
rQP

The result of the particular integration in (10.19) is precise. The final integral cannot be
simply solved, this can be done only numerically. To get an analytical result, we have to
simplify the expression for distance rQP, Fig. 10.6b. We assume r >> a.

rQP = (QX )2 + ( XP )2 = [a 2
]
+ (OX )2 − 2a(OX ) cos(α ) + [r cos ϑ ]2 =

= a 2 + r 2 sin 2 (ϑ ) − 2ar sin (ϑ ) cos(α ) + r 2 cos 2 (ϑ ) = a 2 + r 2 − 2ar sin (ϑ ) cos(α ) ≈


a  a 
≈ r 2 − 2ar sin (ϑ ) cos(α ) = r 1 − 2 sin (ϑ ) cos(α ) = r 1 − sin (ϑ ) cos(α )
r  r 

rQP = r − a sin (ϑ ) cos(α ) . (10.20)

Assuming that cos[ka sin (ϑ ) cos(α )] ≈ 1 and sin[ka sin (ϑ ) cos(α )] ≈ ka sin (ϑ ) cos(α ) as a << λ
we can simplify

= e − jkr e jka sin (ϑ ) cos (α ) = e − jkr {cos[ka sin (ϑ ) cos(α )] + j sin[ka sin (ϑ ) cos(α )]} ≈
− jkrQP
e
,
= e − jkr [1 + jka sin (ϑ ) cos(α )]

127
book - 10
1 1 1 a  1 a
= ≈ 1 + sin (ϑ ) cos(α ) = + 2 sin (ϑ ) cos(α ) .
rQP  a  r r  r r
r 1 − sin (ϑ ) cos(α )
 r 

Now we can simply express (10.19) as

π − jkr
µ e QP
Aα (P ) = aI∫ cos(α )dα =
2π 0
rQP
π
µ 1 a 
= a I ∫ e − jkr [1 + jka sin (ϑ ) cos(α )] + 2 sin (ϑ ) cos(α ) cos(α ) dα
2π 0 r r 

After some manipulations and after integration we get

µ a 2 I  jk 1  − jkr
Aα (P ) =  + 2 e sin (ϑ ) . (10.21)
4  r r 

From this distribution of the vector potential we calculate the magnetic field using
H = rot (A ) µ , and from the magnetic field we calculate the electric field using
E = rot (H ) ( jωε ) . The procedure is the same as that applied in the case of the electric dipole.
Operator rot must be used in the spherical coordinate system (13.89). As a result we obtain
two components of the magnetic field Hr and Hϑ and one component of the electric field Eα

π (ka )2 I  j 1  − jkr
Hr =  + e cos(ϑ ) , (10.22)
λ  (ka ) (ka )3 
2

π (ka )2 I 1 j 1  − jkr
Hϑ = −  − − e sin (ϑ ) , (10.23)
2λ  kr (kr ) (kr )3 
2

ωµ (ka )2 I  1 j  − jkr
Eα =  − 2
e sin (ϑ ) . (10.24)
4  kr (kr ) 

Comparing these formulas with (10.7), (10.8) and (10.9) we can see that the field
excited by the current loop is dual to the field excited by the electric dipole. For this reason
we will not repeat here the discussion from the previous section. It is obvious that, due to this
duality, the field excited by the magnetic dipole can be obtained from the formulas (10.7),
(10.8) and (10.9), using the transformation: H is exchanged by –E, E is exchanged by H, ε is
exchanged by µ, and the dipole moment p = Id/(jω) is exchanged by µm = µπa2I.

_________________________________
Example 10.3: Compare the power radiated by the electric dipole to the power radiated by
the magnetic dipole. Assume the same current and the same length of the conductor. This
means that the length of electric dipole d equals the perimeter of the current loop, i.e., d =
2πa.
The power radiated by the electric dipole is determined by (10.16). The power radiated
by the magnetic dipole can be calculated in the same way that (10.16) was derived. Another
128
book - 10
way is to use the above mentioned duality and transformation. To use this transformation we
have to modify (10.16) to

µ I 2 d 2π p 2ω 4
Pe = = .
ε 3λ2 12πε c 3

Inserting here µm instead of p, µ instead of ε and using d = 2πa we get the power radiated by
the magnetic dipole

Pm = =
( )
ω 4 µ 2 m 2 ω 4 I 2 πa 2 µ
.
2

12πµ c 3 12π c 3

Now we can express the ratio of these powers

2
Pe I 2 d 2ω 2 12π c 3 λ 
= =  .
Pm 12πε c ω I π a µ  πa 
3 4 2 2 4

As λ >> a, the power radiated by an electric dipole is much greater than the power radiated by
a magnetic dipole. For this reason, only an electric dipole is used as a transmitting antenna.
The radiation of the magnetic dipole has very low efficiency. This antenna is used only rarely
and only as a receiving antenna.

_______________________________________

10.3 Radiation of sources with dimensions comparable with the wavelength


In previous sections we calculated the field radiated by electric and magnetic dipoles.
We assumed that these dipoles have dimensions much smaller than the wavelength, and the
current density is therefore constant within the volume of the dipole. This assumption
simplified the calculation of the integral in (1.58). In the case of sources with dimensions
comparable with the wavelength we cannot neglect the current variation, and we have to
consider that distance R in (1.58) varies within the volume of the source and is different from
the radius vector, which defines the position of the point at which the field is calculated.
Practical antenna problems are very complicated, as we do not know the current distribution
in advance. This distribution however depends on the radiated field via the boundary
conditions valid on the antenna surface. The radiated field is determined by the current
distribution. This vicious circle must be overcome by a proper numerical method, which
enables us to calculate the unknown current distribution applying the proper boundary
conditions. Such a problem is however beyond the scope of this textbook.
Let us turn our attention to the calculation of the radiated field in the far field zone,
and let us assume the known distribution of a current passing our source. We are looking for
the field at point P in the far field zone determined by radius vector r, Fig.10.7. The source
has volume V. Let us put an element of the current passing the source volume at point Q
defined by radius vector rQ, Fig. 10.7. The dimensions of the source are comparable with the
wavelength. Assuming that point P is in the far field zone, vectors r and rQP can be
considered to be parallel. Thus we can express the distance between points Q and P as

129
book - 10
rQP ≈ r − rQ cos(α ) , P
(10.25)
z r
and we can assume an even
coarser approximation in the
rQcos(α)
denominator of the expression
for a vector potential
0 α rQP
1 1 y
≈ . J
rQP r rQ
x Q
V
Using (1.58) we get the vector
potential at point P
Fig. 10.7

J (Q ) e
− jkrQP
µ µ e − jkr jkrQ cos (α )
A=
4π ∫∫∫ rQP
dV ≈
4π r ∫∫∫ J(Q)e dV ,
V V

Let us denote a radiation vector

jkrQ cos (α )
N = ∫∫∫ J (Q ) e dV , (10.26)
V

which defines the properties of the source. Then the vector potential is

− jkr
µ e
A= N . (10.27)
4π r

The vector of the electric field is calculated from the vector and scalar potentials by (1.52).
The scalar potential is coupled with the vector potential by the Lorentz calibration condition
(1.53). Inserting this condition into (1.52) we get

j
E = − jω A − grad div(A ) . (10.28)
ωµε

The vector potential (10.27) is inserted into (10.28), taking into account that radiation vector
N does not depend on r. We get

1  ω µ − jkr
E=  −j e (Nϑϑ0 + Nα α 0 ) + 12 [ L ] + L , (10.29)
r 4π  r

Nϑ and Nα are the components of N in the spherical coordinate system. We are looking for the
field in the far field zone. In this zone only functions with the dependence on distance r of the
order of –1 can be taken into account, and the value of the terms of higher orders can be
neglected. This means that only the first term in (10.29) can be taken. This gives us a
physically reasonable result. The electric field has the form of a spherical wave with a zero
longitudinal component and two nonzero transversal components

130
book - 10
Er = 0 ,

ω µ e − jkr Z e − jkr
Eϑ = − j Nϑ = − j Nϑ , (10.30)
4π r 2λ r

ω µ e − jkr Z e − jkr
Eα = − j Nα = − j Nα .
4π r 2λ r

As this spherical wave can be locally approximated by a plane TEM wave, we can determine
the magnetic field using the wave impedance of the free space (2.26)

Eϑ Eα
Hα = , Hϑ = − , (10.31)
Zw Zw

The field in the far field zone has locally the character of a TEM wave with elliptical
polarization, because the particular transversal components Eϑ, and Eα, have generally
different amplitudes and phase.
z
rQP
_________________________________ P
Example 10.4: Calculate the electric field radiated by a
symmetric dipole of finite length, Fig. 10.8, located at
the origin of the coordinate system. Current passing r
L Q
dipole conductors is in the form of a standing wave. ϑ
Let us first calculate the radiation vector, which
has the direction of the passing current. Using (10.26) x y
we get
L
L L I(z)
jkrQ cos (α ) jkz cos (ϑ )
N = z 0 ∫ ∫∫ J z e dSdz = z 0 ∫ I ( z ) e dz
−L S −L

The dipole is represented by a transmission line with an Fig. 10.8


open end termination at both ends. So the current must
fulfill here the boundary condition I(-L) = I(L) = 0, which gives a current distribution in the
form of a standing wave

I = I m sin [k (L − z )] .

The ϑ component of the radiation vector is Nϑ = -Nzsin(ϑ). The electric field has the only
component

Z e − jkr Z e − jkr
Eϑ = − j Nϑ = j N z sin (ϑ ) =
2λ r 2λ r
Z e − jkr  0 L 
= j sin (ϑ ) ∫ sin [k (L + z )]e jkz cos (ϑ ) dz + ∫ sin[k (L − z )]e jkz cos (ϑ ) dz 
2λ r − L 0 

After some rearrangements, this formula reads

131
book - 101
e − jkr cos[kL cos(ϑ )] − cos(kL )
Eϑ = j 60 I m . (10.32)
r sin (ϑ )

The dipole radiation pattern is

cos[kL cos(ϑ )] − cos(kL )


F (ϑ , α ) = . (10.33)
sin (ϑ )

1
For a very short dipole (L << λ) cos(kL ) ≈ 1 − (kL )2 , cos[kL cos(ϑ )] ≈ 1 − 1 (kL )2 cos 2 (ϑ ) ,
2 2
and the radiation pattern is described by the simplified formula

F (ϑ , α ) =
(kL )2 sin (ϑ ) ,
2

which is the radiation pattern of the elemental electric dipole. Fig. 10.9 shows the radiation
patterns of dipoles with particular lengths: L → 0, L = λ / 4 (kL = π/2), L = λ/2 (kL = π), L =
λ3/4 (kL = π3/2), . L = λ (kL = 2π), L = 2λ (kL = 4π). Increasing further the dipole length we

L→0 L = λ/4 L = λ/2

L = λ3/4 L=λ

L = 2λ

Fig. 10.9

finally get the radiation pattern with two main beams directed along the conductors. This tells
us that the power is transmitted only along the conductors.

_______________________________________
132
book - 101
The power radiated by our source into the whole space is calculated from Poynting’s
vector

2 2
Eϑ + Eα
S av
1
2
1
2
(
= Re(E × H ) = r0 Eϑ H α* − Eα H ϑ* = r0) 240π
= S rav r0 . (10.34)

Integrating (10.34) over the surface of a sphere with radius r >> λ we get the radiated power

π 2π
Pr = ∫ ∫ S rav r 2 sin (ϑ ) dϑ dα . (10.35)
0 0

Quantity W = Sravr2 is known as the radiation intensity. It represents the power radiated into
a unit space angle.

10.4 Antenna parameters


We have already spoken about the antenna radiation pattern (10.14). The radiation
pattern determines the dependence of the radiated field on directional angles ϑ and α. It is
usually normalized to the maximum of the radiated field. It is generally a complex function.
Most often we use the amplitude radiation pattern, but the phase radiation pattern can be
also applied. The power radiation pattern corresponds to the second power of (10.14). The
amplitude radiation pattern is expressed in absolute units or in dB. Fig. 10.10 shows the
radiation pattern of
0
an antenna plotted
F (dB)

in dB normalized to FWHP -3 dB
0. It consists of the -5
main lobe and a SLL ϑ = 0°
number of side -1 0
lobes. We are in
most cases
-1 5
interested in the
main lobe, at which
the maximum of -2 0
power is radiated.
The main lobe is -2 5
determined by its
width, usually
-3 0
defined as the width
measured at the
level –3 dB below -3 5

the main maximum -1 8 0 -1 5 0 -1 2 0 -9 0 -6 0 -3 0 0 30 60 90 120 150 180

of the electric field α (° )


Fig. 10.10
amplitude, i.e., at
one half of the maximum power – the full width at half power (FWHP). We are also
interested in the direction of the main lobe. Next we define the side lobe level (SLL),
usually in dB, which determines the separation of the maximum lobe of the side lobes from
the level of the field in the main lobe.
The radiation pattern defines the antenna directivity. The maximum directivity is
defined as the ratio of the maximum of the radiation intensity and its average value

133
book - 101
Wmax S rav max r 2 4πr 2 S rav max 4π
Dmax = = = π 2π = π 2π
=
Wav Pr S rav
r ∫ ∫ S rav sin (ϑ )dϑ dα
2
∫ ∫ S rav max sin (ϑ )dϑ dα

0 0 0 0
, (10.36)

= π 2π

∫ ∫ F (ϑ ,α ) sin (ϑ )dϑ dα
2

0 0

The directivity in the general direction then is

D = Dmax F (ϑ , α ) .
2
(10.37)

The antenna directivity evaluates only the directional distribution of the energy flow. It does
not define the efficiency of the radiation η. This efficiency is included into an antenna gain.
The antenna gain incorporates the efficiency of the conversion of the power supplied from the
source feeding the antenna Ps to radiated power Pr. The gain is defined as

Wmax 4πWmax
G= = = ηDmax . (10.38)
Ps Pr
4π η

Then antenna input impedance is defined as

UA
ZA = , (10.39)
IA

where UA and IA are the voltage and the current at the antenna terminals. It is usually difficult
to determine the current distribution over the antenna that defines its input current. In many
cases we thus substitute the real part of the antenna input impedance by the radiating
resistance (10.17). It is

π 2π
2 Pr 2r 2
Rr =
IA
2
=
IA
2 ∫ ∫ S r sin (ϑ ) dϑ dα . (10.40)
0 0

10.5 Antenna arrays


We have studied up to now only single radiators. The field radiated by these radiators
is calculated as the superposition of the spherical waves coming out from partial elements of
the radiator. These particular waves depend on the local amplitude and phase of the current
passing this element. The possible ways of modifying the antenna radiation pattern and gain
are therefore limited, as we cannot simply change the current distribution over the antenna
body. This is the reason why antenna arrays are used. Such an array consists of a number of
partial radiating elements. Each element has its own feeding source. Now we have huge
possibilities to change the array radiation pattern and its gain. This can be done by changing
the amplitudes and phases of the feeding currents and changing the positions of these
elements. Antenna arrays can be one dimensional, two dimensional or three dimensional.
134
book - 101
Generally, when we increase the number of particular elements we reduce the width of the
radiation pattern.
Let us show the tremendous possibilities of antenna arrays through the following
simple case, Fig. 10.11. There are two dipoles of total length λ/2 located in parallel with the z
axis at distance d symmetrically to the origin, as shown in Fig. 10.11. These dipoles are fed
by different currents I1 and I2. The field radiated by a particular dipole was calculated in
Example 10.4 (10.32). The field radiated by the couple of dipoles is the superposition of the
particular fields (10.32), where kL = π/2

π 
cos  cos(ϑ )
  I e 1 + I e 2 
− jkr − jkr
Eϑ = Eϑ1 + Eϑ 2 = j 60 2  ,
sin (ϑ )  1 r 2
r2 
 1 
P
assuming that the mutual coupling
between the dipoles is neglected.
Looking for the field in the far field r1 r r2
zone, we can express distances r1 and
z
r2, using r as in (10.20)

d
r1 = r + sin (ϑ ) cos(α ) , ϑ
2

d 0
r2 = r − sin (ϑ ) cos(α ) . λ/2
2 I1 I2 y
d
Consequently we get the field
radiated by the couple of dipoles in x α
the form
Fig. 10.11

e − jkr  − jk d sin (ϑ )cos (α ) jk sin (ϑ ) cos (α ) 


d
e − jkr
Eϑ = j 60 F1 (ϑ )  I1 e 2 + I1 e 2  = j 60 F1 (ϑ ) F2 (ϑ ,α ) ,
r   r
(10.41)

where F1 is the radiation pattern of one dipole and F2 is a function known as the array factor.
Expression (10.41) can be used to describe the radiated field in the far field zone for any array
of equal radiators. We get the radiation pattern of such an array in the simple form of a
product

F (ϑ , α ) = F1 (ϑ ,α ) F2 (ϑ , α ) . (10.42)

Let us return to our problem. (10.41) can be simplified assuming currents of the same
amplitude I0 and symmetric phase

I 1 = I 0 e jψ , I 2 = I 0 e − jψ .

