Renato Orta

Lecture Notes
on
Electromagnetic Field Theory
PRELIMINARY VERSION

November 2011

DEPARTMENT OF ELECTRONICS

POLITECNICO DI TORINO

Contents
Contents

1 Fundamental concepts

1.1

Maxwells Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.2

Phasors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.3

Constitutive relations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.4

Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

11

2 Waves in homogeneous media

14

2.1

Plane waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

14

2.2

Cylindrical waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

20

2.3

Spherical waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

21

2.4

Waves in non homogeneous media . . . . . . . . . . . . . . . . . . . . . . . . . . .

22

2.5

Propagation in good conductors . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

23

3 Radiation in free space

28

3.1

Greens functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

28

3.2

Elementary dipoles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

34

3.3

Radiation of generic sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

41

4 Antennas
4.1

49

Antenna parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

49

4.1.1

Input impedance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

50

4.1.2

Radiation pattern, Directivity and Gain . . . . . . . . . . . . . . . . . . . .

51

4.1.3

Effective area, effective height . . . . . . . . . . . . . . . . . . . . . . . . . .

54

4.2

Friis transmission formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

57

4.3

Examples of simple antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

59

4.3.1

Wire antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

60

4.3.2

Aperture antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

67

CONTENTS

5 Waveguides

76

5.1

Waveguide modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

77

5.2

Equivalent transmission lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

80

5.3

Rectangular waveguide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

86

5.3.1

Design of a single mode rectangular waveguide . . . . . . . . . . . . . . . .

92

5.3.2

Tunneling effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

94

5.3.3

Irises and waveguide discontinuities . . . . . . . . . . . . . . . . . . . . . . . 100

A Mathematical Basics

A.1 Coordinate systems and algebra of vector fields . . . . . . . . . . . . . . . . . . . .

A.2 Calculus of vector fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10

A.3 Multidimensional Dirac delta functions . . . . . . . . . . . . . . . . . . . . . . . . .

17

B Solved Exercises

20

B.1 Polarization and Phasors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

20

B.2 Plane Waves

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

23

B.3 Antennas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

28

B.4 Waveguides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

36

Chapter 1

Fundamental concepts
1.1

Maxwells Equations

All electromagnetic phenomena of interest in this course can be modeled by means of Maxwells
equations

E(r,t) = B(r,t) M(r,t)

t
(1.1)

H(r,t) =
D(r,t) + J (r,t)
t
Let us review the meaning of the symbols and the relevant measurement units.

E(r,t)

electric field

V/m

H(r,t)

magnetic field

A/m

D(r,t)

electric induction

C/m2

B(r,t)

magnetic induction

Wb/m2

J (r,t)

electric current density (source)

A/m2

M(r,t)

magnetic current density (source)

[V/m2 ]

Customarily, only electric currents are introduced; it is in particular stated that magnetic
charges and currents do not exist. However, it will be seen in later chapters, that the introduction
of fictitious magnetic currents has some advantages:
The radiation of some antennas, such as loops or horns, is easily obtained
Maxwells equations are more symmetric
3

PRELIMINARY VERSION

(surface) magnetic currents are necessary for the formulation of the equivalence theorem, a
fundamental tool for the rigorous modelling of antennas
In circuit theory, one has two types of ideal generators, i.e. current and voltage ones: likewise, in
electromagnetism one introduces two types of sources.
Concerning the symmetry of Maxwells equations, we cite the principle of duality: performing
the exchanges
E H
B D
J M
Maxwells equations transform into each others.
Experiments show that the electric charge is a conserved quantity. This implies that electric
current density and electric charge (volume density are related by a continuity equation
J (r,t) +

(r,t)
=0
t

(1.2)

By analogy, we assume that also magnetic charges are conserved, so that a similar continuity
equation must be satisfied:
m (r,t)
M(r,t) +
=0
(1.3)
t
It can be proved that eqs.(1.1), (1.2) (1.3) imply the well known divergence equations
B(r,t) = m (r,t)

D(r,t) = (r,t)

(1.4)

Some authors prefer to assume eqs.(1.1), 1.4) as fundamental equations and obtain the conservation
of charge (1.2) (1.3) as a consequence.
Maxwells equations can be written in differential form as above, so that they are assumed to
hold in every point of a domain, or in integral form, so that they refer to a finite volume. The
integral form can be obtained by integrating eq.(1.1) over an open surface o with boundary
and applying Stokes theorem:
E ds

d
dt

B
do
o

M
do
o

(1.5)
H ds =

d
dt

D
do +
o

J
do
o

In words, the first equation says that the line integral of the electric field, i.e. the sum of all voltage
drops along a closed loop, equals the time rate of change of the magnetic induction flux plus the
total magnetic current. The second equation says that the line integral of the magnetic field along
a closed loop equals the time rate of change of the electric induction flux plus the total electric
current.
Then we integrate eq.(1.4) over a volume V with surface and apply the divergence theorem:
B
d =

m dV
V

D
d =

dV
V

(1.6)

PRELIMINARY VERSION

Figure 1.1. Open surface o for the application of Stokes theorem. Notice that the orientations of
and are related by the right-hand-rule: if the thumb points in the direction
of
, the fingers point in that of .

Figure 1.2.

Closed surface for the application of the divergence theorem.

which is the usual formulation of Gauss theorem.

The same procedure on eq.(1.2) produces:
J
d +

d
dt

dV = 0

(1.7)

This says that the total current out of a volume equals the time rate of decrease of the internal
charge.

1.2

Phasors

Field variables can have any time dependence but a particularly important one is the so called
time-harmonic regime. Consider a general time-harmonic electric field in a particular point:
E(t) = Ex0 cos(0 t + x )
x + Ey0 cos(0 t + y )
y + Ez0 cos(0 t + z )
z

PRELIMINARY VERSION

The three cartesian components are sinusoidal functions of time with different amplitudes and
phases, but the same frequency. This equation can be transformed in the following way.
E(t) = R Ex0 ej(0 t+x ) x
+ R Ey0 ej(0 t+y ) y
+ R Ez0 ej(0 t+z )
z
=R

Ex0 ejx x
+ Ey0 ejy y
+ Ez0 ejz
z ej0 t

(1.8)

= R E ej0 t
The complex vector E is called the phasor of the time-harmonic vector E(t) and is measured in
the same units. It can be decomposed into real and imaginary parts:
E = E + jE
with
E = Ex0 cos x x
+ Ey0 cos y y
+ Ez0 cos z
z
and
E = Ex0 sin x x
+ Ey0 sin y y
+ Ez0 sin z
z
Consider again eq.(1.8):
E(t) = R E ej0 t
= R (E + jE ) ej0 t
= R {(E + jE ) (cos 0 t + j sin 0 t)}
= E cos 0 t E sin 0 t
We have obtained a representation of the time-harmonic electric field as the sum of two vectors,
with arbitrary directions, in time quadrature one with respect to the other: in other words both
these vectors are sinusoidal functions of time, with the same frequency but with a relative delay of
a quarter of a period. This representation is useful to study the polarization of the time-harmonic
vector, i.e. the form of the curve traced by the vector E(t) as a function of time. It can be shown

E"

E'

E(t )

Figure 1.3.