Now the array factor is


135
book - 102
 kd 
F2 (ϑ , α ) = 2 I 0 cos  sin (ϑ ) cos(α ) − ψ  . (10.43)
2 

This function is plotted for several values of d and ψ in Fig. 10.12 in the plane xy, i.e., for
ϑ = π 2 which gives F1 = 1. Fig. 10.12 thus shows the radiation pattern of the couple of
dipoles in the plane xy. It is evident that we have really tremendous possibilities in changing
the shape of the array radiation pattern even in the case of two radiators.

d = λ/8 d = λ/4 d = λ3/8 d = λ/2 d = λ3/4 d=λ

2ψ = 0°

2ψ = 45°

2ψ = 90°

2ψ = 135°

2ψ = 180°

Fig. 10.12

______________________________
Example 10.5: Calculate the
radiation pattern of the array of N r rm
dipoles λ/2 in length in the plane
xy. The dipoles are parallel with mkdcos(α)
the z axis. Their centers are
equidistantly located on the x
axis with separation d = λ/2, Fig. α
10.13. The dipoles are fed by
equal current I0. 0 1 2 d m N-1 x
The dipoles radiate in the
xy plane homogeneously in all Fig. 10.13
136
book - 102
directions, so F1(π/2,α) = 1. Using Fig. 10.13 and details of calculation from the derivation of
(10.41), we get

N −1
e − jkrm e − jkr N −1
e − jkr
Eϑ = ∑ A I0 rm
= A I0
r
∑ e jkmd cos(α ) = A I 0 r
F2 (α ) .
m=0 m =0

Substituting q = e jkmd cos (α ) we can read the expression for the array factor normalized to 1 as
a geometric sequence of complex terms qm

N 
sin  kd cos(α )
N N
N −1 −
F2 N
1 1 q −1 1 q2 −q 2 1 2 
= 1 + q 2 + q3 + L q N = = q 2 =
N N N q −1 N q 12
− q −1 2 N 1 
sin  kd cos(α )
2 
 Nπ 
sin  cos(α )
1  4 
=
N π 
sin  cos(α )
4 

The array factors


computed for N = 1.2
 F2

1,2,3,4,10 are plotted in


Fig. 10.14. These 1
N=1
functions are normalized
by dividing their values 0.8
by N. For N = 1 the N=4 N=2
radiation goes
0.6
homogeneously in all N = 10 N=3
directions – one dipole.
0.4
Increasing the number of
radiators, we increase the
amplitude of the main 0.2
lobe of F2 at α = 0, and
side lobes appear, as was 0
mentioned above. The 0 30 60 90 120 150 180
width of the main beam α (°)
decreases with the Fig. 10.14
number of dipoles.

_______________________________________

10.6 Receiving antennas


In the preceding paragraphs we studied the particular sources of an electromagnetic
field which are known as transmitting antennas. Here we show how to calculate the power
received by an antenna. We define the effective length of an antenna and its equivalent circuit.
Let us assume an antenna irradiated by a plane TEM wave, Fig. 10.15. Impedance Zp
represents the input impedance of a receiver. The wave described by vectors Ei, Hi and by
propagation vector ki is incident to the conducting surface of the antenna. Let us assume that
137
book - 102
this wave is not distorted by this incidence. Consequently it does not fulfill the boundary
conditions on the antenna surface. To meet these conditions and to account for the distortion
of the field we have to add the field known as scattered, with vectors Er and Hr. Now the total
field

E = Ei + E r ,
n
H = Hi + Hr , Ei

must fulfill the boundary condition, and it models the ki


precise distortion of the field by the antenna. Note
that the scattered field must vanish far from the Zp Hi
antenna, as the field here must correspond only to the
incident wave. We will calculate the power received
by the antenna and fed to the input of the receiver as
the total power which crosses closed surface S, fully S
covering the antenna. Let us assume that this surface
closely copies the antenna conducting surface. This Fig. 10.15
power is

P=
1
∫∫
2 S
( )1
[
E × H * ⋅ dS = ∫∫ (E i + E r ) × H *i + H *r ⋅ dS =
2 S
( )]
, (10.44)
1
[ 1
] ( 1
)
= ∫∫ (E i + E r ) × H *i ⋅ dS + ∫∫ E i × H *r ⋅ dS + ∫∫ E r × H *r ⋅ dS
2 S 2 S 2 S
( )
In the first integral we can reorder vectors in the mixed product and we can get

[(E i ] [ ]
+ E r ) × H *i ⋅ dS = (E i + E r ) × H *i ⋅ n dS = [n × (E i + E r )] ⋅ H *i dS = 0

due to the fact that n × (E i + E r ) = 0 , as this vector product determines the tangent component
of the total electric field to the conducting surface. The last integral in (10.44) represents the
power radiated by the antenna back in the form of a scattered field. This power can be
represented as

1
∫∫
2 S
( )1
E r × H *r ⋅ dS = Z A I A
2
2
,

where ZA is the antenna input impedance and IA is the antenna input current. The second
integral in (10.44) can be rewritten

1
∫∫
2 S
( ) 1
2 S
( 1
2
) 1
E i × H *r ⋅ dS = ∫∫ E i × H *r ⋅ n dS = − E i ⋅ ∫∫ n × H *r dS = − − E i ⋅ ∫∫ K *A dS =
2
S S
,
1 1
= − E i ⋅ i 0 ∫ i *A dl = − E i ⋅ d eff I *A
2 d
2

138
book - 102
where i0 is a unit vector in the direction along the antenna, IA is the current passing along the
antenna, d is its length and deff represents the antenna effective length. This quantity is
defined as the length which, multiplied by the value of the incident electric field, offers the
internal voltage received by the antenna. The received power is

1 1 2 1
P = U A I *A = I A Z A − E i ⋅ d eff I *A , (10.45)
2 2 2

and the voltage induced on the antenna terminals is

U A = Z A I A − E i ⋅ d eff . (10.46)

(10.45) and (10.46) define the receiving antenna equivalent circuit, Fig. 10.16. The input
impedance of antenna ZA represents the back radiation which in the receiving antenna is
considered as loss.
Similarly as the antenna effective length we can define the antenna effective
aperture by

PR = S av Aeff , (10.46)

ZA IA
where PR is the power supplied by the antenna
to its load, and Sav is the magnitude of
Poynting’s vector of the incident plane
electromagnetic wave. Zp E i ⋅ d eff
UA
_____________________________________
Example 10.6: Calculate the effective aperture
of the elementary electric dipole d in length,
located in air. Assume that the dipole is Fig. 10.16
matched to the load, Fig. 10.16.
According to Fig. 10.16, the current supplied by the receiving antenna to the input
impedance of the receiver and the corresponding power are

U 1 2
IA = , PR = I A Rr ,
ZA + Zp 2

where Rr is radiating resistance of the dipole (10.17), which represents the real part of antenna
impedance ZA, and U = Eincd is the voltage received by the antenna. Assuming the impedance
matching Z p = Z *A and the current is

U
IA = .
2 Rr
Then the received power is

2
1 U E2 d 2
PR = = inc .
2 4 Rr 8 Rr

139
book - 102
2 2
E inc Einc
From (10.46), inserting for the received power, for Poynting’s vector S av = = and
2 Z 0 240π
for the radiating resistance (10.17), we get the effective aperture

2
Einc d 2 240π d 2 30π 3
Aeff = 2
= = λ2 .
2
8 Rr Einc d 8
80π 2  
λ

Using (10.36) we can calculate the directivity of the elementary electric dipole

4π 4π 3
Dmax = π 2π
= π
= .
2
∫ ∫ F (ϑ ,α ) sin (ϑ )dϑ dα 2π ∫ sin 3 (ϑ )dϑ
2

0 0 0

Combining this result for the directivity with the formula for the effective aperture we get

λ2
Aeff = D .

Although this formula has been derived for the elementary electric dipole, it can be applied
for the wide class of antennas.
_________________________________________

10.7 Problems
10.1 A short conductor of length l radiates an electromagnetic wave at the frequency f = 100
kHz, Fig. 10.17. At the distance r1 = 1000 m S
from this antenna we can measure the electric
δ
field amplitude Em = 10 V/m. There is a l
rectangular conducting loop with the surface S = I r1 r2
2
0.5 m located at the distance r2 = 1200 m from S
the antenna. Calculate the angle δ at which the I
loop receives the maximum voltage, and also the
value of this voltage. r1
δ = 0°
U = 0.0105 V Fig. 10.17

10.2 Calculate the average value of the power density radiated by an electric dipole at the
distance r = 20 km in the directions determined by ϑ = 0°, 45°, 90°, 135°, 180°. The
amplitude of the passing current is Im = 50 A, the frequency is f = 500 kHz.
ϑ = 0°, 180° S av = 0 W/m2
ϑ = 45°, 135° S av = 2.04 ⋅ 10 −8 W/m2
ϑ = 90° S av = 4.08 ⋅ 10 −8 W/m2

10.3 A conductor 50 km in length conducts a current with the amplitude Im = 100 A. The
frequency is f = 50 Hz. Calculate the power radiated into the whole space.

140
book - 102
P = 274 W

10.4 Calculate the power radiated to the whole space by a circular loop of radius a = 5 cm.
The passing current has the amplitude Im = 5 A, and the frequency is f = 500 kHz. Calculate
the radiating resistance of this radiator.
P = 0.22 10-10 W
Rr = 1.76 10-12 Ω

11. WAVE PROPAGATION IN NON-ISOTROPIC MEDIA


We have studied the propagation of electromagnetic waves in isotropic materials in
previous chapters. The relations between the field vectors in these materials are described by
(1.9), (1.10) and (1.11), where conductivity σ, permeability µ, and permittivity ε are scalar
quantities. In non-isotropic materials these formulas have the form (1.26)

D=ε E , (11.1)

B=µH . (11.2)

Vectors D, E, and B, H are not parallel (1.27). Tensors of permittivity (1.28) and
permeability can be expressed

ε xx ε xy ε xz 
 
ε = ε yx ε yy ε yz  , (11.3)
ε zx ε zy ε zz 

 µ xx µ xy µ xz 
 
µ =  µ yx µ yy µ yz  . (11.4)
 µ zx µ zy µ zz 

Taking into account the non-isotropy of the materials we are able to clarify a number
of effects occurring in nature. The number of elements applied in microwave technology is
based on the effect of propagation in a non-isotropic material. Such elements are called non-
reciprocal, as a wave propagates through them differently in different directions. Non-
isotropic behaviour of materials can be either natural or induced. Some mono-crystals in
which, e.g., birefringence (double refraction) known from optics, can be observed are
materials with natural non-isotropic behaviour. Plasma (ionized gas) and ferrite material are
examples with induced non-isotropic behaviour. The non-isotropy is induced in these
materials by applying an external magnetic field. Plasma occurs in the upper layers of then
atmosphere, and the non-isotropy is induced by the earth’s magnetic field. We will pay
attention to ferrite materials, e.g., they are widely used in microwave technology in
production of non-reciprocal devices, as isolators, gyrators, hybrids, phase shifters, etc., or
ferrite resonators. Ferrites are mixtures of oxides of ferromagnetic materials (Fe, Ni, Cd, Li,
Mg) sintered at high temperatures. Ferrites must have very low conductivity (10-4 – 10-6 S/m)
to allow an electromagnetic field to propagate through them and to interact with them.

141
book - 11
11.1 Tensor of permeability of a magnetized ferrite
Any medium consists of a system of atoms. Each atom consists of a positively charged
core and a number of electrons bearing a negative charge. These electrons move along certain
orbital paths, and at the same time spin around their axes. This movement and rotation of the
charged electrons represents the flow of an electric current. The flowing current excites a
magnetic field which is perpendicular to the plane of the current loop. Consequently we can
define the orbital magnetic moment and the spin magnetic moment of the electron – mo
and msp. The electron has a certain mass. Its movement causes the orbital mechanical
moment Lo and the spin mechanical moment Lsp. These moments are coupled by the
relations

e
m o = −µ 0 Lo , (11.5)
2m

e
m sp = − µ 0 L sp = −γ L sp , (11.6)
m

where e and m are the charge and the mass of the electron, and γ = µ 0 e m is known as the
gyromagnetic constant. The total moment of the whole
atom is the sum of these particular moments. The Ho z
moments of the core are negligible in comparison with
the moment of the electron. It has been experimentally
proved that the orbital moments are of much lower
value than the spin moments. Therefore we will treat msp
only the spin moments.
Let us now study the behaviour of an electron
with magnetic spin moment msp and with mechanical
spin moment Lsp located in the external magnetic field
parallel to the z axis Ho = Hoz0 which is not parallel to electron y
msp. Under the external field each single magnetic
dipole in the medium rotates. This movement is known
as precession. The end points of vectors msp and Lsp Lsp
move along a circle, Fig. 11.1. Torque T affecting the x
dipole is
Fig. 11.1
e
T = m sp × H o = − µ 0 L sp × H o .
m

It is known from mechanics that the change of Lsp due to T is

dL sp
= T = m sp × H o .
dt

Inserting for Lsp from (11.6) we get

dm sp e
= −µ 0 m sp × H o = −γ m sp × H o . (11.7)
dt m

This vector relation can be rewritten into three scalar equations

142
book - 11
dm spx
+ γ H o mspy = 0 , (11.8)
dt

dm spy
− γ H o m spx = 0 , (11.9)
dt

dm spz
=0 . (11.10)
dt

The solution of (11.8) and (11.9) describes the precession movement of the end point of msp

m spx = m∞ cos(ω o t ) , (11.11)

m spy = m∞ sin (ω o t ) , (11.12)

where ω o = γ H o is known as the frequency of free precession – the Larmor precession


frequency. This frequency is proportional to the magnetizing field. In the case of a ferrite
without losses the precession is not damped, and it would last to infinity. In a real lossy
medium, the end point of msp moves along a spiral and it finally approaches the direction of
the external magnetic field. Instead of the magnetic dipole we can follow the vector of
magnetization as for a unit volume including N particles with the magnetic moments
organized by the external magnetic field into the same direction, and the magnetization is M
= N msp. Equation (11.7) now reads

dM
= −γ M × H o . (11.13)
dt

The solution of this equation is

M x = M ∞ cos(ω o t ) , (11.14)

M y = M ∞ sin (ω o t ) . (11.15)

Let us assume that the ferrite material is exposed to the superposition of a DC


magnetic field Ho and a high frequency field Hm, the amplitude of which is Hm << Ho. The
directions of these fields are different

H c = H o + H m e j ωt = H o z 0 + H m e j ω t . (11.16)

The total magnetic field thus has a time varying value and a time varying direction.
Consequently the precession movement of the magnetization does not have a constant
direction but will follow the varying magnetic field, and its frequency will be equal to the
frequency of the varying magnetic field. This is know as forced precession. Similarly as the
magnetic field we can express the magnetization

M c = M o + M m e j ωt = M o z 0 + M m e j ω t , (11.17)

143
book - 11
where Mm << Mo. Now we insert the total fields (11.16) and (11.17) into (11.13) and we get

jω M m e j ω t = −γ M c × H c =
( )
= −γ M o × H m e j ω t + M m × H o e j ω t + M m × H m e 2 j ω t + M o × H m e j ω t + M o × H o ≈
≈ −γ (z 0 × H m M oe jω t + M m × z 0 H oe jω t )
jω M m = −γ M o z 0 × H m + γ H o z 0 × M m .

This vector equation can be expressed using vector components in the following three
equations

jω M mx = ω M H my − ω o M my ,

jω M my = −ω M H mx + ω o M mx , (11.18)

jω M mz = 0 ,

where ω M = γ M o is the magnetizing constant. Solving the set of equations (11.18) we get

ω oω M ω ωM
M mx = − H mx − j H my ,
ω 2 − ω o2 ω 2 − ω o2

ω ωM ω ω
M mx = j 2 2
H mx − 2o M 2 H my , (11.19)
ω − ωo ω − ωo

M mz = 0 .

The magnetic induction of the high frequency field is

B m = µ 0 (H m + M m ) .

Particular components of Bm using (11.19) are read

 ωω  ωω
Bmx = µ 0 1 − 2o M 2  H mx − jµ 0 2 M 2 H my ,
 ω − ωo  ω − ωo

ω ωM  ω ω 
Bmx = jµ 0 2 2
H mx + µ 0 1 − 2o M 2  H my , (11.20)
ω − ωo  ω − ωo 

Bmz = µ 0 H mz .

Consequently we have the tensor of permeability

144
book - 11
 µ − jµ a 0
µ =  jµ a µ 0  , (11.21)
 0 0 µ 0 

where

 ω o ω M 
µ = µ 0 1 − , (11.22)
 ω 2 − ω o2 

ω ωM
µa = µ0 . (11.23)
ω 2 − ω o2

It is important to remember that the tensor of permeability (11.21) describes the behaviour of
the ferrite material under magnetization by the field (11.16). The DC field has the direction
parallel to the z axis and the tensor of permeability (11.21) couples together the induction and
the intensity of the high frequency magnetic field

Bm = µ Hm . (11.24)

We noted that non-isotropic behaviour is also displayed by plasma magnetized by a


DC magnetic field directed parallel to the z axis. In this case we can express the tensor of
permittivity as

 ε − jε a 0
ε =  jε a ε 0  , (11.25)
 0 0 ε 0 

and the vectors of the high frequency field are coupled by

Dm = ε Em . (11.26)

_______________________________________
Example 11.1: Calculate magnetic field Ho to magnetize a ferrite to magnetic resonance at
the frequency fr = 15 GHz. The mass of an electron is m = 9.109.10-31 kg, the charge of an
electron is e = 1.602.10-19 C.
The gyromagnetic constant is

e
γ = µ0 = 2.21 ⋅ 10 5 m/C .
m

The frequency of the free precession is

ωo
ωo = γ Ho => Ho = = 4.26 ⋅ 10 5 A/m
γ
_______________________________________

145
book - 11
Example 11.2: Calculate the tensor of permeability of a ferrite material with ωο = 1.105 1010
s-1 and ωM = 1.547 1010 s-1 (see Problem 11.1) at the frequency f = 300 MHz.
Applying formulas (11.22) and (11.23) we get

 ω o ω M 
µ = µ 0 1 − = 2.44 µ 0 ,
 ω 2 − ω o2 

ωωM
µa = µ0 = 0.246µ 0 ,
ω 2 − ω o2

and the tensor of permeability (11.21) is

 µ − jµ a 0   2.44 − j 0.246 0

µ =  jµ a µ 0  = µ 0  j 0.246
 2.44 0
 0 0 µ 0   0 0 1
_________________________________________

11.2 Longitudinal propagation of a plane electromagnetic wave in a


magnetized ferrite
Let us assume an unbound space x
filled homogeneously by a ferrite plane of a constant phase
material considered to be lossless. This
material is magnetized in the z direction Ho
by a static magnetic field Ho = Hoz0. A
uniform plane electromagnetic wave
propagates in this space in the direction
z
of the z axis, i.e., in the direction parallel y k
to the magnetizing field, Fig. 11.2. Both
the electric field and the magnetic field of
this wave do not depend on the x and y Fig. 11.2

coordinates. Assuming = 0 and
∂x

= 0 we can rewrite Maxwell’s first and second equations (1.12) and (1.13) into
∂y

∂H y
− = jωε E x , (11.27)
∂z

∂H x
= jωε E y , (11.28)
∂z

0 = jωε E z , (11.29)

∂E y

∂z
(
= − j ω µH x − jµ a H y , ) (11.30)

146
book - 11
∂E x
∂z
(
= − j ω j µ a H x + µH y , ) (11.31)

0 = − jωµ 0 H z , (11.32)

Equations (11.29) and (11.32) tell us that the propagating wave is a transversal
electromagnetic wave, as the longitudinal field components are zero. We are looking for the
solution to above equations in the form of a plane TEM wave

H x = H mx e − jk z z , (11.33)

H y = H my e − jk z z , (11.34)

E x = E mx e − jk z z = Z xy H my e − jk z z , (11.35)

E y = E my e − jk z z = − Z yx H mx e − jk z z , (11.36)

where the wave impedances are defined

E mx
Z xy = , (11.37)
H my

E my
Z yx = − , (11.37)
H mx

and can be determined together with propagation constant kz by solving equations (11.27) –
(11.32). We insert the form of field (11.33), (11.34), (11.35) and (11.36) into (11.27) –
(11.32)

jk z H my e − jk z z = − jωε E mx e − jk z z = − jωε Z xy H my e − jk z z ,

jk z H my == − jωε Z xy H my , (11.38)

− jk z H mx e − jk z z = − jωε E my e − jk z z = jωε Z yx H mx e − jk z z ,

− jk z H mx = jωε Z yx H mx . (11.39)

Similarly from (11.30) and (11.31) we get

(
k z H mx Z yx = ω µH mx − jµ a H my , ) (11.40)

(
k z H my Z xy = ω jµ a H mx + µH my . ) (11.41)

From (11.38) and (11.39) we have the wave impedances

147
book - 11
kz
Z xy = Z yx = , (11.42)
ωε

From (11.40) and (11.41) we exclude the wave impedance by (11.42). We get the set of two
equations for Hmx and Hmy

(k 2
z )
− ω 2 µε H mx + jω 2ε µ a H my = 0 , (11.43)

(
− jω 2ε µ a H mx + k z2 − ω 2 µε H my = 0 . ) (11.44)

Propagation constant kz can be determined from the condition of solubility of the set of the
equations (11.43) and (11.44), which requires that the determinant of the system matrix is
equal to zero

(k 2
z − ω 2 µε ) + ( jω
2 2
ε µa )
2
=0 .

This equation has two solutions defining the propagation constant of a wave propagating
through the space filled by a ferrite material

k zL, R = ω ε (µ ± µ a ) . (11.45)

Inserting the propagation constants (11.45) into (11.43) and (11.44) we get

H myL = j H mxL , (11.46)

H myR = − j H mxR . (11.47)

The propagating wave can thus be decomposed as the superposition of two partial
electromagnetic waves. These waves can be expressed

H L = (x 0 + j y 0 ) H mxL e − jk zL z , (11.48)

H R = (x 0 − j y 0 ) H mxR e − jk zR z , (11.49)

These two waves are circularly polarized waves. The wave denoted by index L is a circularly
polarized wave with left-handed rotation. The wave denoted by index R is a circularly
polarized wave with right-handed rotation. The wave propagating in the ferrite consequently
behaves as the superposition of the two circularly polarized waves with left-handed and right-
handed polarization. Their propagation constants (11.45) are different, and so are the phase
velocities and the wave impedances

ω 1
vL = = , (11.50)
k zL ε (µ + µ a )

ω 1
vR = = , (11.51)
k zR ε (µ − µ a )
148
book - 11
k zL µ + µa
ZL = = , (11.52)
ωε ε

k zR µ − µa
ZR = = . (11.53)
ωε ε

We substitute for the term (µ ± µ a ) , where µ is defined by (11.22) and µa by (11.23),


the effective permeability µef

 ωM 
µ ef = µ ± µ a = µ 0 1 ± . 10
 ω ± ω o  µef/µ0
8
(11.54)
6
µefR
The dependence of 4
relative effective permeability 2 µefL
µ ef µ 0 and of normalized
0
phase velocity vc on 1 ω/ωo
-2
normalized frequency ω ω o is µefR
-4
plotted in Fig. 11.3. There is a
frequency band in which the -6
permeability of the wave with -8
right-handed polarization is
-10
negative. The corresponding
phase velocity (11.51) together
with the wave impedance 3
(11.53) are imaginary in this v/c
frequency range. This means that
the wave does not propagate.
2
There is a sharp resonance in the
plots in Fig. 11.3 at ω = ωo. It is vR
known as ferromagnetic
resonance. At this frequency, 1
equal to Larmor precession vL
vR
frequency ωo, vector Hm
performs the precession
0
movement, which is not damped ω/ωo
in the case of a lossless ferrite. 1
The wave with right-handed Fig. 11.3
polarization is intensively
attenuated at this resonance. The propagation of the wave with left-handed polarization is not
influenced by the ferromagnetic resonance, as the direction of the magnetic field vector
rotation is opposite to the precession movement. It follows from this that in practical
applications of the ferrite material we have to choose a frequency fairly different from the
Larmor precession frequency ωo. This removes the strong wave attenuation together with the
severe dependence of all wave parameters on frequency. Note that all this concerns the case
of a wave propagating in the direction parallel to DC magnetizing field Ho.
The phase velocities of the partial waves with right-handed and left-handed
polarization (11.50) and (11.51) are different. Let us now assume that a plane TEM wave with
149
book - 11
linear polarization propagates in the ferrite material. This wave is the superposition of the two
above defined waves which can be expressed using (11.48) and (11.49)

H = H R + H L = (x 0 + jy 0 )H mL + (x 0 − jy 0 )H mR .

Passing the distance d along the z axis, the two particular waves change their phase
differently, as they have different phase velocities and the wave can consequently be
expressed assuming that the amplitudes are equal HmL = HmR = Hmx

H (d ) = H L e − jk zL d + H R e − jk zR d =
[ ( ) ( )]
= H mx x 0 e − jk zR d + e − jk zL d − jy 0 e − jk zR d − e − jk zL d = , (11.55)
  k − k zL   k − k zL 
= 2 H mx e − jka d x 0 cos zR d  − y 0 sin  zR d 
  2   2 

where

k zL + k zR
ka = .
2

Equation (11.55) shows that the x component and the y component of the magnetic field of
the propagating wave stay in phase. This means that the polarization remains linear. The plane
of the polarization is twisted by angle ψ

Hy  k − k zL  k − k zL
tg (ψ ) = = tg zR d  => ψ = zR d = KF d , (11.56)
Hx  2  2

where KF is the Faraday constant.


This effect is known as the Faraday effect. The polarization plane of the wave with
linear polarization propagating in the ferrite material in the direction parallel with the DC
magnetizing field is gradually rotated. The measure of this rotation depends on frequency f,
magnetizing field Ho and the parameters of the ferrite.

________________________________________
Example 11.3: Calculate the Faraday constant determining the rotation of the polarization
plane of the wave propagating in the ferrite material from Example 11.2 in parallel with the
magnetizing field at the frequency 300 MHz. Calculate the distance through which the
polarization plane is rotated by 90°.
Using (11.45), we get the propagation constants of the wave with left-handed
polarization and the wave with right-handed polarization

k zL = ω ε (µ + µ a ) = 10.28 m-1,

k zR = ω ε (µ − µ a ) = 9.29 m-1.

µ and µa calculated in Example 11.2 were used. Using (11.56) we get

150
book - 11
k zR − k zL
KF = = −0.495 rad/m KF = -28.5 °/m .
2

The polarization plane is rotated by 90° through the distance 3.16 m.


________________________________________

11.3 Transversal propagation of a plane electromagnetic wave in a


magnetized ferrite
Let us assume that a plane
electromagnetic wave propagates in a x
ferrite material perpendicular to a DC
k
magnetizing field Ho = Hoz0. Let us say in
the direction of the x axis, Fig. 11.4. Both Ho
the electric field and the magnetic field of
this wave do not depend on the y and z
∂ z
coordinates. Assuming = 0 and y
∂y plane of a constant phase

= 0 we can rewrite Maxwell’s first and Fig. 11.4
∂z
second equations (1.12) and (1.13)
similarly as in the case of longitudinal propagation into

Ex = 0 , (11.57)

∂H z
− = jωε E y , (11.58)
∂x

∂H y
= jωε E z , (11.59)
∂x

( )
0 = − jω µ H x − jµ a H y , (11.60)

∂E z
∂x
(
= jω jµ a H x + µ H y , ) (11.61)

∂E y
= − jω µ 0 H z , (11.62)
∂x

Assuming dependence on the x coordinate in the form e − jk x x which describes a plane


electromagnetic wave propagating in the x direction, the equations (11.57) to (11.62) read

E mx = 0 , (11.63)

k x H mz = ωε E my , (11.64)

151
book - 11
k x H my = −ωε E mz , (11.65)

µH mx = jµ a H my , (11.66)

(
− k x E mz = ω jµ a H mx + µH my , ) (11.67)

k x E my = ω µ 0 H mz , (11.68)

The set of equations (11.63) to (11.68) can be divided into two sets. The equations
(11.64) and (11.68) define the system of Emy and Hmz. The equations (11.65), (11.66) and
(11.67) define the system of Emz, Hmx and Hmy.
Dividing equations (11.64) and (11.68) we get the propagation constant

k x = k xo = ω µ 0 ε . (11.69)

Using the propagation constant we can define the phase velocity

ω 1
vo = = . (11.70)
k xo µ 0ε

The wave impedance is

E my µ0
Zo = = . (11.71)
H mz ε

Equations (11.69), (11.70) and (11.71) define the propagation constant, the phase velocity and
the wave impedance of a standard plane TEM wave. This wave is known as an ordinary
wave. The ordinary wave propagates perpendicularly to the DC magnetizing field and its
magnetic field is parallel with Ho. The propagation of the ordinary wave does not depend on
Ho .
The second set of equations describes the wave propagating perpendicular to Ho. It has
the longitudinal component of magnetic field Hmx shifted by π/2 to transversal component
Hmy. Electric field Emz is parallel to Ho. This wave is known as an extraordinary wave. From
(11.65), (11.66) and (11.67) we get the propagation constant, the phase velocity and the wave
impedance of the extraordinary wave

µ 2 − µ a2
k xe = ω ε , (11.72)
µ

ω 1
ve = = , (11.73)
k mx µ 2 − µ a2
ε
µ

E k µ 2 − µ a2
Z e = − mz = xe = . (11.74)
H my ωε εµ

152
book - 11
These quantities depend on Ho. At µ = 0 we have ve = 0 and the extraordinary wave does not
propagate. This happens at the frequency

ω t = ω o (ω o + ω M ) . (11.75)

This effect is transversal ferromagnetic resonance. The extraordinary wave can be


described by the field components

H y = H my e − jk xe x , (11.76)

µa
H x = H mx e − jk xe x = j H my e − jk xe x , (11.77)
µ

E z = H mz e − jk xe x = − Z e H my e − jk xe x . (11.78)
x
The plane TEM wave with linear polarization
propagating perpendicularly to the DC magnetizing
field can be decomposed into ordinary and kxe
extraordinary waves, Fig. 11.5. These waves have extra- Hxe
different phase velocities and therefore they have ordinary Eze
different phases. The ratio of the electric fields of wave
these waves is Hye

E yo E my E my
= e j (k xe − k xo )x = e jψ . kxo
E ze E mz E mz
ordinary
Hzo
This shows that the phase between the two wave
components of the electric field varies with the x Eyo
coordinate, and the TEM wave which is the
superposition of the ordinary wave and the z
extraordinary wave changes its polarization as the Ho
wave propagates, Fig. 11.6. The wave has linear y
polarization at the points at which
ψ = (k xe − k xo )x = nπ . The character of the wave Fig. 11.5
polarization changes regularly with x from linear
polarization, to elliptical polarization, circular polarization, again to elliptical polarization,
etc., Fig. 11.6.

________________________________________
Example 11.4: A plate TEM wave is incident from the air at the angle ϑi = 45° to the surface
of the ferrite material from Example 10.2, the magnetizing field is parallel to the ferrite
surface, Fig. 11.7, and the ferrite permittivity is εr = 10. The frequency is f = 300 MHz. The
wave has arbitrary orientation of vector E to the plane of incidence. Calculate the difference
between refraction angles of an ordinary wave and of an extraordinary wave.

153
book - 11
0<ψ<π/2 ψ=π/2 π/2<ψ<π ψ=π π<ψ<3π/2 ψ=3π/2 3π/2<ψ<2π ψ=2π
y

ψ=0
45° 45° 45°

E
x
z
Ho

Fig. 11.6
The refracted wave propagates in the ferrite perpendicular to the magnetizing field.
The incident wave must be expressed as the superposition of the wave with horizontal
polarization and of the wave with vertical polarization. The wave with horizontal polarization
has vector E perpendicular to the plane of incidence, i.e.,
parallel with Ho. It represents an extraordinary wave in the
ferrite. The wave with vertical polarization has vector E
parallel to the plane of incidence, i.e., perpendicular to Ho. It ϑi
represents an ordinary wave in the ferrite. The propagation
constants of these waves (11.69) (10.72) are
ferrite
k = ω µ ε = 19.83 m-1 , H o
o 0 ϑo
ϑe
µ 2 − µ a2
ke = ω ε = 9.7 m-1 .
µ Fig. 10.7

From Snell’s law (3.49) we get refraction angles ϑto = 12.9° and ϑte = 27°. The difference
between these angles is 14.1°.
_________________________________________

11.4 Applications of non-reciprocal devices


There are devices working with a longitudinally magnetized ferrite which make use of
the Faraday effect, and devices with a transversally magnetized ferrite. The latter have simpler
construction and are more universal.
Among nonreciprocal devices, we can mention a gyrator marked in circuits as shown
in Fig. 11.8a. The gyrator
shifts the phase of a 1
transmitted wave in one
direction by 180° and there π
is no phase shift in the
3 2
return direction. The
insulator, Fig. 11.8b, gyrator insulator
transmits a wave only in circulator
one direction and does not a b c
transmit in the return Fig. 11.8
154
book - 11
direction. The circulator, Fig. 11.8c, transmits a wave only in the direction of an arrow from
gate 1 to gate 2, from gate 2 to gate 3, and from gate 3 to gate 1. Controlling circuits are
steered by changing the electric current in the winding of an electromagnet magnetizing a
ferrite. Ferrite resonators were mentioned in Chapter 9.
Fig. 11.9 shows a sketch of an insulator with a longitudinally magnetized ferrite
working with the use of the Faraday effect. The wave in the form of the TE10 mode in a
rectangular waveguide propagates from the left. It is transformed to the TE11 mode of the
waveguide with a circular cross-section. Its electric field, Fig. 11.9, is perpendicular to the
resistive plate and is not attenuated by this plate. This mode travels in the ferrite and its plane
of polarization is rotated by 45°, as its electric field is again perpendicular to the second
resistive plate and is transformed to the TE10 mode of an output rectangular waveguide. The
rotation of the polarization plane of the wave traveling back is in the opposite direction, so the
electric field is parallel to the resistive plate and, moreover, the TE01 mode which cannot
propagate is excited in the output waveguide. The wave is thus highly attenuated.

Ho

TE10 TE11 TE11 TE10

TE01 TE11 TE11 TE10

Fig. 11.9

11.5 Problems
11.1 Calculate the frequency of the ferromagnetic resonance and the magnetizing constant of
a ferrite magnetized by the field Ho = 5 104 A/m. The magnetization is Mo = 7 104 A/m.
ωο = 1.105 1010 s-1
fr = 1.76 GHz
ωM = 1.547 1010 s-1

11.2 Recalculate the tensor of permeability of the ferrite material from Example 11.2 at the
frequency 3 GHz.
0.266 − j1.25 0
µ = µ 0  j1.25 0.266 0
 0 0 1

155
book - 11
11.3 Calculate for the ferrite material from Example 11.2 the propagation constants of the
waves with left-handed polarization and with right-handed polarization propagating in parallel
with a magnetizing field at the frequency 3 GHz.
kL = 77.2 m-1
kR = j62 m-1
The wave with right-handed polarization does not
propagate, as kR is a purely imaginary number.