Elliptically polarized field

that, in general, this curve is an ellipse inscribed in a parallelogram that has E and E as half
medians, as shown in Fig. 1.3. We see easily from the previous equation that

PRELIMINARY VERSION

for t = 0, E = E
for t = T /4, E = E
for t = T /2, E = E
for t = 3T /4, E = E
where T = 2/ is the period. Hence the sense of rotation is from E to E . In this case the
field is said to be elliptically polarized.
There are particular cases. When E and E are parallel or one of the two is zero, the parallelogram degenerates into a line and the polarization is linear. The two cases can be condensed in
the single condition (cross product, i.e. vector product):
E E =0
The other particular case is that in which
|E | = |E |

and

E E =0

The parallelogram becomes a square and the ellipse a circle: the field is circularly polarized.
The plane defined by the two vectors E and E is called polarization plane. Suppose we
introduce a cartesian reference in this plane with the z axis orthogonal to it. The phasor E has
only x and y components,
E = Ex x
+ Ey y

where Ex and Ey are complex numbers. This means that the original time-harmonic field is
represented as the sum of two sinusoidally varying orthogonal vectors, with arbitrary magnitudes
and phases. On this basis, the type of polarization is ascertained with the following rules:
if the phase difference between the two components = arg Ey arg Ex is 0 or the
polarization is linear, along a line that forms an angle = arctan(|Ey |/|Ex |) (if = 0) or
= arctan(|Ey |/|Ex |) (if = )
if = /2 and |Ey | = |Ex | the polarization is circular, clockwise if = /2, counterclockwise if = /2
in the other cases, the polarization is elliptical
To illustrate these concepts consider the following example.

Example
The phasor of a magnetic field is H = (1 + j)
x + (1 3j)
y. Determine the type of polarization,
write the expression of the time varying field H(t) and draw the polarization curve defined by this
vector.
Solution Find real and imaginary part of the phasor
H =x
+y

H =x
3
y

PRELIMINARY VERSION

Compute
H H = (
x+y
) (
x 3
y) = (3 1)
z=0
H H = (
x+y
) (
x 3
y) = 1 3 = 2 = 0
The polarization is neither linear nor circular, hence it is elliptical counterclockwise (H(t) goes
from H to H ).
The time varying field is
H(t) = (
x+y
) cos 0 t (
x 3
y) sin 0 t
The plot is shown in Fig. 1.4
y
4

1
x
0

4
4

Figure 1.4.

Polarization curve defined by H(t) above

The time-harmonic regime is important because of the property

d
d
E(r,t) = R E(r) ej0 t = R j0 E(r) ej0 t
dt
dt

(1.9)

so that time derivatives become algebraic operations. If we substitute the representation (1.8) into
(1.1) we obtain, after canceling common factors exp(j0 t):
E(r)

j0 B(r) M(r)
(1.10)

H(r) =

j0 D(r) + J(r)

If the time dependence is of general type, eq. (1.8) is generalized by the spectral representation
(inverse Fourier transform)

1
E(r,) ejt d
(1.11)
E(r,t) =
2

PRELIMINARY VERSION

In words, a generic time varying electric field is represented as a sum of an infinite number of
harmonic components of all frequencies and amplitude E(r,)d/(2), where

E(r,) =

E(r,t) ejt dt

(1.12)

(direct Fourier transform). E(r,) is a spectral density of electric field, hence it is measured in
V/(m Hz). Due to the fact that E(r,t) is real, E(r, ) = E (r,), so that the previous equation
can also be written

1
E(r,t) = 2R
E(r,) ejt d
2 0
in terms of positive frequencies only.
The importance of the spectral representation is related to the property
d 1
d
E(r,t) =
dt
dt 2

E(r,) ejt d =

1
2

jE(r,) ejt d

If we take the Fourier transform of (1.1), we get

E(r,)

jB(r,) M(r,)
(1.13)

H(r,) =

jD(r,) + J(r,)

While 0 is a specific frequency value, (1.13) must be satisfied for all frequencies. We refer to
these system as Maxwells equations in the frequency domain. The variables will be interpreted as
phasors or Fourier transforms, depending on the application.

1.3

Constitutive relations

Clearly Maxwells equations as written in the previous section form an underdetermined system:
indeed there are only two equations but four unknowns, the two fields and the two inductions. It
is necessary to introduce the constitutive relations, i.e. the equations linking the inductions to the
fields. From a general point of view, matter becomes polarized when it is introduced into a region
in which there is an electromagnetic field, that is, the electric charges at molecular and atomic
level are set in motion by the field and produce an additional field that is summed to the original
one. The inductions describe the electromagnetic behaviour of matter.
The simplest case is that of free space in which
B(r,t)

0 H(r,t)
(1.14)

D(r,t) =
where

0,

E(r,t)

dielectric permittivity, and 0 magnetic permeability, have the values

0

4 107 H/m
1
1

109
0 c2
36

F/m

PRELIMINARY VERSION

10

and the speed of light in free space c has the value

c = 2.99792458 108

m/s.

In the case of linear, isotropic, dielectrics, the constitutive relations (1.14) are substituted by
B(r,)

() H(r,)

D(r,) =

() E(r,)

(1.15)
where
() =
() =

0 r ()
0 r ()

and r (), r () (pure numbers) are the relative permittivity and permeabilities. All non ferromagnetic materials have values of r very close to 1. Notice that since molecular and atomic
charges have some inertia, they cannot respond instantaneously to the applied field, so that the
response depends on the time rate of change of the excitation. The description of such a mechanism
is best performed in the frequency domain, where () and () play the role of transfer functions.
The fact that they depend on frequency is called dispersion: hence free space is non dispersive. In
general () and () are complex:
= j

= j

It can be shown that the volume density of dissipated power in a medium is related to their
imaginary part
1
1
Pdiss = |E|2 + |H|2
2
2
Notice that , are positive in a passive medium because of the phasor time convention
exp(j0 t). Some authors use the convention exp(j0 t): in this case passive media have negative , . Clearly with the time convention we adopt, negative , characterize active
media, such as laser media.
When the dielectric contains free charges, the presence of an electric field E(r,) gives rise to
a conduction current density Jc (r,):
Jc (r,) = () E(r,)
where () is the conductivity of the dielectric, measured in S/m. The previous equation is the
microscopic form of Ohms law of circuit theory.
The conduction current enters into the second Maxwells equation (1.13), which becomes
H = jE + E + J
where the term J is the (independent) source. It is customary to incorporate the conduction
current into the polarization current by means of an equivalent permittivity. Indeed we can write
jE + E = j j

E = jeq E

with eq = j( + /). In practice the subscript eq is always dropped: the imaginary part of
takes into account all loss mechanisms, both those due to molecular and atomic transitions and
those due to Joule effect.