12. APPLICATIONS OF ELECTROMAGNETIC FIELDS


Knowledge of electromagnetic field theory is applied in the analysis and design of
systems in all branches of high frequency technology. These are namely: microwave
technology, antennas, propagation of electromagnetic waves, and optical communications. In
particular paragraphs we will briefly introduce these branches.
All these particular branches contribute to the design of, e.g., the communication link
shown in Fig. 12.1. Microwave technology covers the design of feeders, the output high
transmitting receiving
antenna antenna

feeder feeder
transmitter receiver
(generator)

propagating wave
(guiding medium)

Fig. 12.1
frequency part of a transmitter, and the input high frequency part of a receiver. Antennas are
treated usually separately, as their analysis, optimization and design are very specific.
Similarly we treat the propagation of waves in the atmosphere above ground in a special way.
In a communication system antennas can be substituted by proper converters to excite a light
signal in an optical fiber. This problem belongs to optoelectronic technology. Proper
transmission lines, e.g., coaxial cable, were used in older communication links.
Special measurement techniques using special measuring devices and systems are
applied to verify all steps in the development process of all high frequency circuits and
systems. It is beyond the scope of this textbook to treat these techniques. In the following
paragraphs we briefly introduce problems connected with microwave technology, antennas,
propagation of waves and optoelectronic technology.

12.1 Introduction to microwave technology


The spectrum of electromagnetic waves is shown in Fig. 1.1 together with typical
applications. Up to the frequency of about 1 GHz most circuits are constructed using lumped
parameter circuit components. In the frequency range from 1 up to 100 GHz, lumped circuit
elements are usually replaced by transmission line and waveguide components. Thus the term
microwave technology refers generally to the engineering and design of information handling

156
book - 12
systems in the frequency range from 1 to 100 GHz, corresponding to wavelengths as long as
30 cm and as short as 3 mm. At shorter wavelengths we have what can be called optical or
quasi-optical technology since many of the techniques used here are derived from classical
optical techniques. The characteristic feature of microwave technology is the short
wavelengths involved, these being of the same order of magnitude as the circuit elements and
devices employed. Conventional low frequency circuit analysis based on Kirchhoff’s laws
and voltage current concepts cannot be applied. It is necessary instead to carry out the
analysis in terms of a description of electric and magnetic fields.
There is no distinct frequency boundary at which lumped parameter circuit elements
must be replaced by distributed circuit elements. With modern technological processes it is
possible to construct printed circuit inductors and on chip capacitors that are so small that
they retain their lumped parameter characteristics at frequencies as high as 10 GHz or even
higher. Likewise, optical components, such as parabolic reflectors and lenses, are used to
focus waves with wavelengths as long as 1 m or more.
The first development of microwave technology was started with the development of
radars, as the need for high resolution radar capable of detecting small targets is coupled with
the need to raise the frequency. This arises predominantly from the need to have antennas of
sufficiently small dimensions that will radiate essentially all the transmitted power into a
narrow pencil-like beam.
In recent years microwave frequencies have also come into widespread use in
communication links. This follows from the demand to increase the amount and the speed of
transmitted information. Satellite and mobile communication systems are based on
microwave technology. At the present time most communication systems use the
transmission of digital signals.
Waveguides periodically loaded with shunt susceptance elements support slow waves
having velocities much less than the velocity of light, and are used in linear accelerators.
These produce high energy beams of charged particles for use in atomic and nuclear research.
The slow traveling electromagnetic waves interact very efficiently with charged-particle
beams having the same velocity, and thereby give up energy to the beam. Another possibility
is for the energy in an electron beam to be given up to an electromagnetic wave, with
resultant amplification. This latter device is the traveling-wave tube.
Sensitive microwave receivers are used in radio astronomy to detect and study the
electromagnetic radiation from the sun and a number of stars that emit radiation in this band.
Microwave radiometers are also used to map atmospheric temperature profiles, moisture
conditions in soils and crops, and for other remote sensing applications.
Molecular, atomic, and nuclear systems exhibit various resonance phenomena under
the action of periodic forces arising from an applied electromagnetic field. Many of these
resonances occur in the microwave range. We have thus a very powerful experimental probe
for studying the basic properties of materials. Out of this research on materials have come
many useful devices employing ferrites, see Chapter 11.
The development of the laser, a generator of essentially monochromatic (single-
frequency) light waves, together with semiconductor technology, has stimulated great interest
in the possibilities of developing communication systems at optical wavelengths.
There are plenty of the applications of microwaves in industrial processes –
microwave ovens, drying of materials, manufacturing wood and paper products, material
curing. Microwave radiation has also found some application for medical hyperthermia or
localized heating of tumors.
At frequencies where the wavelength is several orders of magnitude larger than the
greatest dimensions of the circuit or system being examined, conventional circuit elements
such as capacitors, inductors, resistors, transistors or diodes are the basic building blocks for
the information transmitting, receiving, and processing circuits used. The description or

157
book - 12
analysis of such circuits may be adequately carried out in terms of loop currents and node
voltages without consideration of propagation effects. The time delay between cause and
effect at different points in these circuits is so small compared with the period of the applied
signal as to be negligible. It might be noted here that an electromagnetic wave propagates a
distance of one wavelength in a time interval equal to one period of a sinusoidally time
varying applied signal. As a consequence, when the distances involved are short compared
with a wavelength, the time delay is not significant. As the frequency is raised to a point
where the wavelength is no longer large compared with the circuit dimensions, propagation
effects can no longer be ignored. A further effect is the great relative increase in the
impedance of the connecting leads, terminals, etc., and the effect of distributed capacitance
and inductance. In addition, currents circulating in unshielded circuits comparable in size
with a wavelength are very effective in radiating electromagnetic waves. The net effect of all
this is to make most conventional low frequency circuit elements and circuits hopelessly
inadequate at microwave frequencies. As is well known, lumped circuit elements do not
behave in the desired manner at high frequencies. For example, a coil of wire may be an
excellent inductor at 1 MHz, but at 50 MHz it may be an equally good capacitor because of
the predominating effect of interturn capacitance. Even though practical low frequency
resistors, inductors and capacitors do not function in the desired manner at microwave
frequencies, this does not mean that such energy dissipating and storage elements cannot be
constructed at microwave frequencies. On the contrary, there are many equivalent inductive
and capacitive devices for use at microwave frequencies. Their geometrical form is quite
different, but they can be and are used for much the same purposes, such as impedance
matching, resonant circuits, etc.
For low power applications, microwave tubes have been completely replaced by
transistors, diodes, and negative resistance diodes. However, for high power applications
microwave tubes are still necessary.
One of the essential requirements in a microwave circuit is the ability to transfer signal
power from one point to another without radiation loss. This requires the transport of
electromagnetic energy in the form of a propagating wave. A variety of such structures have
been developed that can guide electromagnetic waves from one point to another without
radiation loss. The simplest guiding structure, from an analysis point of view, is the
transmission line. Several of these, such as the open two-conductor line, the coaxial line, the
parallel plate line, and the strip line, illustrated in Fig. 5.1b, are in common use at the lower
microwave frequencies.
At the higher microwave frequencies hollow-pipe waveguides, as illustrated in Fig.
5.1c, may be used. Today, these waveguides are used predominantly only in special systems,
where high power signals are transmitted, or where extremely low losses are required.
The development of semiconductor high frequency active devices, such as bipolar
transistors, field-effect transistors, and diodes, has had a dramatic impact on the microwave
engineering field. With the availability of microwave transistors, the focus on waveguides
and waveguide components has changed to a focus on planar transmission line structures,
such as microstrip lines and coplanar waveguides. These structures, shown in Fig. 5.1b, can
be manufactured using printed circuit techniques. They are compatible with microwave
semiconductor devices, and can be miniaturized, they are cheap, and are thus suitable for
mass production. Microwave circuits are usually designed as hybrid integrated microwave
circuits. In hybrid circuit construction the transmission lines and transmission line
components, such as matching elements, are manufactured first and then the semiconductor
devices are soldered into place. The current trend is toward the use of monolithic microwave
integrated circuits (MMIC) in which both the transmission line circuits and the active devices
are fabricated on a single chip.

158
book - 12
A unique property of the transmission line is that a satisfactory analysis of its
properties may by carried out by treating it as a network with distributed parameters and
solving for the voltage and current waves that may propagate along the line, see Chapter 6. In
the case of planar transmission lines this can be done at lower microwave frequencies only.
At higher frequencies they must be, similarly as hollow-pipe waveguides, treated as
electromagnetic boundary value problems, and a solution for the electromagnetic fields must
be determined, see Chapter 7. The reason is that it is not possible to define a unique voltage
and current that have the same significance as at low frequencies. However using a proper
normalization the graphical means of waveguide analysis, such as the Smith chart, see
Chapter 6.3, are commonly used.
Associated with transmission lines and waveguides are a number of interesting
problems related to methods of exciting fields in guides and methods of coupling energy out.
Signals from generators and measuring devices are usually carried with the use of coaxial
cables. A number of connectors have been produced which are compatible to connecting
planar transmission lines by soldering them directly into their layout, Fig. 12.2. These lines
can be simple coupled using the proximity effect, see the microstrip line coupler in Fig. 12.3

1 2

3 4
Fig. 12.3
Fig. 12.2
aperture
with ports 1 to 4. Three
basic coupling methods
are used in the case of
waveguides, Fig. 12.4:
probe coupling, see Fig.
12.4a and Fig. 7.8, loop a b c
coupling, see Fig. 12.4b, Fig. 12.4
and aperture coupling
between adjacent guides,
see Fig. 12.4c. These coupling devices are actually small antennas that radiate into a
waveguide.
Inductive and capacitive elements take a variety of forms at microwave frequencies.
Perhaps the simplest are short circuited sections of a transmission line and a waveguide, see
Chapter 6.2.3. These exhibit a range of susceptance values from minus to plus infinity,
depending on the length of the line, and hence may act as either inductive or capacitive
elements. They may be connected as either series or shunt elements, as illustrated in Fig. 12.5.
They are commonly referred to as stubs and are widely used as impedance matching
elements, see Example 6.8. In a rectangular waveguide thin conducting windows, or
diaphragms, as illustrated in Fig.
12.6, also act as shunt susceptive
elements. Their inductive, Fig. 12.6a,
or capacitive, Fig. 12.6b, nature
depends on whether there is more
magnetic energy or electric energy a - series stub b – shunt stub
stored in the local fringing fields.
Fig. 12.5
159
book - 12
Inductors and capacitors are designed namely
for microwave integrated circuits in forms resembling
lumped parameter elements. The example is an
interdigital capacitor designed for the coplanar
waveguide illustrated in Fig. 12.7. An example of an
a b
inductor is a spiral inductor designed for a microstrip
line technology, shown in Fig. 12.8. Fig. 12.6
Resonant circuits are used both at low
frequencies and at microwave frequencies to control the frequency of an oscillator and for
frequency filtering. At low frequencies this function is performed by an inductor and a
capacitor in a series or parallel combination. Resonance occurs when there are equal average
amounts of electric and magnetic energy stored. This energy oscillates back and forth between
the magnetic field around the inductor and the
electric field between the capacitor plates. At
microwave frequencies the LC circuit may be
replaced by a closed conducting enclosure, or
cavity, see Chapter 9.1. The electric and
magnetic energy is stored in the field within the
cavity. At an infinite number of the resonant
frequencies (9.1) there are equal average
amounts of electric and magnetic energy stored
in the cavity volume. In the vicinity of any one
resonant frequency, the input impedance to the
cavity has the same properties as for a
conventional LC resonant circuit. One
significant feature worth noting is the very much metallization slot
larger Q values that may be obtained, these
Fig. 12.7
being often in excess of 104, as compared with
those obtainable from low frequency LC circuits. Dielectric and
ferrite resonators are mentioned in Chapter 9.1. Resonators can be
formed even using the planar technology of printed transmission
lines. An example is a patch resonator, frequently used as a
microstrip patch antenna, Fig. 12.9. The Q factor of these
resonators is rather low, due to their radiation. Fig. 12.9 shows a
rectangular patch resonator, the resonators of various different
shapes are even used.
When a number of microwave devices are connected by
means of sections of transmission lines or waveguides, we obtain a Fig. 12.8
microwave circuit. An analysis of the behaviour of such circuits is
carried out either in terms of
the equivalent transmission
line voltage and current
waves or in terms of the
conducting
amplitudes of the propagating
patch
waves. The first approach feeding
leads to an equivalent microstrip
impedance description, and
the second emphasizes the dielectric substrate
wave nature of the fields and
results in a scattering matrix backward metallization
formulation.
Fig. 12.9
160
book - 12
Particular microwave circuits perform a variety of particular tasks in complex systems
or subsystems. We can include among these circuits: oscillators, amplifiers, modulators,
demodulators, mixers, filters, switches, multiplexers, demultiplexers, etc. It is beyond the
scope of this textbook to treat these circuits in detail.
Scattering parameters:
A microwave circuit can be generally
represented by an N-port, i.e., a circuit with N
ports, Fig. 12.10. The internal structure of this b1 bn
circuit may be rather complex and it can be U1 a1 an Un
very difficult to find the solution of the
Maxwell equations describing the field. Let .
us try to describe such a circuit. Let us b2 .
assume that each port is connected by a U2 a2 .
transmission line to the rest of a system.
There is a wave, which is incident to this . bN
aN UN
circuit port propagating along this line toward .
the port, and there is a wave reflected from .
the port and propagating back. These waves
are determined by amplitudes U n+ and U n− . Fig. 12.10
Let us normalize these amplitudes by

U n+
an = , (12.1)
Z Cn

U n−
bn = , (12.2)
Z Cn

where ZCn is the characteristic impedance of the line connected to the n-th port. The voltage
and the current at the n-th port are

U n = U n+ + U n− = Z Cn (a n + bn ) , (12.3)

In =
1
Z Cn
(U n+ − U n− = ) 1
Z Cn
(an − bn ) . (12.4)

The power at the n-th port is

Pn = Pn+ − Pn− =
1
2
[ 1
]
Re U n I n* = Re a n
2
[ 2
− bn
2
( )]
+ b a n* − a n bn* =
1
2
an
2

1
2
bn
2
.

So the normalization was done to be able to express simply the power transmitted by the
incident wave and the power transmitted by the reflected wave

1 2
Pn+ = an , (12.5)
2

161
book - 12
1 2
Pn− = bn . (12.6)
2

It is reasonable to assume that all incident waves contribute to each reflected wave. In
the case of a linear circuit, this dependence must be linear. It reads

bn = S n1a1 + S n 2 a 2 + S n3 a 3 + ... + S nj a j + ... + S nN a N , n = 1, 2, 3, ….,N . (12.7)

In the matrix form we have

 b1   S11 S12 . S1N   a1 


b   S S 22 . S 2 N   a 2 
 2  =  21 . (12.8)
 .   . . . .  . 
    
bN   S N 1 SN2 . S NN  a N 

Matrix [S ] is known as a scattering matrix. Particular elements of the scattering matrix are
defined as

bi
S ij = for ak = 0 until k≠ j . (12.9)
aj

These elements represent the transmission from the j-th port to the i-th port. The main
diagonal elements Sii represent the reflection coefficient at the i-th port. This assumes that all
other ports are matched and no energy comes to them.
The elements of a scattering matrix are generally complex numbers and can be
displayed on a Smith chart as functions of frequency. Their amplitudes are often expressed in
dB. The scattering matrix can be simply transformed to a transition matrix or to a impedance
matrix.