PRELIMINARY VERSION

11

In the case of low loss dielectrics one often introduces the loss tangent
tan =

so that we can write

= (1 j tan )
Values of tan

10

: 10

characterize low loss dielectrics.

It is interesting to note that for fundamental physical reasons, there is relationship between
the real and the imaginary part of the dielectric permittivity and of the magnetic permeability.
Indeed, in the case of the permittivity, just as a consequence of causality, () 0 and () are
Hilbert transforms of each other, that is
() 0 =
() =

1
P

1
P

()
d

() 0
d

These equations are called Kramers-Kronig relations. The symbol P denotes the Cauchy principal
value of the integral, that is, for f (0) = 0

f (x)
dx = lim
a0
x

f (x)
dx +
x

f (x)
dx
x

The constitutive equations (1.15) imply that the inductions are parallel to the applied fields, which
is true in the case of isotropic media but not in the case of crystals. These media are said to be
anisotropic and are characterized by a regular periodical arrangement of their atoms. In this case
the permittivity constitutive equations must be written in matrix form:

Dx
xx xy xz
Ex
Dy = yx yy yz Ey
Dz
zx zy zz
Ez
In the case of an isotropic dielectric, the matrix is a multiple of the identity: = I.

1.4

Boundary conditions

Maxwells equations (1.13) are partial differential equations (PDE), valid in every point of a given
domain, which can be finite or infinite. If it is finite, we must supply information about the nature
of the material that forms the boundary. In mathematical terms, we must specify the boundary
conditions, i.e. the values of the state variables on the boundary.
Often the boundary is a perfect electric conductor (PEC), that is a material with infinite
conductivity. Note that copper is such a good conductor that up to microwave frequencies it can
be modeled as a PEC. If the conductivity is infinite, the electric field must vanish everywhere in
the volume of a PEC, otherwise the conduction current would be infinite. The first Maxwells
equation shows that also the magnetic field is identically zero, provided the frequency is not zero.
Since we are essentially interested in time-varying fields, we conclude that in a PEC both fields
vanish identically. At the surface, since the conduction current cannot have a normal component,

PRELIMINARY VERSION

12

only the tangential component of the electric field must be zero. If

is the unit normal at the
boundary, this condition can be written

E=0

on the boundary

(1.16)

Indeed,
E is tangential to the boundary, as shown in Fig. 1.5.

E tg

W

Figure 1.5. Boundary condition at the surface of a perfect electric conductor. Relationship between
the tangential component of the electric field Etg and
E

Sometimes the permittivity or the permeability change abruptly crossing a surface in the domain. By applying some integral theorems to Maxwells equations, it can be shown that the
following continuity conditions hold

(H(r+ ) H(r )) = 0

(B(r+ ) B(r )) = 0

(E(r+ ) E(r )) = 0

(1.17)

(D(r+ ) D(r )) = 0

(1.18)

where
is the normal to the surface and r are infinitely close points, lying on opposite sides of
the surface, as shown in Fig. 1.6. It can also be proved that if a surface current Js or Ms exist,
then the fields are discontinuous

2 2

r+

1 1
Figure 1.6.

Boundary conditions at the surface of separation between different dielectric media

(H(r+ ) H(r )) = Js

(E(r+ ) E(r )) = Ms

Since it can be proved that when a PEC is present in a magnetic field, the induced current flows
only on the surface of it and the magnetic field is identically zero in the PEC, we can write

H(r+ ) = Js

(1.19)

PRELIMINARY VERSION

13

Notice that this is an equation that does not force a condition on H but allows the computation of
Js . Hence the boundary condition to be enforced at the surface of a PEC is only (1.16). Notice also
that the units of a surface electric current are A/m and those of a surface magnetic current V/m.
This is necessary for the validity of the previous equations, but it is also obvious for geometrical
reasons, as Fig. 1.7 shows.

Js
W

Figure 1.7. Surface electric current induced on a perfect electric conductor (PEC). is a curve
lying on the PEC surface. Js is a current density per unit length measured along (A/m)

Sometimes an approximate boundary condition of impedance type is used: it is a linear relation

between the tangential electric and magnetic fields, called also a Leontovich boundary condition,
that can be written

E(r+ ) = Zs (
H(r+ ))

(1.20)
in terms of a suitable surface impedance. This condition is typically applied when the boundary is
a real metal and one desires a more accurate model than just a PEC. The double vector product
on the right hand side makes the tangential electric and magnetic field orthogonal on the surface.
If the surface is not smooth but has a sub-wavelength wire structure, the surface impedance is not
a scalar but a tensor (matrix).
If the domain is infinite and sources are all at finite distance from the origin, the only necessary
boundary condition is that the field is only outgoing at large distance from the origin.
Sometimes the geometry of dielectric and metal bodies possesses sharp edges or sharp vertices
(e.g. plates, cubes, cones), as shown in Fig. 1.8. In this case some field components can become
infinite at the geometric singularity: however the fields must always be locally square integrable.
In physical terms this condition means that the electromagnetic energy contained in a finite neighborhood of the singularity must always be finite. Apart from these cases of true singularities of
the geometry, fields are always regular, i.e. differentiable. This is to be noted in particular when
the geometry singularity is only apparent because it is related to the particular coordinate system.
For example if we use cylindrical coordinates in a homogeneous medium the fields must be regular
in the origin even if the coordinates have a singularity there.

Figure 1.8.

Chapter 2

Waves in homogeneous media

At this point we have finished the preliminaries. We have decided to use the electric and magnetic
field as state variables and we have the relevant equations:

E = jH M

H = jE + J
(2.1)

+
boundary conditions
where for brevity we have dropped the dependence of all variables on r, , but it is understood.
The line boundary conditions contains the form of the domain where we want to compute the
fields created by the given sources J,M and information on the nature of the material that forms
the boundary. In general permittivity and permeability are functions of r and provide information
on the shape of the bodies in the domain and on their nature. In this way the problem is well posed
and it can be shown it has a unique solution, provided there is at least a region in the domain
where energy is dissipated. Needless to say the problem (2.1) can be very complicated and can be
solved only by approximate numerical techniques. For this reason it is useful to proceed by small
steps, by analyzing first a very idealized problem that is so simple to allow an analytical solution.