12.2 Antennas
An antenna is an element that transforms energy from an electric circuit to energy
transmitted by a radiated wave into space, or vice versa. The basic theory of wave radiation,
antennas, and antenna arrays was presented in Chapter 10 together with definitions of the
basic antenna parameters. These are: radiation pattern, input impedance, antenna directivity
and gain, effective length and area of the receiving antenna. All these parameters depend on
frequency. An antenna is an integral part of communication links, navigation systems, radars,
telescopes, and various sensors.
Antennas are often designed as metallic bodies. The basic problem in this case is to
determine the distribution of the electric current on the surface of the antenna. Once we know
the distribution of this current we can relatively simply calculate the field radiated by an
antenna using formulas (10.26) to (10.30) derived in Chapter 10.
Let us assume an antenna as a perfectly conducting body with surface S, Fig. 12.11,
located in an unbounded space filled by homogeneous lossless material with wave impedance
Z0 and propagation constant k. This antenna is irradiated by a plane electromagnetic wave
described by vector Ei. This field induces an electric current with density K on the surface of
an antenna. This current produces electric field Es radiated, or scattered, back by an antenna.
Consequently the total field is the sum of these particular fields

162
book - 12
E = Ei + Es . (12.10)

Let us assume for the sake of simplicity, Ei


the problem that irradiating field Ei
remains unchanged by the incidence to k K
the surface of an antenna, and so the Hi n
disturbance of a field is caused purely by R
scattered field Es. The total field must
fulfill the boundary condition on the
antenna surface (1.22) r
r’
n×E = 0 ,
n × E s = −n × E i . (12.11) S
0

The scattered field is determined by


(10.28) Fig. 12.11

j
E s = − jω A − grad div(A ) , (12.12)
ωµε

where vector potential A is determined by the surface equivalent of (1.58), Fig. 12.12
µ K (r ') e − jkR
A(r ) =
4π ∫∫
dS , (12.13)
S
R

where R = r – r’. Inserting for vector potential (12.13) into (12.12) and consequently for the
scattered field into (12.11) we get an integral equation for the unknown distribution of electric
current K on the antenna surface

 µ K e − jkR j  µ K e − jkR 
  = −n × E i .
n ×  − jω
 4π ∫∫ R dS −
ωµε
grad div
 4π ∫∫ R dS

S  S 

We can exchange the order of derivatives and integration in the second term as we integrate
according to r’ and the derivatives included in operators gradient and divergence are
calculated according to r. So we have

µ  K e − jkR 1  K e − jkR 
j n × ∫∫  jω + grad div  dS = n × E i .
 (12.14)
4π S  R ωµε  R 

This integro-differential equation cannot be solved analytically. A proper numerical method


must be used to solve (12.14). Simplified methods are used in practice to determine the
distribution of current on the surface of antennas. These methods depend on the kind of
antenna.
The solution of the antenna problem is rather difficult. Therefore we often divide this
problem into two parts. Solving an inner antenna problem we in fact solve equation (12.14).
The aim is to determine the distribution of currents or charges on the antenna body or on
some equivalent surfaces. This problem must often be simplified. In the second step we solve
the outer antenna problem, which calculates of the field excited by the source currents. If

163
book - 12
we would like to determine the field in the radiating (far field) zone, we can use methods
explained in Chapter 10. The behaviour of transmitting and receiving antennas is similar. A
radiation pattern for transmission is equal to a radiation pattern for receiving, and the same is
valid for an antenna input impedance. Therefore we analyze antennas mostly as transmitting.
In the case of a receiving antenna we determine in addition the effective length or aperture
and possibly the noise temperature.
It is very difficult to list various kinds of antennas, as there is a great variety of them
and it is difficult to categorize them in some cases. We did not mention aperture antennas in
Chapter 10. The wave passes through an aperture, e.g., through an open-ended waveguide.
Examples of this are horn, reflector and slot antennas. A variety of wire antennas are used
in particular applications. Ultra wide band antennas must by applied in modern radars and
communication systems. Special types of antennas are lens and traveling wave antennas.
Planar antennas are widely used, as they are compatible with planar microwave circuits and
microwave integrated circuits. There is a huge variety of these antennas. All kinds of antennas
can be combined to create an antenna array, see Chapter 10.5. In the following text we briefly
introduce particular types of antennas.

12.2.1 Wire antennas


Dipole antennas as examples of wire antennas are widely
used. They were analyzed in Example 10.4. Greater directivity and L
lower side lobe levels than in the case of a straight dipole can be
gained using the vee dipole, Fig. 12.12. It is possible to find the
angle γ to get maximum directivity.
An extremely practical wire antenna is the folded dipole. It γ
consists of two parallel dipoles connected at the ends forming a
narrow wire loop, as shown in Fig. 12.13, with dimension d much
smaller than L. For L = λ/2 the input impedance of this dipole is
four times higher than the impedance of a straight dipole. This Fig. 12.12
antenna is suitable for TV and FM radio receiving, as it is well
matched to 300 Ω twin-lead transmission line. d
We saw in Chapter 10.5 that array antennas can be used to increase
directivity. The arrays we have examined had all elements active,
requiring a direct connection to each element by a feed network. The feed
networks for arrays are considerably simplified L
reflector if only a few elements are fed directly. Such an
driven element array is referred to as a parasitic array. The
directors elements that are not directly driven, called
parasites, receive their excitation by near-field Fig. 12.13
coupling from the driven elements. A parasitic
linear array of parallel dipoles is known as a Yagi-Uda antenna.
These antennas are simple and have a relatively high gain. The
basic unit consists of three elements one is driven and two are
parasitic, a director and a reflector. More than one director is
Fig. 12.14 usually used. Fig. 12.14 shows a Yagi-Uda antenna with five
directors. The current induced in closely spaced parasitic elements
has the opposite phase to the current of the driven element. From the radiation patterns shown
at the bottom of Fig. 10.12 it is clear that such an array radiates endfire. Lengthening one
parasitic element known as the reflector, the radiation dual endfire beam is changed to a more
desirable single endfire beam. This is more pronounced in a parasitic element shorter than the
driven element, known as a director. An example of the radiation pattern of a Yagi-Uda

164
book - 12
antenna with 5 directors is shown in Fig. 12.15. The radiation pattern was measured in the
plane of antenna symmetry perpendicular to
the antenna elements.

12.2.2 Aperture antennas


Aperture antennas can be solved using
the equivalence principle. Let
electromagnetic sources be contained in
volume V bounded by surface S with outward
normal n, Fig. 12.16a. The fields E and H
exterior to S can be found by removing the
Fig. 12.15
sources in V and placing the following surface
current densities on S, see Fig. 12.16b,

K = n × Ht , (12.15)

M = Et × n , (12.16)

where K is current surface


E, H Et n E, H M n
density on S, M is
equivalent magnetic current S Ht S K
density on S, Et and Ht are no sources
sources zero field
the electric and magnetic
fields produced by the V V
original sources on surface
S. Thus, with a knowledge a b
of the tangential fields over Fig. 12.16
a surface due to the original
sources, we can find the fields everywhere external to the surface through the use of
equivalent surface current densities K and M. These densities define vector potentials, see
(1.58),

µ K e − jkr
A=
4π ∫∫ r dS , (12.17)
S

µ M e − jkr
F=
4π ∫∫ r dS , (12.18)
S

where F is the electric vector potential, which is a quantity analogical to magnetic vector
potential A. The magnetic and electric vector potentials determine the field vectors, see
(10.28),

j 1
E = − jω A − grad div(A ) − rot (F ) , (12.19)
ωµε ε

j 1
H = − jω F − grad div (F ) + rot (A ) . (12.20)
ωµε µ

165
book - 12
Equations (12.19) and (12.20) follow from the symmetry of Maxwell’s equation gained by
introducing equivalent current densities (12.15) and (12.16). The calculation of the field in the
radiating zone follows the technique described in Chapter 10.3.
The gain of the aperture antenna can be estimated by, see Example 10.6 where the
homogeneous field distribution over the antenna aperture was assumed, not exactly fulfilled
in most aperture antennas,


G= Aeff , (12.21)
λ2

where Aeff = ηAeffm is the effective aperture area of an antenna, Aeffm being the maximum
effective aperture (10.46) area, and η is the
efficiency of radiation.
A typical example of aperture antennas B
is a pyramidal horn antenna, shown in Fig. E
12.17. This antenna is fed from the
rectangular waveguide. The pyramidal horn
antenna is flared in both the E- and H-planes. A
This configuration will lead to narrow
beamwidths in both principal planes, thus Fig. 12.17
forming a pencil beam. The gain can be
simply evaluated by (12.21), where Aeff = ηΑΒ. A horn antenna can also be designed on the
basis of a waveguide of circular cross-section.
In long-distance radio communication and high-resolution radar applications, antennas
with high gain are required. Reflector antennas are
perhaps the most widely used high gain antennas, reflector aperture
which routinely achieve gains far in excess of 30 dB plain
in the microwave region. The simplest reflector radiated ray
antenna consists of two components: a large (relative
to the wavelength) reflecting surface and a much
smaller feeding antenna. The most prominent 2a
example is the parabolic reflector antenna, see its focal point –
cross section in Fig. 12.18. The reflector has a feeding point
paraboloid of revolution shape. A feeder is placed at
the focal point. For large reflectors (a>>λ)
geometrical optics principles can be applied and
radiation from the antenna is analyzed by the ray Fig. 12.18
tracing method, see Chapter 4. The gain can be
simply evaluated by
(12.21), where Aeffm =
4πa2, Fig. 12.18. The radiating slot
radiation pattern of these
antennas is very narrow –
units of degrees.
There are various microstrip
versions of slot antennas. feeding the
Radiating slots can be metallization slot from a
made in the walls of a back side
waveguide or can be dielectric substrate
designed in a planar
version, Fig. 12.19. Fig. 12.19

166
book - 12
12.2.3 Broadband antennas
In many applications an antenna must operate effectively over a wide range of
frequencies. An antenna with wide bandwidth is referred to as a broadband, or an ultra wide
band (UWB), antenna. The term broadband is a relative measure of bandwidth and varies with
circumstances. Let fU and fL be the upper and lower frequencies of operation for which
satisfactory performance is obtained. The center frequency is fC. Then the bandwidth as a
percentage of the center frequency is

fU − f L
BW = × 100 . (12.22)
fC

Resonant antennas have small bandwidths, while, antennas that have traveling waves on them
operate over wider frequency ranges.
It is frequently desirable to have the radiation pattern and input impedance of an
antenna remaining constant over a very
wide range of frequencies, say 10:1 or
higher. An antenna of this type is referred
to as a frequency independent antenna.
This is, for example, the case of an
infinite biconical antenna, see Fig.
12.20. This antenna must however be Fig. 12.20
truncated, forming a finite biconical
antenna and most of its broadband behaviour
disappears. The condition of frequency
independence is based on a shape determined fully
by angles. The concept of angle emphasis has been
exploited in recent years and has led to a family of
wide bandwidth antennas. These are spiral
antennas and log-periodic antennas.
Planar spiral antennas can be simply
constructed using printed circuit techniques. Their
radiation pattern is bi-directional. To obtain
radiation in one direction, the spiral antenna is
placed on a conical surface. The shape of a spiral
antenna is determined by the equation of the spiral
in the polar coordinate system

r = r0 e aϕ , (12.23) Fig. 12.21

where r0 is the radius for ϕ = 0 and a is a constant giving the flare rate of the spiral. This
curve can be used to make the angular antenna shown in Fig. 12.21, which is referred as a
planar spiral antenna. The four edges of the metal each have an equation for their curves of
the form (12.23), r1 = r0 e aϕ , r2 = r0 e a (ϕ −π / 2 ) , r3 = r0 e a (ϕ −π ) , r4 = r0 e a (ϕ −π / 2−π ) . Maximal
radius R determines a lower frequency R = λ L / 4 and the distance of input ports determines
the upper frequency.
A log-periodic antenna is an antenna having a structural geometry such that its
impedance and radiation characteristics repeat periodically as the logarithm of the frequency.
There are several configurations of this antenna. Fig 12.22a shows a planar log-periodic
toothed trapezoid antenna, Fig. 12.22b shows a log-periodic trapezoid wire antenna, Fig.
12.22c shows log-periodic dipole array geometry. The wide band operation of these

167
book - 12
antennas is obtained due their linearly varied dimensions. Each part of the antenna is active at
different frequency band.

a b c
Fig. 12.22

12.2.4 Planar antennas dielectric substrate


Planar antennas
are designed in a planar radiating
technology. Due to this patch
they are compatible with
planar transmission lines
and microwave circuits
based on planar
technology. They can be
directly integrated into
these circuits and can air spacer microstrip
even be a part of coupling slot feeding the
monolithic microwave slot from
metallization
integrated circuits. back side
A planar
dielectric substrate
microstrip patch
antenna is shown in Fig. Fig. 12.23
12.9. The drawback of
this antenna is its rather narrow band. A number of modifications of a microstrip patch
antenna have been designed to widen its frequency band, to get the antenna radiating a
circularly polarized wave, to get dual frequency operation.
The simplest way to widen the microstrip patch antenna frequency band is to use a
thicker substrate, or possibly to divide this substrate into two layers, a thin dielectric substrate
and a thick air (foam) layer. This approach is very often combined with an aperture coupling.
This antenna is sketched in Fig. 12.23. The thick substrate reduces the antenna quality factor
and in this way widens the frequency band. At the same time, the slot through which energy is
coupled from the feeding microstrip line to the patch, together with the open ended microstrip
line stub terminating the line, is tuned to a frequency slightly different from the patch resonant
frequency. This offset can considerably increase the antenna frequency band.
There are a number of other types of planar antennas. Some from them were
introduced above. An electric dipole alone or as a part of an Yagi-Uda antenna can be
designed in planar geometry. Fig. 12.19 shows a slot antenna. Broad band antennas are often
designed in planar technology, see the spiral antenna in Fig. 12.21, or the planar log-periodic
toothed trapezoid antenna shown in Fig. 12.22a.
168
book - 12
12.3 Propagation of electromagnetic waves in the atmosphere
The problem of propagation of electromagnetic waves deals with the part of a
communication link shown in Fig 12.1 between the transmitting and receiving antennas. It
includes propagation of electromagnetic waves in the space above the Earth. Wave
propagation is influenced mainly by two factors. The first factor includes the parameters of
the medium itself in which the wave propagates. The second factor covers the geometry of the
whole scene. Here we have to assume the actual profile and cover of the ground, including
vegetation. The atmosphere changes its parameters – complex permeability – depending on
altitude. This causes, e.g., the bending of wave rays in lower parts of the atmosphere (the
troposphere), and wave reflection in the ionosphere. The atmosphere is a time varying
medium. The time changes are slow, and are caused by alternating seasons of the year, and by
changes between day and night. Rapid variation can be caused by changing weather, above all
by hydro-meteors present in the atmosphere – rain, snow, fog, water vapour. They cause
attenuation and scattering of waves. The propagation of an electromagnetic wave is thus not
the matter of a single ray simulating more or less straight propagation of a wave. The wave
received by the antenna is in most cases the sum of the particular waves caused by particular
reflections. Some of these effects are frequency very selective, some are not and are active in
wide frequency bands. Some of these effects vary rapidly with time, some are stable.
We can distinguish several basic types of propagation: surface wave, direct wave,
reflected and scattered wave, space wave, tropospheric wave, ionospheric wave, Fig. 12.24.
The ground surface is a boundary between two electrically different materials. Such a

ionosphere 50 - 400 km
ionospheric wave

troposphere up to 10 km
tropospheric wave

direct wave
surface wave

reflected wave

Earth

Fig. 12.24
boundary is able to guide a surface wave, see Chapter 8. The surface wave follows the bent
Earth surface. Its polarization is vertical. This kind of connection can be made via long
distances namely in the case of long waves, Fig. 1.1.
A direct wave represents a connection over short distances between places with direct
visibility and in the case of satellite communication. This can happen only at very high
frequencies. In practical cases a wave does not propagate in a straight direction. Due to space
variations of refractive index of the atmosphere, the trajectories of rays representing the wave

169
book – 12.1
propagation are bent, see Chapter 4.
In the case of a wave transmitted from elevated positions we have to consider not only
the direct wave, but the contribution of waves reflected from the ground or scattered from
various obstacles. These are the reflected wave and the scattered wave. These waves are
superimposed on the direct wave. This interference can cause unpredictable losses of the
received signal.
The simultaneous presence of the direct and reflected wave, the complex sum of these
waves, creates the total field. This, is known as propagation by the space wave. It is typified
namely by the direct visibility between the transmitter and the receiver, ensured by elevated
antennas at frequencies above 30 GHz. In this case we have so called multi-path
propagation, as there could be a number of reflected and scattered waves approaching the
receiving antenna at the same time.
The propagation of a tropospheric wave is used in communication over very long
distances (thousands of kilometers) for short waves. Propagation of this type uses scattering
of a wave on non-homogeneities in the troposphere that have different refractive indices. Only
a very small part of the transmitted power reaches the receiver, but a connection can be made
at distances far beyond the optical horizon.
An electromagnetic wave can propagate to long distances due to reflections from the
ionosphere. The layer of ionized air causes intensive continuous bending of the trajectories of
the rays as finally the rays are returned toward the Earth’s surface. Waves with a wavelength
longer than 10 m can be propagated by an ionospheric wave.
At present communication over long distances runs via satellites in the microwave
frequency ranges – satellite communication links.
Let us now discuss the power balance of a communication channel between two
antennas located at distance r in free an unbounded space filled by a lossless material
assuming that r is in the far field zone of the two antennas. In this case, we can assume
propagation of a spherical wave with its amplitude modified by angular dependence defined
by the transmitting antenna radiation pattern – antenna directivity DT. Assuming dependence
of the radiated field on distance in the form (2.49) we get the average value of Poynting’s
vector at the position of the receiving antenna

Pr DT PT GT
S av = 2
= , (12.24)
4πr 4πr 2

where Pr is the power radiated by the transmitting antenna, and PT is the total power supplied
to the transmitting antenna. The power received by the receiving antenna is

PR = S av Aeffmη R , (12.25)

where Aeffm is the receiving antenna maximum effective aperture (10.46) and ηR is the
efficiency of the receiving antenna. The effective aperture can be simply calculated for an
elementary electric dipole. It has the form

λ2
Aeffm = D , (12.26)

valid for any antenna. Using (10.38), (12.26) and (12.24) we get from (12.25)

170
book – 12.1
PT GT G R λ2
PR = S av Aeffmη R = .
4πr 2 4π

Consequently we get the power received by the receiving antenna

2
 λ 
PR =   GT G R PT . (12.27)
 4πr 

This term determines the received power assuming that all kinds of losses are neglected. In
spite of this fact the received power is lower than the transmitted power. The reason is that the
transmitted power is spread into the whole space, see Chapter 2.4, and consequently the field
amplitude decreases as 1/r and Poynting’s vector as 1/r2. The term (λ 4πr )2 is known as the
free space loss.
The transmission formula (12.27) assumes that antennas are targeted mutually in the
directions of the maximum of their radiation. Using (10.38) we can introduce antenna
directivities with their dependences on angles ϑ and ϕ

2
 λ 
PR = PT   ηT η R DT (ϑT , ϕ T ) DR (ϑ R , ϕ R ) , (12.28)
 4πr 

where index R denotes the receiving antenna and index T denotes the transmitting antenna.
Equation (12.28) is known as the Friis transmission formula. It is applicable in the case of
the general directions of the
antennas defined by angles transmitting antenna receiving antenna
ϑT , ϕ T ,ϑ R , ϕ R , Fig. 12.25. In
practical cases of propagation
all kinds of losses including ϑR, ϕR
the polarization mismatch ϑT, ϕΤ
must be included in (12.28).
Moreover, the multi-path
propagation must be taken into r
account and particular waves
must be added with their Fig. 12.25
proper phase and target
attenuation.
transmitter
Radar can be
assumed as a special case of
a communication channel.
An electromagnetic wave is receiver
transmitted by a transmitter,
reflected back by a target r
and received by a receiver,
Fig. 12.26. The time delay Fig. 12.26
between transmission and
reception of the pulses is proportional to the distance of the target. We will calculate the
power received by a radar receiver. To simplify this problem we assume that the transmitting
and receiving antennas are pointed such that the radiating pattern maxima are directed toward
the target. The power density incident on the target is then

171
book – 12.1
PT PT AeffT
S inc = GT = , (12.29)
4πr 2 λ2 r 2

where the gain has been expressed using the transmitting antenna effective aperture (12.21).
The power intercepted by the target is proportional to the incident power density (12.29), so

Pinc = σ S inc , (12.30)

where the proportionality constant σ is the radar cross section and is the equivalent area of the
target as if the target reradiated the incident power isotropically. Although power Pinc is not
really scattered isotropically, the receiver samples the scattered power in only one direction
and we are only concerned about that direction and assume the target scatters isotropically.
Because Pinc appears to be scattered isotropically, the power density arriving at the receiver is

Pinc
SR = . (12.31)
4πr 2

The power available at the receiver is then

PR = AeffR S R . (12.32)

Combining the above four equations gives

σS inc AeffT AeffRσ


PR = AeffR = PT , (12.33)
4πr 2 4πr 4 λ2

which is referred to as the radar equation. Usually the transmitting and receiving antennas
are identical, that is, AeffT = AeffR and GT = GR = G. Using (12.21) we can rewrite (12.33) in a
convenient form as

λ 2 G 2σ
PR = PT . (12.34)
(4π )3 r 4
The combination of (12.30) and (12.31) actually forms the definition of the radar cross
section

4πr 2 S R
σ = , (12.35)
S inc

which is the ratio of 4π times the radiation intensity, r2SR, in the receiver direction to the
incident power density from the transmitter direction.
The radar equation enables us to determine the range of the radar supposing we know
the transmitted power, the antenna parameters, the least power that must be received by the
receiver, and the target radar cross section. Equation (12.33) neglects the attenuation of a
wave by the atmosphere.

172
book – 12.1
12.4 Optoelectronics
With increasing demands to transmit more and more of information, transmission
channels must use wider and wider frequency bands. This is achieved by increasing the
transmission frequency. Huge possibilities are offered by using optical beams as a medium for
transmitting information. Here we are in the frequency band of the order 1014 Hz. This gives
us an extremely wide frequency band, which can be used to transmit a signal, i.e., the great
information capacity of an optical link. Optical signals can be exploited when we master the
special technology. This concerns generation, modulation, transmission, demodulation and
detection of optical signals.
Based on the knowledge gained in the preceding text, in this chapter we will introduce
basic problems of optoelectronics. We start with optical waveguides, and will continue with
coupling the optical signal to them, followed by optical detectors, optical amplifiers, lasers,
optical modulators and, finally, optical sensors.

12.4.1 Optical waveguides


A system designer must be aware of the fundamental bandwidth limitations in optical
waveguides. The most prominent limitation is dispersion, which represents the dependence
of the relative propagation constant k/k0 on frequency, where k0 is the phase constant in a
vacuum. Dispersion distorts the shape of the pulses transmitted through the waveguide. The
pulses are spread in time when traveling along the waveguide, so that they even cannot be
recognized at the output. Temporal spreading effectively establishes the maximum data rate
for a communication link. There are three types of dispersion: material dispersion, modal
dispersion, and waveguide dispersion. In material dispersion, different wavelengths of light
travel at different velocities. Consequently, the pulse effectively spreads out (or disperses) in
time and space. Modal dispersion arises in waveguides with more than one propagating
mode. Each allowed mode in the waveguide travels at a different group velocity. The pulse
energy in a waveguide is distributed among the various allowed modes, either through the
initial excitation, or through mode coupling that occurs within the waveguide. The modes
arrive at the end of the waveguide slightly delayed relative to each other. This effectively
spreads the temporal duration of the pulse, which again limits the bandwidth. Waveguide
dispersion is a more subtle effect. The propagation constant depends on the wavelength, so
even within a single mode different wavelengths propagate at slightly different speeds.
Waveguide dispersion is usually smaller than material and modal dispersion. However, in the
vicinity of the so called zero dispersion point for materials, waveguide dispersion can be the
dominant effect in a single mode system. Waveguide dispersion can be used to cancel
material dispersion, allowing the design of special dispersion shifted waveguides.
In Chapter 8 we studied basic dielectric waveguides: the dielectric layer – the
dielectric slab waveguide and the dielectric cylinder. In these two waveguides the core has a
constant permittivity. Such waveguides are known as step-index waveguides. A very common
parameter for characterizing waveguides is the numerical aperture NA. This concept is
based on ray tracing and refraction, so it is applicable to multimode waveguides. The
geometry of this problem is shown in Fig. 12.27. The numerical aperture is defined as the
sinus of the maximum angle θmax
under which the ray is coupled into n=1 fiber cladding n2
the waveguide. Let us assume a fiber
θ ≥ θc
with core refraction index n1,
refraction index of the surrounding θ inc ≤ θ max fiber core n1
material n2, and this fiber faces a
fiber cladding n2
medium with n = 1. The ray will be
guided if it strikes the fiber interface
at an angle greater than the critical Fig. 12.27

173
book 122
angle (3.68). Applying the Snell laws we get

NA = sin (θ max ) = n1 sin (90 − θ c ) = n1 cos(θ c ) =


. (12.36)
= n1 1 − sin 2 (θ c ) = n1 1 − n22 n12 = n12 − n 22

The numerical aperture is a useful parameter for large core multimode waveguides, namely
for describing the coupling of light into the fiber. The light to be coupled into the waveguide
must be focused in such a way that all incident rays lie within this angle θ max.
There are two ways to significantly reduce modal dispersion in a waveguide: by using
only single mode waveguides, or by using a graded-index waveguide. The single mode
waveguide appears to be the simplest, but it is not always a practical solution. It is very
difficult to couple light into single mode waveguides, as their core has small dimensions. The
second method for reducing modal dispersion is to use graded-index waveguides. These
waveguides can be made with relatively large dimensions, easing the coupling and alignment
problems common in single mode devices, and they can dramatically reduce modal
dispersion.
In a graded-index waveguide the permittivity is a n(x)
smooth function of position. At the waveguide center it has
the maximum value, and it decreases smoothly with
distance away from the central axis, see the sketch in Fig.
12.28. These waveguides can be analyzed as has been
described in Chapter 4. The path of the ray propagating in
the graded-index waveguide has been determined in
Example 4.1. It is described by a sinus function.
The modal dispersion arose due to the path
differences between the high order rays that followed a 0 x
long zigzag down the waveguide and the low order rays
that traveled straight. Fig. 12.29a illustrates this case for Fig. 12.28
two extreme modes. In the graded-index structure, a ray
traveling near the axis will higher order mode higher order mode
spend more time in a high
index material, Fig.
12.29b, and will travel
more slowly than will a
ray that is farther from the low order mode low order mode
axis. However, rays far step index graded index
from the axis follow a a b
longer sinusoidal path. Fig. 12.29
Through optimal
adjustment of the index gradient, it is possible to minimize the difference in the group delay
between the extreme rays. This will reduce modal dispersion and effectively increase the
information capacity of the waveguide.
Dispersion of the signal can sometimes be compensated or eliminated through clever
design, but attenuation simply leads to a loss of signal. Eventually the energy in the signal
becomes so weak that it cannot be distinguished with sufficient reliability from the noise
always present in the system. Attenuation therefore determines the maximum distance at
which an optical link can be operated without amplification. Attenuation arises from several
different physical effects. In an optical waveguide, one must consider: intrinsic material
absorption, absorption due to impurities, Rayleigh scattering, bending and waveguide
scattering losses, and microbending loss. In terms of priority, intrinsic material absorption

174
book 122
and Rayleigh scattering are the most serious causes of power loss for long distance systems.
Impurity absorption has become less of a problem as improved material processing techniques
have been developed. Total optical attenuation is characterized

Pout = Pin e −αz , (12.37)

where Pout and Pin are the output and input powers of the optical wave, respectively, and α is
the attenuation coefficient. The attenuation coefficient depends strongly on the wavelength
of the light and the material system involved.
The intrinsic material absorption arises from the atomic, molecular, and vibrational
transitions. These transitions can absorb electromagnetic energy from the applied field and
store it in the excited state. This energy is eventually dissipated through emission of a photon
or through the creation of lattice vibrations, and represents a loss to the electromagnetic field.
Sometimes the attenuation due to impurities is more severe than the attenuation of the basic
material itself. One of the problems of manufacturing optical glass fibers for long distance
communication systems is to provide material with sufficient purity. These losses have been
represented by complex permittivity (1.23) in Chapter 1. Similarly we can introduce the
complex index of refraction (3.11)

n = n’ – jn’’ . (12.38)

The imaginary part of n can lead to attenuation or gain, depending on its sign. In the case of a
passive medium it leads to attenuation and its value is positive. Inserting for wave
propagating constant k = k0n = k0(n’-n’’) we can describe the wave traveling along the
waveguide

E ( z ) = E0e − jkz = E0e − jk 0 (n '− jn '')z = E0e − k 0 n '' z e − jk 0 n ' z (12.39)

Rayleigh scattering is a fundamentally different attenuation mechanism. Instead of


light being absorbed and converted into stored energy within a medium, it is simply scattered
away from its original direction. This is the scattering of light off the random density
fluctuations that exist in a dielectric material. These losses are fundamental and cannot be
compensated or eliminated.
The present attenuation value of 0.2 dB/km in fused silica is the fundamental limit of
the performance of such glass fibers.
Optical power can be lost through leakage due to fiber bending. If the fiber is bent, the
spatial mode is not appreciably changed in shape compared to the straight fiber. However, the
plane wavefronts associated with the mode are now pivoted about the center of curvature of
the bend. To keep up with the mode, the phase front on the outside of the bend must travel a
little faster than the phase front in the core. At some critical distance from the core of the
fiber, the phase front will have to travel faster than the local speed of light, c/nclad. Since this
is not possible, the field beyond this critical radius breaks away and enters a radiating mode.
The power that breaks away is a loss to the waveguide. Microbending loss is caused by
putting an optical fiber in a cable and wrapping it about the central cord. The degree of
attenuation depends on the specific cabling geometry.
There can be a significant loss in optical connections due to misalignment or due to
mismatch between the two devices. Misalignment between a source and a single mode
waveguide in the case of dimensions less than 1 µm can cause a coupling loss exceeding
1 dB. Coupling problems are exaggerated by the small dimensions of the typical optical
waveguides and sources, which makes alignment a critical and challenging task. The

175
book 122
calculation of the coupling between two optical waveguides is based on a modal description
of the waveguides. The couple depends on alignment, dimension differences, and geometric
shapes. In the calculation process we have to include the forward waves of the waveguides,
the reflected waves and the radiation modes originating from both the reflected wave and the
transmitted field.
The crucial problem is to couple the light from a source into a waveguide. This
coupling is defined by the radiation pattern of the light source, which is defined by the angular
dependence of brightness B(ϕ,θ). Brightness is defined as the optical power radiated into a
unit solid angle per unit surface area. It is specified in watts per square centimeter per
steradian. Consider the case
in Fig. 12.30, showing an source radiation
optical source end-fire pattern
coupling onto the end of a
waveguide. The source optical source fiber
emits light into a cone, core acceptance
angle
which partially overlaps the cladding
numerical aperture of the lost power
waveguide. Any light
falling outside either the Fig. 12.30
numerical aperture or the
physical core dimensions obviously will not couple to the waveguide. The total power
coupled to the waveguide is, assuming waveguide circular symmetry, given by

rmin 2π  2πθ max 


P= ∫ ∫  ∫ ∫ B (ϕ , θ ) sin (θ ) dθ dϕ  r dr dϕ S (12.40)
0 00 0 

where the subscript S refers to the source, rmin is the smaller radius of either the fiber core or
the source. Brightness B is here integrated over the acceptance solid angle of the fiber. The
maximum acceptance angle θmax is defined through the numerical aperture (12.36).

12.4.2 Optical detectors


The minimum received power necessary to achieve the desired quality of information
in a communication link is established by noise. There are many ways to detect light. Any
process which converts optical energy into another useful form of energy can be considered to
be a detector. For optoelectronic systems, the most useful detectors are those that convert
optical energy directly into electric current or voltage. There are two fundamental classes of
detector: quantum, or photon, detectors, which respond to the number of photons that are
absorbed, and thermal detectors, which respond to the energy that is absorbed.
The quantum detector measures the number of photons received per second N, which
is proportional to the power of the optical signal P

ηP ηPλ
N= = , (12.41)
hf hc

where η is the quantum efficiency of the detector, defined as the probability that a free
electron or hole is generated per absorbed photon. h is Planck’s constant, hf is the energy of
the photon. The number of received photons increases with the wavelength up to the threshold
at which the energy of the photon is not sufficient to excite a free electron or hole.

176
book 122
An ideal thermal detector responds to the total power, independent of the wavelength.
Real detectors have spectral response curves which are limited only by the spectral absorption
of their coating. The absorbed optical power is converted to heat. In general, thermal detectors
have a much slower response time than most quantum detectors. Optical communication
systems almost exclusively use quantum detectors.
The quality of a signal is degraded by noise, i.e., random fluctuations that cannot be
distinguished from the signal. There are many sources of noise. Shot noise is a fundamental
noise that exists in all optical detection processes. Its origin is in the randomness of the arrival
of photons in the signal. It is fundamental and, consequently, there is no way to avoid it. The
absolute noise value is rarely a concern – normally we are worried about the relative noise
compared to the signal. The significant parameter is the signal-to-noise ratio - S/N.
Four basic parameters characterize the performance of an optical detector. These are:
responsivity, spectral response, detectivity, and time response. Responsivity tells us how
much signal is obtained per unit optical power. It is measured in A/W or in V/w. Detectivity
tells us the minimum detectable power required to achieve S/N = 1. The quantum detectors
used in optical communication systems work on the basis of the internal photoeffect. In this
case a photon creates a free charge carrier in the material. The creation of excess carriers leads
to a change in conductivity, p-n junction voltage, or junction currents. There are three major
categories of optoelectronic detectors: photoconductors, pin diodes, and avalanche
photodiodes.
Photoconductive detectors employ semiconductor materials with a bandgap suited to
the wavelength of the light that is being detected. A photon with sufficient energy can lift an
electron from the valence band to the conducting band, and thus increases the conductivity. p-
n junction detectors are very often used in optical systems. A pair electron-hole is generated
by a photon in the depletion layer. It is separated by an internal field and contributes to the
current excited in this way by the incident light. Adding an intrinsic layer between the p- and
n-layers we get a pin photodiode. The intrinsic layer widens the depleted layer and in this
way reduces the diode capacitance and shortens the diode time response. An avalanche
photodiode uses the internal multiplication effect to achieve higher detectivity than pin
photodiodes. The structure of this diode is designed so that the extra electron-hole pairs
excited by the detected light are accelerated through the depletion layer by the intense electric
field as they are able to generate extra free electron-hole pairs. These subsequent electrons
and holes are themselves accelerated, and can lead to further pair creation. As a result of this
impact ionization, the total current in the photodiode is greater than would be produced by
photoionization alone.