2.1

Plane waves

Let us start by assuming that the domain of interest is infinite and the medium is homogeneous
and lossless, so that , are real constants. The only boundary condition to enforce is that the
field is regular everywhere, in particular at infinity.
Moreover we assume that sources are identically zero. This is, at first sight, a strange assumption since it would seem to imply that also the fields must be identically zero! However, if there
are no sources, Maxwells equations become a homogeneous system of differential equations: it is
well known that homogeneous differential equations have nontrivial solutions. Let us review some
examples:

d
f (x) = 0, x R f (x) = const
dx
14

PRELIMINARY VERSION

15

Harmonic oscillator
d2
+ 02 f (t) = 0, t R f (t) = A cos 0 t + B sin 0 t
dt2
Transmission line

V =
dz

dI =
dz

V (z)

j L I
, zR
j C V

I(z)

= V0+ ejkz + V0 e+jkz

=

Y V0+ ejkz Y V0 e+jkz

These are actually generalizations of the simple case of a linear system of algebraic equations
Ax=0
which has nontrivial solutions if the matrix A is non invertible.
So the problem we want to solve is
E

H =

jH
jE

(2.2)

Equations (2.2) are written in a coordinate-free language. However, in order to solve them it
is necessary to select a coordinate system. Several choices are at our disposal, the more common
being cartesian, cylindrical and spherical coordinates. The corresponding solutions of (2.2) will
be plane, cylindrical and spherical waves, respectively. The simplest case is the first and we start
with that.
Recalling the expression of the operator in cartesian coordinates
=x

+y

+
z
x
y
z

and that the medium is homogeneous, it is clear that (2.2) is a linear system of constant coefficient
equations. On the basis of the experience with ordinary differential equations, we can expect that
the solution is of exponential type, hence we assume tentatively
E(r) = E0 exp(jkx x) exp(jky y) exp(jkz z)

(2.3)

and likewise for the magnetic field, where E0 and kx ,ky ,kz are constants to be found. The constants
kx ,ky ,kz have dimensions rad/m and are wavenumbers along the three coordinate axes. It is
convenient to work with a vector formalism, even if the coordinate system is fixed. By recalling
that r = x
x + y
y + z
z and defining the wavevector k = kx x
+ ky y
+ kz
z, the assumed form of the
solution is
E(r) = E0 exp(jk r)
H(r) = H0 exp(jk r)
(2.4)
Before substituting it into (2.2) it is useful to compute
exp(jk r) =

+y

+
z
x
y
z

exp(jkx x) exp(jky y) exp(jkz z)

= (jkx x
jky y
jkz
z) exp(jkx x) exp(jky y) exp(jkz z)
= jk exp(jk r)

PRELIMINARY VERSION

16

Moreover we recall the identity of vector calculus

(A(r)(r)) = (r) A(r) + (r) A(r)
so that the substitution of (2.4) into (2.2) yields
jk E0 exp(jk r)

jk H0 exp(jk r) =

jH0 exp(jk r)
jE0 exp(jk r)

Canceling common factors

jk E0

jH0

jk H0

jE0

Note that the fact that these equations do not contain the variable r any longer confirms the
correctness of the assumption (2.4).
Recalling the properties of vector products we learn that E0 , H0 , k form a righthanded triple
of mutually orthogonal vectors. Next, to proceed, we eliminate H0 between the two equations. To
this end, we solve the first equation with respect to H0 :
H0 =

k E0

(2.5)

and substitute into the second one

k (k E0 ) + 2 E0 = 0
The double vector product can be expanded
k (k E0 ) (k k) E0 + 2 E0 = 0
The first term is zero because of the orthogonality of k and E0 , hence
k k 2 E0 = 0

(2.6)

We are interested in nontrivial solutions of this equation, so that the following condition must hold
k k = 2

(2.7)

A relationship between frequency and wavenumbers is called in general a dispersion relation. We

can read it from right to left or vice versa: in the first instance it tells us what the frequency
must be so that the field distribution (2.4) with a specific wavevector k is a solution of Maxwells
equations. From this point of view, the value of can be considered to be a resonance frequency of
the structure. Notice that the requirement that the solution be regular everywhere (in particular
at infinity) forces the wavector to be real. Apart from this condition there is no constrain on the
possible values of the wavenumbers, hence the resonance frequencies of the system are infinite in
number and even distributed continuously. We start seeing a property that characterizes all field
problems. Whereas lumped element circuits have a finite number of resonances, distributed systems
always have an infinite number of them. Moreover, if the structure has finite size its resonances
are denumerably infinite: this means that they can be labeled with integers 1 , . . . ,n , . . .. If the
structure, as in this case, has infinite size, then the resonances should be labeled with a continuous

PRELIMINARY VERSION

17

variable. To simplify the notation, we omit this labeling variable and remember that can take
every real value.
The dispersion equation can also be read from left to right: in this case the frequency is
considered fixed and we look for the wavevectors that satisfy eq.(2.7). It is convenient to introduce
a unit vector
s, called direction of propagation, directed along k, so that k = k
s; then the dispersion
relation becomes

k =

(2.8)

Clearly the direction

s can be whatever, only the wavenumber k is specified. In other words for
any given frequency there are an infinite number of waves with arbitrary directions of propagation.
Considering again eq.(2.6), we see that if the dispersion relation is satisfied, the vector E0
can be arbitrarily chosen, provided it is orthogonal to
s. For a given
s, there are two linearly
independent waves, but they are degenerate, because they have the same value of the wavenumber
k. The corresponding magnetic field is obtained by the impedance relation (2.5):

s E0

H0 =
=
s E0 = Y s E0
(2.9)

where the wave admittance has been introduced. Its inverse is the wave impedance Z = 1/Y . In
free space it has the value
0
Z0 =
120 377
0
In conclusion a solution of the problem (2.2) is
Es(r) = E0 exp(jk
s r)
(2.10)
Hs(r) = H0 exp(jk
s r)

where
s is the direction of propagation, k = , H0 = Y s E0 and E0 , H0 , s are mutually
orthogonal: electromagnetic waves are transverse. Since the problem is linear, the general solution
of (2.2) can be written as a linear combination of waves of the type (2.10) with all possible directions

s.
Wavefronts are defined to be the surfaces on which the phase (r) of the wave is constant. In
this case
(r) = k
s r = constant
is the equation of a family of planes perpendicular to s: hence the fields (2.10) are called plane
waves because the wavefronts are planes.
Let us now study the polarization of plane waves. This requires going back to time domain via
eq.(1.8)
E(r,t) = R {E0 exp(jk
s r) exp(j0 t)}
= E 0 cos (0 t k
s r) E 0 sin (0 t k
s r)

(2.11)

We see clearly from this equation that the type of polarization in every point of space is specified
by E0 , which is the electric field in the origin. What changes from point to point is the time
evolution, due to the propagation delay. The constant phase surfaces of the time varying field are
0 t k
s r = const, from which we find
0
t const

sr=
k

PRELIMINARY VERSION

18

s
W

H
Figure 2.1. Wavefronts of a plane wave with direction of propagation
s. They move at the phase
velocity along
s. The vector u
denotes an arbitrary direction

This means that the wavefronts are not fixed but are moving. Indeed, consider a specific wavefront,
say the zero-phase one, i.e. the one for which const=0, as shown in Fig.2.1; the vector r that denotes
its points is such that its projection on
s increases linearly with time. In other words, the plane
moves as a whole with speed
vph =

0
0
1
c
= = =
k
0

r r

(2.12)

This velocity is called phase velocity of the wave, because it has been defined by means of the
constant phase surfaces. Now consider an arbitrary straight line with direction u
, an let P be its
intersection with the zero-phase wavefront. The velocity of P is
vph (
u) =

1
c

u

s r r

Clearly, for all directions u

=
s this velocity is larger than c/ r r . Notice, however, that even
when this velocity is larger than the speed of light in empty space, the theory of relativity is not
violated. Indeed no matter or energy is moving in the direction of u
, but only a mathematical
point, e.g. a maximum or a node of the oscillation. This concept is at the basis of the fact that
the phase velocity in a waveguide is always greater that the speed of light in vacuum.
Fig.2.2 shows the fields of a linearly polarized plane wave propagating in the
s=
z direction:
remember that the electric and magnetic field are orthogonal. For clarity, the field vectors have
been drawn only for a number of points on the z axis, even if they are defined in every point of
space.
Wave propagation is always associated to energy flow. The Poynting vector has the meaning
of power density (per unit surface) associated to the wave. Let us compute the Poynting vector S
in the case of a plane wave:
1
1
(E H ) = (E0 exp(jk
s r) H0 exp(jk
s r))
2
2
1
1
= (E0 H0 ) = (E0 (
s E0 )Y )
2
2
1 |E0 |2
=

s
2 Z

S=

(2.13)

where we used the impedance relation (2.9) and the property

E0 (
s E0 ) = |E0 |2 s (E0 s) E0 = |E0 |2 s

(2.14)

PRELIMINARY VERSION

19

E( z, t )

H( z , t )

Figure 2.2.

Snapshot of a linearly polarized plane wave propagating in the

s=
z direction

because of the orthogonality between E0 and s. Note that the Poynting vector is real, hence
no reactive power is associated to plane waves in a lossless medium. We see that the active
power density (magnitude of S) associated to a plane wave is constant: this implies that the total
power, obtained by integration over the whole wavefront, is infinite. Hence, a single plane wave
is not a physically realizable field. This property, however, does not destroy the usefulness of the
concept. Indeed, since Maxwells equations are linear, the superposition principle holds and linear
combinations of plane waves are also solutions. It turns out that a continuous sum (integral) of
plane waves not necessarily has infinite power: indeed all physically realizable fields can always be
represented as integrals of plane waves.
We recall that to each plane wave not only a power flow is associated, but also a flow of
linear momentum and of angular momentum. In particular the linear momentum flow, which has
direction
s, is responsible for the radiation pressure, that explains, for instance, the shape of comet
tails and has been considered as a possible engine for interplanetary travels.
The properties of plane waves do not change much if the dielectric is lossy. In this case the
permittivity is complex and the dispersion relation (2.8) becomes
k=

( j ) = j,

where is the true phase constant, measured in rad/m and is the attenuation constant, measured
in Nepers/m. The electric field, for instance, obeys the propagation law
E(r) = E0 exp(j
s r) exp(
s r)

(2.15)

Clearly the magnitude of the field is no longer constant in space and the wavefronts are also surfaces
of constant field magnitude. Obviously, it must be remarked that the value of the phase velocity
cannot be computed by (2.12), but it is given by
vph =
and the wavelength is
=

vph
2
=

PRELIMINARY VERSION

20

Indeed, both the phase velocity and the wavelength are defined on the basis of the phase of the
wave and is exactly the phase rate-of-change, measured, as said above, in rad/m.
Let us compute the power flow.
1
1
(E H ) = (E0 exp(jk
s r) H0 exp(jk s r))
2
2
1
1
= (E0 (
s E0 )Y exp(2
s r)) = Y |E0 |2 exp(2
s r)
s
2
2

S=

(2.16)

where we have used again the property (2.14). In this case the Poynting vector is complex. The
active power density of the wave decreases during propagation because part of it is transferred to
the dielectric in the form of heat.
All the plane waves considered up to now are called uniform because their propagation direction s is real. If we go through all the steps of the derivation, we realize that even if s is complex
(whatever this means!) the expression (2.10) is a valid solution of Maxwells equations, although
only in a halfspace. Such a generalization leading to non uniform plane waves is required when
solving a scattering problem where a plane wave is incident on the interface separating two dielectrics. It is to be remarked that also the plane wave (2.15) is a valid solution only in a halfspace.
Indeed, if s r , the field diverges, which is not physically acceptable.

2.2

Cylindrical waves

To solve eqs.(2.2) it is also possible to use a cylindrical coordinate system instead of a cartesian
one. The mathematics is considerably more complicated in this case. The reason is that the unit
vectors of the cylindrical coordinate system are not constant but change from point to point. As
a consequence, the expression of the differential operators is no longer with constant coefficients
and the solutions are no longer of exponential type, but are expressed in terms of Bessel functions.
These special functions of mathematical physics were actually introduced, along with many others,
in order to solve the wave equation.
If a cylindrical or spherical coordinate system is used, Maxwells equations (2.2) are not attacked
directly but are first transformed into a single second order equation. We write them again for
convenience
E = jH
H =

jE

Since it is a system of equations, we can eliminate one of the two unknowns. We solve the first
equation with respect to H and substitute in the second

H
=

( E) = 2 E
As expected, the second equation contains only the electric field: the price to pay for it is that it
is second order in the space derivatives; it is called the curl-curl equation. However its form can
be simplified recalling the identity
( E) = ( E) 2 E = 2 E

PRELIMINARY VERSION

21

where we have used the fact that (2.2) do not have sources, hence also the charge density (r) is
zero and the electric field has zero divergence, E = = 0. We obtain in this way the vector
Helmholtz equation
2 E + 2 E = 0
(2.17)
Even if it is written in coordinate-free language, its meaning is easily understood in cartesian
coordinates only, where
E(r) = Ex (r)
x + Ey (r)
y + Ez (r)
Since the unit vectors are not function of r, each cartesian component of the electric field satisfies
Helmholtz equation, which then becomes scalar:
2 + 2 = 0
where (r) denotes any component of E. It is interesting to note that even if we are using cartesian
components to represent the electric field, we are not forced to use necessarily cartesian coordinates
to specify the observation point, i.e. the components of r. By using the classical method of the
separation of variables in cylindrical coordinates, we can find
(2)
(k )ejkz z ejm
(,,z) = 0 Hm

(2.18)

where m = 0, 1, 2, . . ., k [0,) and kz identify the various outgoing cylindrical waves. These
(2)
three parameters play the role of kx , ky , kz in the case of plane waves. The function Hm (k ) is
a Hankel function of second kind and order m. Its asymptotic expansion is
(2)
Hm
(k )

2

exp j(k m )
k
2
4

The dispersion relation is

2 k2

kz =

Notice that all three components have this form, but the values of 0 for each must be interrelated
so that the resulting vectors E and H satisfy Maxwells equations.
We are not going to describe in detail the properties of these waves. To explain the name, it
is enough to say that the wavefronts are cylinders having z as axis, at least in the case m = 0 and

k = .