12.4.3 Optical amplifiers and sources


A photon with sufficient energy absorbed in a material can excite an atom from its
basic energetic state to the upper excited stage. Spontaneous emission of radiation occurs
when an excited atom spontaneously relaxes back to the lower basic state, and in the process
emits a single photon. This effect of the spontaneous emission of light radiation is used in
standard light emitting diodes. The electromagnetic field of the incident optical wave can
induce a transition between the upper and lower state, and the atom gives up one quantum of
energy to the field. This is known as stimulated emission of radiation. Optical amplifiers and
lasers are based on the stimulated emission of radiation. Let us take a monochromatic optical
beam that propagates through a material in the z direction. Its electrical field then varies as

E ( z ) = E (0 ) e gz , (12.42)

where g is the optical gain. This quantity is proportional to the difference between the
populations of atoms in the lower basic state N1 and the atoms in the upper excited state N2

177
book 122
g = σ ( N 2 − N1 ) , (12.43)

where σ is a constant. From (12.43) it follows that for N2 > N1 we have g > 0. This condition
is known as the inversion population of states, and the optical beam is amplified due to the
stimulated emission. The inversion population can be obtained by adding the necessary
energy from outside, depending on the kind of active material – pumping the laser or the
amplifier.
The simplest optical amplifier compatible with optical communication links using
optical fibers is the erbium-doped fiber optical amplifier. This is a standard optical fiber with
the core doped by erbium. Erbium ions in the glass lattice assure appropriate energy levels.
This active material is pumped optically by
irradiating, e.g., by a tungsten bulb. The energy of the E3
photons of the pumping light must be higher than the E2
energy of the photons of the amplified light. The pump stimulated
pumping assures transition of atoms from the basic emission
energy state E0 to state E3, see Fig. 12.31, and the E0 E1
stimulated emission then goes from state E2 to state
E1. The spontaneous emission in the optical amplifier Fig. 12.31
is the source of the noise added to the amplified
signal.
Lasers are optical oscillators. The acronym laser stands for light amplification by the
stimulated emission of radiation. Like other oscillators, the laser is an optical amplifier with
positive feedback. Amplification is assured by the gain, which is established through
population inversion due to pumping an active medium. Positive feedback is accomplished in
the optical frequency range using mirrors appropriately aligned to resonate one, or a few,
cavity modes. Part of the generated power is in this way selectively returned back to the input
of the amplifier.
In optoelectronics, lasers tend to be based on waveguide structures which are already
one-dimensional in nature. The spatial confinement of the oscillating mode is provided by the
structure of the device. The longitudinal mode selection and feedback is accomplished by an
open resonator known as the Fabry-Perot resonator, Fig. 12.32. It consists of two mirrors with
reflection coefficients R1 and R2, separated by
R1 R2
distance d. Only light with an integer number of
half-wavelengths in the cavity will be resonant in d
the structure. The field incident to the mirrors is
partially reflected, assuring the feedback, and mirror mirror
partially transmitted, providing the laser output.
The wave reflected back is amplified, which Fig. 12.32
maintains the oscillations in the cavity.
Modern optoelectronics is based on the application of semiconductor injection lasers.
These lasers are compatible with driving electronic circuits. They are efficient, small and
cheap. The light emitted by these lasers can be relatively simply coupled to optical fibers. The
inversion population of states is achieved by injecting free charge carriers across two
heterojunctions into a narrow active layer. For this reason, these lasers are known as double
heterostructure semiconductor lasers. Semiconductor physics is beyond the scope of this
course.

12.4.4 Optical modulators and sensors


There are two common methods for encoding a signal onto an optical beam: either
directly modulate the optical source, or externally modulate a continuous wave optical source.
Direct modulation is the most widespread method of modulation today, but it introduces

178
book 122
demanding constraints on semiconductor lasers. For example, it is difficult to directly
modulate a semiconductor laser at frequencies above a few GHz. Furthermore, it is difficult to
maintain single mode operation of these pulsed lasers. Multiple mode lasers have a larger
spectral bandwidth, which leads to increased pulse spreading due to dispersion. External
modulators offer several advantages over direct modulation. First, one can use a relatively
simple and inexpensive continuous wave laser as the primary optical source. Second, since a
modulator can encode information based on a number of externally controlled effects, it is not
compromised by the need to maintain population inversion or single mode control.
The simplest electrooptic modulator is based on the linear electrooptic effect, or
Pockel’s effect. This represents the change in the refractive index due to an externally applied
electric field. Devices which directly modulate the phase, the intensity, or the polarization of
the light are designed with the use of the Pockel’s effect. The phase modulator is simple. The
external electric field changes the refractive index, and in this way the phase velocity of the
wave propagating through the modulator. This causes changes in the output signal. The phase
modulation introduced by the electrooptic effect can be used to create intensity modulation
via changes in polarization and through interferometric effects. Polarization modulation can
be achieved using the differential retardation between two orthogonal polarizations of the
optical wave. To convert polarization modulation into intensity modulation, it is necessary to
run the output through a linear polarizer. Phase modulation can be converted into intensity
modulation through constructive interference between two waves. The Fabry-Perot
interferometer and the Mach-Zender interferometer represent two examples for converting
phase modulation into intensity modulation. Fig. 12.33 shows a schematic Mach-Zender
interferometer. The single mode waveguide input is split into two single mode waveguides by
a 3 dB Y junction. The split beams
3 dB coupler
travel different paths, and then
recombine at another Y junction. The
relative phase difference of the two input
beams can be electrooptically V
controlled by applying a voltage to output
E
the center electrode in the structure
shown in Fig. 12.33. Because the
change in refractive index n depends
on the direction of the applied electric Fig. 12.33
field, index n increases in one arm
and decreases in the other arm. This differential change in index is used to alter the relative
phase of the recombining fields and thus the amplitude of the output wave.
Like electrooptic modulators, acoustooptic modulators control the transmission of
light by local changes in the refractive index of the transmission medium. The modulation
occurs by means of a traveling sound wave which induces stress related modifications of the
local refraction index. Optical wave interaction can be produced by either bulk acoustic waves
traveling in the volume of the material, or by surface acoustic waves which propagate on the
surface within approximately one acoustic wavelength of the surface. Surface acoustic wave
devices are well suited to integrated optics applications because the energy of the acoustic
field is concentrated in the region of the optical waveguide.
Anything that can perturb the optical beam in a fiber can be exploited to make a
sensor. Common interactions are length and refractive index modification through stress,
strain, or temperature. The designer must know how to make the fiber interact selectively with
the measured quantity of interest. An optical sensor can be based on changes in intensity,
polarization, phase, wavelength, and direction of propagation. In extrinsic sensors the light
leaves the fiber and is modulated prior to being recoupled onto a fiber. In an intrinsic sensor
the light is modulated inside the fiber.

179
book 122
Three basic effects are used to intensity modulate the optical signal: mechanical, based
on movement of an aperture or object which interferes with the optical path; refractive, where
the cladding layer of the fiber is modified in such a way as to cause the guided light to leave
the core; and absorptive, where the optical signal is absorbed by a substance. The critical
weakness of the intensity sensor is that the system has no way of knowing if a change in
intensity is in response to a change of the measured quantity, or if it is due to some other
cause.
This weakness can be reduced by using a phase modulation index based, e.g., on the
Mach-Zender interferometer, see Fig. 12.33. The measured physical quantity changes the
phase of the optical wave in one arm of the interferometer. This causes a change in the output
signal.

13. MATHEMATICAL APPENDIX


This section summarizes the basic knowledge of mathematics necessary to understand
the text.

Computation with vectors:


A vector quantity is written as a linear combination of basic vectors x0, y0, z0

A = Ax x 0 + Ay y 0 + Az z 0 . (13.1)

We add vectors by adding their coordinates

A + B = ( Ax + B x ) x 0 + (Ay + B y )y 0 + ( Az + B z ) z 0 . (13.2)

The modulus of a vector is

A = A = Ax2 + Ay2 + Az2 . (13.3)

The scalar product of two vectors is

A ⋅ B = AB cos(α ) = Ax B x + Ay B y + Az B z = B ⋅ A , (13.4)

where α is the angle between the two vectors A and B. It follows from definition (13.4) that
the scalar product of two vectors equals zero for perpendicular vectors. The scalar
product is linear

A ⋅ (B + C) = A ⋅ B + A ⋅ C , (13.5)

From (13.4) we can determine the angle between vectors

A⋅B
α = arccos . (13.6)
AB

180
book 122
According to Fig. 13.1 the scalar product determines the projection
of vector A in the direction of vector B. Let n be a unit vector A
normal to a surface, then it follows from Fig. 13.1 that the Asin(α)
component of A normal to the surface is defined by α •
An = n ⋅ A . (13.7) Acos(α) B

Fig. 13.1
The vector product of two vectors is defined

A × B = AB sin (α ) i n = −B × A , (13.8)

where in is a unit vector perpendicular to the plane in which the two vectors lie. The vector
product can be calculated in the rectangular coordinate system as

x0 y0 z0
A × B = Ax Ay Az = (Ay B z − Az B y )x 0 + ( Az B x − Ax B z ) y 0 + (Ax B y − Ay B x )z 0 . (13.9)
Bx By Bz

If the vector product of two nonzero vectors equals zero, it


follows from (13.8) that these two vectors are parallel. From Fig. A×B
13.1 it is clear that the value of the vector product determines the A
projection of vector A into the direction perpendicular to vector α
B. Therefore using the vector product we can compute the
tangential component of a vector to the plane determined by unit B
normal vector n
Fig. 13.2

At = n × A . (13.10)

The value of the vector product (13.8) equals the area of


the parallelogram in Fig. 13.2. The vector product is linear
A
as the scalar product (see (13.5)).
B
The double vector product is defined as

A × (B × C) = B(A ⋅ C) − C(A ⋅ B ) . (13.11) C

The mixed product is defined as Fig. 13.3

Ax Ay Az
A ⋅ (B × C) = B ⋅ (C × A ) = C ⋅ (A × B ) = B x By Bz . (13.12)
Cx Cy Cz

The value of the mixed product equals the volume of the parallelepiped shown in Fig. 13.3.

Complex numbers:
The basic knowledge is, see Fig. 13.4 defining particular quantities,

181
book-13
 y
z = x + jy = ρ e jϕ , ρ = x 2 + y 2 , ϕ = arctg   , (13.13)
 x

e jϕ = cos(ϕ ) + j sin (ϕ ) . (13.14)


z
Goniometric functions: jy
Here is a list of basic formulas ρ
ϕ
sin 2 (α ) + cos 2 (α ) = 1 , (13.15)
x
1
sin 2 (α ) = [1 − cos(2α )] , (13.16) Fig. 13.4
2

1
cos 2 (α ) = [1 + cos(2α )] , (13.17)
2

sin (α ± β ) = sin (α ) cos(β ) ± cos(α ) sin (β ) , (13.18)

cos(α ± β ) = cos(α ) cos(β ) m sin (α ) sin (β ) , (13.19)

α ± β  α m β 
sin (α ) ± sin (β ) = 2 sin   cos  , (13.20)
 2   2 

α + β  α − β 
cos(α ) + cos(β ) = 2 cos  cos  , (13.21)
 2   2 

α + β  α − β 
cos(α ) − cos(β ) = −2 sin   sin   , (13.22)
 2   2 

 B
A sin (ωt ) + B cos(ωt ) = C sin (ωt + ϕ ) , C = A 2 + B 2 , ϕ = arctg  , (13.23)
 A

1 jx
sin ( x ) =
2j
(e − e − jx ) , (13.24)

1 jx
cos( x ) =
2
(
e + e − jx .) (13.25)

Hyperbolic functions:

1 x
sinh ( x ) =
2
(
e − e −x ,) (13.26)

1 x
cosh ( x ) =
2
(
e + e−x . ) (13.27)

182
book-13
cosh 2 ( x ) = sinh 2 ( x ) + 1 , (13.28)

sinh ( x ± y ) = sinh ( x ) cosh ( y ) ± cosh ( x )sinh ( y ) , (13.29)

cosh ( x ± y ) = cosh ( x ) cosh ( y ) ± sinh (x ) sinh ( y ) . (13.30)

From definitions of hyperbolic functions (13.26) and from the definitions of goniometric
functions (13.24) and (13.25) it follows

sin ( jx ) = j sinh ( x ) , cos( jx ) = cosh (x ) , (13.31)

sin ( x + jy ) = sin ( x ) cosh ( y ) + j cos( x )sinh ( y ) , (13.32)

cos( x + jy ) = cos( x ) cosh ( y ) − j sin ( x )sinh ( y ) , (13.33)

sin (2 x ) + j sinh (2 y )
tg ( x + jy ) = , (13.34)
cos(2 x ) + cosh (2 y )

sinh ( x + jy ) = sinh ( x ) cos( y ) + j cosh ( x )sin ( y ) , (13.35)

cosh ( x + jy ) = cosh ( x ) cos( y ) + j sinh ( x ) sin ( y ) . (13.36)

Bessel’s functions:
A linear combination of Bessel’s functions Jn(x) and Yn(x) of the n-th order is a
solution of the differential equation

x 2 y ' '+ x y '+(x 2 − n 2 ) y = 0 , (13.37)

y = C1 J n ( x ) + C 2 Yn ( x ) . (13.38)

Functions Jn(x) are defined by the sum

n+2k


(− 1)k  x 
J n (x ) = ∑  2 . (13.39)
k = 0 k!Γ(n + k + 1)

Functions Yn(x) follow from Jn(x) by the formula

J n ( x ) cos(nπ ) − J − n ( x )
Yn ( x ) = . (13.40)
sin (nπ )

The integrals of Bessel’s functions are

∫x
n
J n−1 ( x ) dx = x n J n ( x ) , ∫x
−n
J n +1 ( x ) dx = − x − n J n ( x ) . (13.41)

183
book-13
Plots of these functions of the three lowest orders are shown in Fig. 13.5 and Fig. 13.6.

1
J0 , J1, J2

Y 0, Y 1, Y 2
J0 0.4 Y0
0.8
0.2 Y1
J1
0.6
J2 0
0.4 Y2
-0.2
0.2
-0.4
0 -0.6
-0.2 -0.8
-0.4 -1
0 2 4 6 8 10 0 2 4 6 8 10
x x
Fig. 13.5 Fig. 13.6

To calculate the propagation constants of modes propagating in a waveguide with a


circular cross-section, see paragraph 7.3, we need to know the zero points αmn of the Bessel
'
function Jm and the zero points α mn of the derivative of this function. The zero points αmn of
'
the Bessel function Jm are listed in Tab. 13.1. The zero points α mn of the derivative J m' are
listed in Tab. 13.2.

m 0 1 2 3 4 5
n=1 2.40482 3.83171 5.13562 6.38016 7.58834 8.77148
n=2 5.52007 7.01559 8.41724 9.76102 11.06471 12.33860
n=3 8.65372 10.17347 11.61984 13.01530 14.37254 15.70017
n=4 11.79153 13.32369 14.79595 16.22347 17.61597 18.98013
n=5 14.93091 16.47063 17.95980 19.40942 20.82693 22.21780

Tab. 13.1 - αmn

m 0 1 2 3 4 5
n=1 3.83170 1.84118 3.05424 4.20119 5.31755 6.41562
n=2 7.01558 5.33144 6.70613 8.01524 9.28240 10.51986
n=3 10.17346 8.53632 9.96947 11.34592 12.68191 13.98719
n=4 13.32369 11.70600 13.17037 14.58585 15.96411 17.31284
n=5 16.47063 14.86359 16.34752 17.78875 19.19603 20.57551
'
Tab. 13.2 - α mn

The asymptotic formulas of Bessel’s functions Jn(x) and Yn(x) for x increasing to infinity are

2   1π 
J n (x ) = cos  x −  n +   , (13.42)
πx   2 2

2   1π 
Yn (x ) = sin  x −  n +   , (13.43)
πx   2 2

Hankel’s functions (Bessel’s functions of the 3rd kind):


184
book-13
The solution of equation (13.37) is any linear combination of Bessel’s functions Jn and
Yn. This enables us to introduce Hankel’s functions by

H n1 ( x) = J n ( x ) + jYn ( x ) , (13.44)

H n2 ( x) = J n ( x ) − jYn (x ) . (13.45)

Their approximate values for a high argument are

2 j ( x −π / 4−πn / 2 )
H n1 ( x ) ≈ e , (13.46)
πx

2 − j ( x −π / 4−πn / 2 )
H n2 ( x ) ≈ e . (13.47)
πx

It is evident from (13.44) and (13.45) that the relations between Hankel’s functions and
Bessel’s functions Jn and Yn are similar to the relations between the exponential function of a
complex argument and goniometric functions. Bessel’s functions Jn and Yn describe solutions
of the wave equation cylindrical as standing waves, whereas Hankel’s functions describe a
traveling wave.
Hankel’s functions of the order ½ are defined by

2 jx
H 11/ 2 ( x ) = − j e , (13.48)
πx

2 − jx
H 12/ 2 ( x ) = j e . (13.48)
πx

Modified Bessel’s functions:


Modified Bessel’s functions are Bessel’s functions of an imaginary argument.
Inserting into (13.37) x = jz we get

( )
z 2 y ' '+ zy '− z 2 + n 2 y = 0 . (13.49)

The solution of this equation is

y = A I n (z ) + B K n (z ) . (13.50)

The modified Bessel’s function of the first kind is defined

I n ( z ) = j − n J n ( jz ) , (13.51)

which gives a real value for real z. Similarly

π
K n ( z ) = j n+1 H n1 ( jz ) (13.52)
2

185
book-13
is the modified Bessel’s function of the second kind, again real for real z. This function can be
defined by function In

π I −n (z ) − I n (z )
K n (z ) = . (13.53)
2 sin (nπ )

Their approximate values for a high argument are

ez
I n (z ) = , (13.54)
2πz

π
K n (z ) = e−z . (13.55)
2z

Plots of modified Bessel’s functions of the two lowest orders are shown in Fig. 13.7 and Fig.
13.8.
30 4
I0 , I 1

K0, K1
25
3
20
I0
15 2
I1 K1
10
1
5 K0

0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
x x

Fig. 13.7 Fig. 13.8

Coordinate systems:
We use the three basic coordinate systems: rectangular, cylindrical and spherical.
In the rectangular coordinate system each point is determined by three coordinates
x, y, z, which represent the corresponding segments on the axes, see Fig. 13.9. Elemental
displacements in the directions of the axes are dx, dy and dz. The volume element is
dV=dx.dy.dz.
The cylindrical coordinates are r, α, z, see Fig. 13.10. r represents the perpendicular
distance from the axis z, α is the angle measured in the plane x y starting from the positive x
direction, and z has the same meaning as in the rectangular system. The cylindrical
coordinates are bound with the rectangular coordinates by the relations

x = r cos(α ) ,

y = r sin (α ) , (13.56)

z=z .