2.3

Spherical waves

The case of spherical waves is similar, from a certain point view, to that of cylindrical waves. Again
the mathematics is fairly complicated and new special functions are introduced. In this case the
scalar Helmholtz equation is solved in spherical coordinates and the result is
(2)

(r,,) = 0 hl (kr)Plm (cos )ejm

where l = 0,1,2, . . . and l m l identify the various outgoing spherical waves. The functions
(2)
hl (k) are spherical Hankel functions of second kind and order l, whereas Plm (cos ) are associated
Legendre polynomials of degree l and order m. Again, the various solutions for the three cartesian

PRELIMINARY VERSION

22

1
1 r

2
2

3
3

t 4

i
Figure 2.3. Scattering from a stratified dielectric: i, incident wave; r reflected wave; t,
transmitted wave. For clarity, the couple of plane waves existing in each of the internal
layers has not been indicated

components must be related so that the resulting E and H satisfy Maxwells equations. The
asymptotic expansion of the spherical Hankel functions is
(2)

hl (kr)

1

exp j(kr m )
kr
2
4

Hence the wavefronts are spheres with center in the origin and this justifies the name.

2.4

Waves in non homogeneous media

The case that has been considered, namely a homogeneous medium filling the whole space is highly
idealized. In a realistic situation, (r), (r) are not constant and obviously the plane waves (2.10)
are not solution of Maxwells equations (2.2). In order to consider a simple case, let us assume
that the medium is piecewise homogeneous and that the interfaces between the different materials
are planar: the structure is called a stratified dielectric. In the left half space an incident plane
wave is assumed. In each layer, plane waves are solutions of (2.2), but the continuity conditions
(1.17) must be obeyed. It can be proved that in each one of the internal layers two plane waves
are present, one forward (incident on the following interface) the other backward (reflected from
the following interface); in the right half space only one, because the medium extends to infinity
and there is no other interface. All these plane waves have the same transverse (to z) component
of the wavevector and their amplitudes can be easily determined so that the continuity conditions
are satisfied. The single wave in the fourth medium is the transmitted field, the second one in the
first medium is the reflected field, as sketched in Fig. 2.3.
If the interfaces are not planar, the problem becomes much more difficult. Consider, for example, the case of Fig. 2.4, where a plane wave is incident on a cylinder with parameters 2 , 2 ,
embedded in a homogeneous medium with parameters 1 , 1 . It can be shown that the continuity
conditions require that an infinite number of plane waves are excited, each one with the right amplitude. Collectively. these are called scattered waves. Hence the difficulty of the problem stems
from the necessity to solve a linear system with an infinite number of unknowns.
If the medium is not even piecewise homogeneous but arbitrarily inhomogeneous, no analytical
solution is at our disposal. It is, however, to be mentioned that when the variations of (r), (r)
are small on the wavelength scale, a well known approximate method can be used, i.e. Geometrical
Optics. Whereas the plane waves discussed up to now can be defined global plane waves since each
one is defined over the whole space, the elementary geometrical optics field is a local plane wave.
For instance, a spherical wave in free space can be approximated by a collection of local plane

PRELIMINARY VERSION

23

2
2

1
1

Figure 2.4. Scattering from a non planar interface: i, incident wave; s, scattered waves. For clarity,
the plane waves existing inside the cylinder have not been indicated

waves because its wavefront (a sphere) can be approximated locally by the relevant tangent plane.
The k vectors of these local plane waves define a vector field, whose field lines are the geometrical
optics rays. It turns out that rays are also the field lines of the Poynting vector field: hence a plot
of the rays provides information about the power flow.
Geometrical optics is a very powerful technique, but sometimes yields definitely wrong results.
This happens when rays cross in a point or along a line, because in this case it predicts a field of
infinite intensity. These singularities are called caustics and the focus of a converging lens is an
example: in such a point the electromagnetic field can be very large but is certainly finite. Hence
geometrical optics can be safely used only away from caustics.

2.5

Propagation in good conductors

Apart from the case of optical fibers, guided wave propagation is possible in structures containing
metal conductors. Examples are coaxial cables, parallel wire transmission lines, microstrip lines,
waveguides with any cross section. Since the metals used in the applications (such as copper) are
characterized by a very large conductivity, in a first approximation they can be considered to be
perfect conductors (PEC), an assumption that greatly simplifies the study. However, in order to
build more accurate numerical models of real devices, it is necessary to take into account the finite
conductivity of real metals. In this section we consider the propagation of plane waves in good
conductors, in order to draw some conclusions pertaining to transmission systems.
Metals are characterized by so a large conductivity that the displacement currents can be safely
neglected with respect to the conduction currents, so that some simplifications in the general
formulas of Section 2.1 are possible. Starting with the wavenumber,
km =
if

( j/)

(j/)

(good conductor)

Recalling that
1j
j =
2

PRELIMINARY VERSION

24

and that Imk 0 for a passive medium, we find

1j
1j
km =
=

(2.19)

where we have introduced the so called skin depth

2

(2.20)

which, of course, should not be confused with the loss angle, introduced in Section 1.3, indicated
with the same symbol. This relation can also be written
1
= const

f =

where f is the frequency and the constant depends only on the material. For instance, in the case
of copper, = 5.8 107 S/m and = 0 = 4 107 H/m, hence

f = 0.0661 Hzm
(2.21)
The reason for the name will be explained below.
The wave impedance is computed with the same approximation:
Zm =

j/

=
j/

1+j
j
=

that is
Zm = Rs (1 + j)

(2.22)

where we have introduced the surface resistance

1
=
2

Rs =
for which we can write

R
s =
f

= const

Et

Hi
Hr

Ht

Er

Figure 2.5. Good conductor in a plane wave field. In the free space region both an incident and
a reflected wave exist, in the metal only the transmitted one. The wavevector of the transmitted
wave is drawn dashed, to indicate that it is complex.

PRELIMINARY VERSION

25

Again, in the case of copper,

R
s = 2.6090 107 / Hz
f

Consider now a (highly idealized) conductor in the form of a half space, which faces free space,
with a linearly plane wave incident normally on it, as shown in Fig. 2.5. The tangential electric
and magnetic fields are continuous at the interface, then the ratio of their magnitudes is the same
in z = 0 and in z = 0+ . But in z = 0+ this ratio is Zm by definition, so we can easily understand
that the expressions of the electric fields are
Ei = E0 ejk0 z x

Er = E0 ejk0 z x

Et = (1 + )E0 ejkm z x

where the reflection coefficient is

=
Since |Zm |

Zm Z0
Zm + Z0

Z0 , is close to 1. Indeed,
1+=

2Zm
2Zm

= 2(1 + j)
Zm + Z0
Z0

0 f

from which
1 + 2(1 + j)
In the case of copper,

0 f

1 + 2(1 + j) 6.9252 1010

(frequency in Hz). We can also say that the metal enforces an impedance type boundary condition
(see (1.20)) with Zm as surface impedance.
The magnetic field is
Hi = Y0 E0 ejk0 z y

Hr = Y0 E0 ejk0 z y

H = Y0 (1 )E0 e

(2.23)

jkm z

x
2Y0 E0 e

jkm z

Note that the total magnetic field at the interface is approximately twice the incident one because
is very close to 1.
The electric field in the metal produces a conduction current in the x
direction
Jc = Et = 2(1 + j)

0 f E0 ejkm z x

In the case of copper, this becomes

Jc = 2(1 + j) 0.402

f E0 ejz/ ez/ x

(frequency in Hz) where we have used (2.19). The magnitude of this current density is maximum
at the interface and then decays exponentially in the metal. At a depth z = , it has reduced by a
factor 1/e = 0.368. Eq.(2.21) allows a simple computation of for various frequencies, reported in

PRELIMINARY VERSION

Table 2.1.

26

Skin depth for copper at various frequencies

Frequency

Skin depth

50 Hz
1 kHz
1 MHz
1 GHz

9.3 mm
2.1 mm
66.1 m
2.1 m

Table 2.1. We see clearly that as the frequency increases, the current density remains appreciable
only in a very thin layer close to the metal surface, which can be considered as its skin. Even if
this analysis strictly refers to a metal half space, we can use it to draw qualitative conclusions in
the case of finite thickness conductors or even round conductors, provided the thickness is much
larger than the skin depth. At the power frequency of 50Hz, the skin depth is so large that the
current has a uniform distribution in ordinary wires. At the frequency of 1MHz, instead, most of
the conductor copper is not used. At microwave frequencies, a few microns of copper deposited on
an insulator perform as an excellent conductor.
The consequence of the skin depth change with frequency is that the resistance of a conductor is
an increasing function of frequency: indeed, the effective cross-section of the conductor decreases
as the frequency increases. This phenomenon is generally called skin effect.
Let us compute the impedance of the structure of Fig. 2.5, viewed as a current carrying conductor. Since the fields and the current density does not depend on y, we consider a strip of unit
length in this direction. We compute first the total current I, flowing in the x
direction, per unit
length along y:

Jc0
I=
Jc (x,z) x
dz =
Jc0 ej(1j)z/ dz =
(2.24)
1+j
0
0
Notice that the dimensions of I are correctly A/m, since Jc0 is a surface current density with value
Jc0 = 2(1 + j) 0 f E0

(2.25)

Next, consider a unit length in the x

direction of this conductor and compute the potential difference along this length by integrating the electric field Ex along the x axis (y = 0, z = 0). Note
that Ex does not depend on x, hence Ex itself coincides numerically with this potential difference.
Finally, the impedance per unit width in the y direction and unit length in the x direction is given
by
Jc0 /
1+j
Ex (0,0)
=
=
= Zm
Zpul =
I
Jc0 /(1 + j)

where we used (2.22). In conclusion we have this remarkable result: the impedance seen by a
current flowing through a square of unit sides coincides with the wave impedance in the metal.
Notice that, apart from the similarity in the symbols,
Zm =

Ex
Hy

hence it is a completely different concept. Moreover, since the conductor we are considering has
unit width in the y direction, unit length in the x direction (and infinite thickness in the z direction)

PRELIMINARY VERSION

27

the previous analysis shows why often the value of the surface resistance Rs is expressed in /
(read Ohm per square).
As a final remark, we note that the material becomes a perfect conductor when . In
these conditions, the skin depth vanishes and the value of the current density at the interface tends
to infinity, according to (2.25). Nevertheless, we see from (2.24) that the total current is finite and
its value, independent of is
Jc0
0
I=
=2
E0
1+j
0
where (2.20) and (2.25) have been used. This means that in a perfect conductor the current density
can be written
0
Jc (x,z) = 2
E0 (z) = J (z)
0
On the other hand, from (2.23) we see that in the limit the magnitude of the total magnetic
field at the z = 0 interface coincides with J . Taking the directions of the vectors into account,
we conclude that if a perfect conductor is immersed in an electromagnetic field, on its surface a
current density J (A/m) appears, such that
J =
H
where
is the normal to the PEC surface, pointing toward free space. In practice, this is the proof
of Eq.(1.19).
Another example that we consider now is that of sea water: because of the salt contained in it,
the conductivity is = 5 S/m, whereas the relative permittivity, up to the microwave region, does
not change very much and will be taken to be r = 80. We compute the complex wavenumber and
the attenuation constant by the general equation
km =

( j/)

The results are the following:

At f = 100Hz, /(2f 0 r ) = 1.1234 107 , so sea water behaves as a good conductor;
k = (4.4429 j4.4429) 102 m1

= 0.3859dB/m

At f = 10kHz, /(2f 0 r ) = 1.1234 105 , so sea water behaves as a good conductor;

k = (0.4429 j0.4429)m1

= 3.8590dB/m

At f = 1GHz, /(2f 0 r ) = 1.1234, so the displacement currents cannot be neglected;

k = (209.7536 j94.1066)m1

= 817.3998dB/m

At f = 10GHz, /(2f 0 r ) = 0.1123, so the displacement currents cannot be neglected;

k = (1877.5266 j105.1341)m1

= 913.1833dB/m

Obviously, at microwave frequency, the attenuation of sea water precludes the possibility of communicating with submarines during subsurface navigation. This becomes possible at low frequencies,
where, however, the available bandwidth is very narrow.

Chapter 3

Radiation in free space

The fundamental problem in electromagnetics is computing the fields created by a specified set of
sources in a given region of space. This means that the functions (r), (r) are assigned, as well
as the form of the region boundary and the material of which it consists. Then the sources are
specified in terms of electric and magnetic current densities J(r), M(r).
In order to understand the basic mechanism of radiation, it is convenient to consider first a
highly idealized problem, wherein the sources radiate in an infinite homogeneous medium. Later
we will see how to apply the results of this chapter to the real antenna problem.