Elemental displacements in the directions of the axes are dr, rdα and dz. The volume element
is dV=dr.r.dα.dz. If the integrated function depends only on r, we can use the volume element

186
book-13
in the form dV=2πrldr, which is the volume between two coaxial cylinders with radii r and
r+dr of length l.
The spherical coordinates are r, α, ϑ, see Fig. 13.11. r represents the distance from
the origin, α is the angle measured in the plane x y starting from the positive x direction, and
ϑ is the angle measured from the positive z direction. The spherical coordinates are bound
with the rectangular coordinates by the relations

x = r cos(α ) sin (ϑ ) ,

y = r sin (α ) sin (ϑ ) , (13.57)

z = r cos(ϑ ) .

z z z
z0 z0 α0
α0
r0
A(x,y,z)
x0 A(r,α,z) A(r,ϑ,α)
y0 ϑ
z r0 ϑ0
r
r
y α z y α y
x
y
x x x
Fig. 13.9 Fig. 13.10 Fig. 13.11

Elemental displacements in the directions of the axes are dr, rsin(ϑ)dα and rdϑ. The volume
element is dV=r2.sin(ϑ).dr.dα.dϑ. If the integrated function depends only on r, we can use the
volume element in the form dV=4πr2dr, which is the volume between two concentric spheres
with radii r and r+dr.

The flux of a vector (a surface integral):


Let J be the distribution of electric current. The total current passing surface S can be
calculated in the following way. We discretize the J
surface, Fig. 13.12, to elementary surfaces dS
S
determined by vectors dS=dSn, where n is the unit n
normal vector. The current passing the elemental
dS
surface dS is

dI = J n dS = J ⋅ n dS = J ⋅ dS . dS

The total current is the summation of dI, which is a


surface integral

I = ∫∫ J ⋅ dS . (13.58) Fig. 13.12


S

In this way we have presented the physical meaning of the surface integral from the scalar
product of a vector and the vector of a surface element. In the case of a closed surface we

187
book-13
have the notation

I = ∫∫ J ⋅ dS .
S

Path integral: F
We demonstrate the physical meaning of this
integral by calculating the work performed by force F dl t0
along path c. We divide the path into elemental arcs of
length dl. The work performed along an arch is, see Fig.
13.13, dl
c
dW = Ft dl = F ⋅ t 0 dl = F ⋅ dl . Fig. 13.13

The work performed by force F along the whole path c is the summation of dW, which is an
integral

W = ∫ F ⋅ dl . (13.59)
c

In the case of a closed path the integral is known as the circulation of vector F

W = ∫ F ⋅ dl . (13.60)
c

Gradient of a scalar function:


The gradient of a scalar function is a vector which determines the direction in which a
function value increases at maximum speed, and the modulus of this vector determines the
magnitude of this speed. This vector is perpendicular to the planes of the constant function
values. In the rectangular coordinate system the gradient is defined by

∂ϕ ∂ϕ ∂ϕ
grad ϕ = x0 + y0 + z0 . (13.61)
∂x ∂y ∂z

The increase of a scalar function corresponding to a shift by dr can be determined neglecting


differentials of higher orders

ϕ (r + dr ) = ϕ ( x + dx, y + dy, z + dz ) =
∂ϕ ∂ϕ ∂ϕ ,
= ϕ ( x, y , z ) + dx + dy + dz + ...
∂x ∂y ∂z

dϕ = ϕ ( x + dx, y + dy, z + dz ) − ϕ ( x, y, z ) ≈
∂ϕ ∂ϕ ∂ϕ
≈ dx + dy + dz = grad ϕ ⋅ dr = grad ϕ dr cos(Θ )
∂x ∂y ∂z

So the increase of a function is greatest in the direction of gradϕ. The gradient is defined by
partial derivatives, so it is a linear operator

188
book-13
grad(λ1ϕ 1 + λ 2ϕ 2 ) = λ1 grad ϕ1 + λ2 grad ϕ 2 , (13.62)

where λ1 and λ2 are two scalar constants. For the product of two functions we have

grad(ϕψ ) = ϕ gradψ + ψ grad ϕ . (13.63)

In the literature we can find the symbolic vector ∇ , known as Hamilton’s “nabla” operator. In
the rectangular coordinate system this operator is defined as

∂ ∂ ∂
∇= x0 + y 0 + z 0 . (13.64)
∂x ∂y ∂z

Using this operator the gradient can be expressed as

grad ϕ = ∇ϕ . (13.65)

Divergence of a vector function:


The divergence of vector function J is defined as the volume density of this vector
quantity flux flowing out of a point, so it characterizes the sources of a vector,

1 
div J = lim
V →0 V

∫∫ J ⋅ dS  ,
S
(13.66)

where S is the boundary of volume V. In the case of J being a current density, divJ
corresponds to the current flowing out of a unit volume. The Gauss-Ostrogradsky theorem
follows from this

∫∫ J ⋅ dS = ∫∫∫ div J dV
S V
. (13.67)

In the rectangular coordinate system the divergence can be calculated as

∂J x ∂J y ∂J z
div J = + + . (13.68)
∂x ∂y ∂z

This operator is a linear operator, as it is defined by partial derivatives,

div(λJ 1 + µJ 2 ) = λ div J 1 + µ div J 2 , (13.69)

where λ and µ are scalar constants. The divergence applied to the product of scalar function f
and vector function J is

div( f J ) = grad f ⋅ J + f div J . (13.70)

The divergence applied to the vector product of two vectors is

div(A × B ) = B ⋅ rot A − A ⋅ rot B . (13.71)

189
book-13
The divergence applied to the gradient of a scalar function defines Laplace’s operator

∂ 2ϕ ∂ 2ϕ ∂ 2ϕ
div(grad ϕ ) = ∆ϕ = + + . (13.72)
∂x 2 ∂y 2 ∂z 2

The divergence can be expressed using Hamilton’s operator

div J = ∇ ⋅ J . (13.73)

Rotation of a vector function:


The rotation of a vector function is defined as the surface density of this vector
circulation

1 
rot F = r0 max lim ∫ F ⋅ dl  , (13.74)
S →0 S
 c 

where c is a curve that describes the boundary of surface S, and r0 is a unit vector. This vector
determines the direction of vector dS and is selected to obtain the maximum value of
1 
lim ∫ F ⋅ dl  . As the rotation represents the surface density of a circulation, Stokes’
S →0 S
 c 
theorem follows from the definition of the rotation

∫ F ⋅ dl = ∫∫ rot F ⋅ dS .
c S
(13.75)

In the rectangular coordinate system the operator rotation is defined as a symbolic


determinant

x0 y0 z0
∂ ∂ ∂
rot F = . (13.76)
∂x ∂y ∂z
Fx Fy Fz

This operator is again a linear operator, as it is defined by partial derivatives,

rot (λF1 + µF2 ) = λ rot F1 + µ rot F2 . (13.77)

The rotation applied to the product of a scalar function and a vector function is

rot ( f F ) = f rot F + grad f × F . (13.78)

There are mathematical identities

rot (grad f ) = 0 , (13.79)

div(rot F ) = 0 , (13.80)

190
book-13
rot (rot F ) = grad(div F ) − ∇ 2 F . (13.81)
In the rectangular coordinate system we have ∇ 2 = ∆ The rotation can be expressed by
Hamilton’s operator as
`
rot F = ∇ × F , (13.82)

which follows from (13.63) and (13.9).

Differential operators in the cylindrical coordinate system:

∂ϕ 1 ∂ϕ ∂ϕ
grad ϕ = r0 + α0 + z0 , (13.83)
∂r r ∂α ∂z

∂F
div F =
1 ∂
(r Fr ) + 1 α + ∂Fz , (13.84)
r ∂r r ∂α ∂z

 1 ∂Fz ∂Fα   ∂F ∂F  1∂ ∂F 


rot F =  −  r0 +  r − z  α 0 +  (r Fα ) − r  z 0 , (13.85)
 r ∂α ∂z   ∂z ∂r  r  ∂r ∂α 

1 ∂  ∂ϕ  1 ∂ 2ϕ ∂ 2ϕ
∆ϕ = r + + . (13.86)
r ∂r  ∂r  r 2 ∂α 2 ∂z 2

Differential operators in the spherical coordinate system:

∂ϕ 1 ∂ϕ 1 ∂ϕ
grad ϕ = r0 + α0 + ϑ0 , (13.87)
∂r r sin ϑ ∂α r ∂ϑ

1 ∂ 2 1 ∂Fα ∂
div F =
r ∂r
2
(
r Fr + ) +
1
r sin ϑ ∂α r sin ϑ ∂ϑ
(sin ϑ Fϑ ) , (13.88)

1 ∂ ∂Fϑ  1 ∂ ∂Fr 
rot F = (sin ϑ F ) − r + (r F ) − α0 +
r sin ϑ  ∂ϑ ∂α  r  ∂r ∂ϑ 
α 0 ϑ

, (13.89)
 1 ∂Fr 1 ∂ 
+ − (r Fα ) ϑ0
 r sin ϑ ∂α r ∂r 

1 ∂  2 ∂ϕ  1 ∂ 2ϕ 1 ∂  ∂ϕ 
∆ϕ =  r  + + 2  sin ϑ  . (13.90)
r ∂r  ∂r  r sin ϑ ∂α
2 2 2 2
r sin ϑ ∂ϑ  ∂ϑ 

14. BASIC PROBLEMS


This section lists the basic problems collected throughout the text. The ability to answer them
verifies the student’s knowledge and readiness to sit for the exam.

1 Maxwell’s equations in a differential form for a time-harmonic field expressed by

191
book-13
phasors.
2 Maxwell’s equations in an integral form for time-harmonic field expressed by phasors.
3 Boundary conditions of tangential components of E, H in a nonstationary field.
4 Boundary conditions of normal components of E, H in a nonstationary field.
5 Electromagnetic field on a surface of an ideal conductor.
6 Poynting’s vector. Definition and expression using field vectors.
7 Energy balance of active power. Physical meaning of particular items.
8 Energy balance of reactive power. Physical meaning of particular items.
9 The wave equation for E or H in a general material outside of a source region. Time-
harmonic field. Expression using phasors.
10 Continuity equation for harmonic field
11 Description of vectors E and H by potentials A and ϕ in time varying field.
12 Wave equations for potentials, sources assumed.
13 General solution of wave equations for potentials, harmonic field is assumed.
14 The expression of E of a time-harmonic plane wave in a general material. Phasor and
instantaneous value. Meaning of particular items.
16 What is a surface of constant amplitude and constant phase in a plane wave? What are a
uniform wave and a nonuniform wave?
17 What is a phase velocity? How is it defined?
18 What is a group velocity? How is it defined?
19 Draw the orientation of E, H and k in a plane electromagnetic wave. What is the
relation of these vectors?
20 What is the wave impedance of a general material? How is it defined?
21 The expressions for k, vf and Z in an ideal dielectric.
22 The expressions for k, vf and Z in a good conductor.
23 What is a penetration depth? How is it defined?
24 Active power transmitted by a plane electromagnetic wave through a surface 1 m2.
25 What is the polarization of an electromagnetic wave? Which kinds of polarization of an
electromagnetic wave are there?
26 Under which conditions can two linearly polarized waves create a wave with linear,
circular and elliptical polarization?
27 Equation of eikonal.
28 Differential equation describing the ray.
29 Coefficient of reflection for electric field in a perpendicular incidence, general
materials.
30 Coefficient of transmission for electric field in a perpendicular incidence, general
materials.
31 Coefficients of reflection and transmission for a perpendicular incidence, lossless
dielectrics.
32 What is the standing wave?
33 What is the standing wave ratio, its relation to R?
34 What is λ/4 transformer?
35 Snell’s laws.
36 What is the Brewster’s polarization angle, expression?
37 Draw a plot of R=f(ϕi) for ε1<ε2 for both horizontal and vertical polarization.
38 Draw a plot of R=f(ϕi) for ε1>ε2 for both horizontal and vertical polarization.
39 What is the total reflection, the condition for it?
40 What is the surface wave?
41 What is the character of field in the second material in the case of the total reflection?
42 Write a formula describing the spherical wave.
43 Write a formula describing the cylindrical wave.

192
book-13
44 What transmission line can transmit TEM wave?
45 Cut-off frequency of a mode in a parallel plate waveguide.
46 Cut-off wavelength of a mode in a parallel plate waveguide.
47 Sketch the field distribution of the TEM wave propagating in a parallel plate
waveguide.
48 Phase velocity of a wave propagating between two parallel plates.
49 Wavelength of a wave propagating between two parallel plates.
50 Propagation constant of a wave propagating between two parallel plates.
51 Wave impedance of a TM and TE waves propagating between two parallel plates.
52 What mode is dominant in the rectangular waveguide?
53 Cut-off frequency of a mode in the rectangular waveguide.
54 Cut-off wavelength of a mode in the rectangular waveguide.
55 Cut-off frequency of the dominant mode in the rectangular waveguide.
56 Cut-off wavelength of the dominant mode in the rectangular waveguide.
57 Phase velocity of a wave propagating in a rectangular waveguide.
58 Wavelength of a wave propagating in a rectangular waveguide.
59 Propagation constant of a wave propagating in a rectangular waveguide.
60 Wave impedance of a TM and TE waves propagating in a rectangular waveguide.
61 Power carried by the dominant mode in a rectangular waveguide.
62 What transmission line can transmit the TEM wave?
63 Cut-off frequency of a TM mode in the circular waveguide.
64 Cut-off frequency of a TE mode in the circular waveguide.
65 What mode is dominant in the circular waveguide?
66 Phase velocity of a wave propagating in a circular waveguide.
67 Wavelength of a wave propagating in a circular waveguide.
68 Propagation constant of a wave propagating in a circular waveguide.
69 Wave impedance of a TM and TE waves propagating in a circular waveguide.
70 Explain the guiding of waves in a dielectric waveguide.
71 Draw the plot of Ey field as a function of the transversal coordinate in the dielectric
layer for the first three TE modes.
72 Which parameters determine the resonant frequency of a cavity resonator?
73 Resonant frequency of a rectangular cavity resonator.
74 The telegraph equations for time-harmonic u and i on a TEM transmission line. The
meaning of particular items.
75 The characteristic impedance of a transmission line - the lossy and lossless case.
Expression and the definition.
76 The input impedance of a lossy and a lossless line of length L terminated by an
impedance ZL.
77 The input impedance of a lossless line of length L with a short and open end.
78 Draw lines of constant amplitude of a reflection coefficient in the Smith Chart.
79 Draw lines of constant phase of a reflection coefficient in the Smith Chart.
80 Draw lines of constant real part or a normalized impedance in the Smith Chart.
81 Draw lines of constant imaginary part or a normalized impedance in the Smith Chart.
82 Characteristic regions around the radiating elementary electric dipole.
83 Which components of electric and magnetic field carry power from the radiating
electric dipole?
84 Draw radiation patterns of the radiating elementary electric dipole.
85 What is the radiating resistance of the dipole?
86 What is the antenna directivity?
87 What is the antenna gain?
88 Relation between E and D in anisotropic material.

193
book-13
89 Relation between B and H in anisotropic material.
90 Tensor of permeability of magnetized ferrite.
91 Tensor of permittivity of a magnetized plasma.
92 What is the ferromagnetic resonance?
93 What is the Faraday effect?
94 What is the ordinary wave?
95 What is the extraordinary wave?

14. LIST OF RECOMMENDED LITERATURE


C. A. Balanis: Advanced Engineering Electromagnetics, John Wiley & Sons, Inc., New York,
USA, 1989.
J. D. Jackson: Classical Electrodynamics, 3rd ed., John Wiley & Sons, Inc., New York, USA,
1998.
L. B. Felsen, N. Marcuvitz: Radiation and scattering of waves, IEEE Press, Piscataway, N. J.,
1994.
R. E. COLIN: Field Theory of Guided Waves, 2nd ed., IEEE Press, Piscataway, N. J., 1991.
R. E. COLIN: Foundations for Microwave Engineering, 2nd ed., IEEE Press, Piscataway, N.
J., 2001.
D. M. Pozar: Microwave engineering, 2nd ed., John Wiley & Sons, Inc., New York, 1998.
C. A. Balanis: Antenna theory analysis and design, 2nd ed., John Wiley & Sons, Inc., New
York, USA, 1997.
W. L. Stutzman, G. A. Thiele: Antenna theory and design, John Wiley & Sons, Inc., New
York, USA, 1981.
J. Lavergnat, M. Sylvain: Radiowave Propagation, John Wiley & Sons, Inc., New York, USA,
2000.
L.W. Barclay (Ed): Propagation of radiwaves, 2nd ed., IEE Books, G.B., 2003.
C. R. Pollock: Fundamentals of optoelectronics, Richard D. Irwin, Inc., USA, 1995.
C. –L. Chen: Elements of optoelectronics and fiber optics, Richard D. Irwin, Chicago, USA,
1996.

194
book-13

You might also like