3.1

Greens functions

The radiation problem is mathematically formulated as

E

= j0 H M

= j0 E + J

(3.1)

in an infinite homogeneous domain that we assume to be free space. These equations are linear with
constant coefficients and the independent variable is r. We can interpret them as the equations of
a Linear Space Invariant system (LSI), where the source currents play the role of input and the
radiated fields that of output, see Fig. 3.1. The box represents a system with two inputs and two
outputs.

G EJ

J(r)

G HJ
M(r)

Figure 3.1.

E(r)
G EM

G HM

H(r)

Linear system view of the radiation phenomenon

28

PRELIMINARY VERSION

29

LSI systems are clearly a multidimensional generalization of Linear Time Invariant (LTI) systems. Let us review briefly the properties of the latter. LTI systems, as shown in Fig. 3.2 are
completely characterized in time domain by their impulse response h(t), that is the output that is
obtained when the input is a Dirac delta function (t). An arbitrary (continuous) input x(t) can
be represented as a linear combination of pulses thanks to the sifting property of the delta function

x(t) =

x(t )(t t )dt

Because of linearity, the response y(t) to x(t) can be found by convolution

y(t) = h(t) x(t) =

h(t t )x(t )dt

Alternatively, an LTI system can be characterized by its transfer function: when the input is x(t) =
exp(jt), the output is proportional to it and the constant of proportionality is, by definition, H(),
so that y(t) = H() exp(jt). It can be proved that the impulse response and the transfer function
of a system are related by a Fourier transform

H() =

h(t) exp(jt)dt

h(t )

x(t )

Figure 3.2.

y (t )

H ( )

X ( )

Y ( )

Time domain and frequency domain description of an LTI system

As a preparation for (3.1), let us consider the simpler case of an infinite, uniform transmission
line excited by a distribution of voltage and current generators, vs (z) and is (z), as shown in Fig. 3.3.
Since these generators are distributed continuously, vs (z) and is (z) are densities per unit length
of generators described, as usual, in terms of their open circuit voltage (V/m) and short circuit
current (A/m), respectively. The differential equations of the system are

dV

= jLI + vs
dz
(3.2)

dI = jCV + i
s
dz

vs ( z )

+ + + +
is ( z )

Figure 3.3.

Infinite uniform transmission line with distributed voltage and current generators.

PRELIMINARY VERSION

30

GVis

is ( z )
vs ( z )

Figure 3.4.

V ( z)

GVvs

GIis

I ( z)

GIvs

Linear system view of the transmission line with distributed generators.

Generally, transmission lines are excited at one end by a generator that acts as a transmitter.
The model shown in Fig. 3.3 refers to a situation of electromagnetic compatibility, where a line
is excited by an electromagnetic wave that couples to the line along a certain segment of it. It
is easy to recognize that this is the one dimensional analogue of Maxwells equations (3.1). The
problem is again LSI and can be schematized as in Fig. 3.4. Hence, as suggested by this picture,
the solution can be expressed as

V (z) = Z

GV is (z z )is (z )dz +

I(z) =

GIis (z z )is (z )dz + Y

GV vs (z z )vs (z )dz

GIvs (z z )vs (z )dz

The system here has two inputs and two outputs: each output depends on both inputs, so that
in practice there are four Greens functions, each one a pure number. They can be obtained
by applying the spatial Fourier transform to the system equations (3.2). However, by the very
definition of Greens function
GV is (z) is the voltage wave V (z)/Z created by a unit amplitude current generator located
in z = 0
GV vs (z) is the voltage wave V (z) created by a unit amplitude voltage generator located in
z=0
GIis (z) is the current wave I(z) created by a unit amplitude current generator located in
z=0
GIvs (z) is the current wave I(z)Y created by a unit amplitude voltage generator located in
z=0
so that they can be found by simple circuit theory, just recalling that the input impedance of an
infinitely long line is Z . The resulting expressions are
1
GV is (z) = GIvs (z) = ejk|z|
2
1
GV vs (z) = GIis (z) = sgn(z)ejk|z|
2
where sgn(z) is the sign function
sgn(z) =

1 if z > 0
1 if z < 0

PRELIMINARY VERSION

31

As another preparatory example before tackling (3.1), let us consider the case of sound waves.
It can be shown that the excess pressure p(r) with respect to the background pressure satisfies the
scalar Helmholtz equation
2
2 + 2 p(r) = S(r)
(3.3)
Vs
where Vs is the sound velocity and S(r) is a source term. This equation corresponds to the picture
of Fig. 3.5, where S(r) is the input and p(r) the output. In this case the system has only one
input and one output but it is multidimensional, since both depend on the three independent
variables x, y, z. In perfect analogy with LTI systems, LSI systems are completely characterized
in space domain by their impulse response G(r), which is traditionally called Greens function.
This is the output of the system when the input is a point source located at the origin of the
coordinate system, which can be represented mathematically by a three-dimensional Dirac delta
function S(r) = (r) = (x)(y)(z). The fundamental property of this multidimensional Dirac
function is

(x)(y)(z)dx dy dz = 1

(r)dr =

When the input is an arbitrary function, the output is found by (three-dimensional) convolution
p(r) =

G(r r )S(r )dr

Alternatively, an LSI system can be characterized in the spectral domain. When the input is a
harmonic function of x, y, z, that is S(r) = exp(j(kx x + ky y + kz z)) = exp(jk r), the output
is proportional to it and the coefficient of proportionality is, by definition, the transfer function
H(k). Again, transfer function and Greens function of the same system are related by a Fourier
transform: however, in this case, it is triple, since it operates on the three variables x, y, z. The
couple of inverse and direct 3-D Fourier transform is given by
G(r) =
H(k) =

1
(2)3

H(k) exp(jk r)dk

(3.4)

G(r) exp(jk r)dr

where dk = dkx dky dkz . It can be shown that in the case of free space, the transfer function is
H(k) =

1
2 /Vs2

(3.5)

exp(jk0 r)
4r

(3.6)

k2

and the corresponding Greens function is

G(r) =

S (r )

Figure 3.5.

G (r )

p (r )

S (k )

H (k )

p(k )

Space domain and spatial frequency domain description of the sound radiation phenomenon

PRELIMINARY VERSION

32

with k0 = /Vs denoting the wavenumber. This expression describes a diverging spherical wave.
Indeed, the constant phase surfaces are obviously r = const, a series of concentric spheres with
center in the origin. Moreover, assuming that the source is harmonic with frequency 0 , the
expression of the Greens function in time domain is
g(r,t) = R

exp(jk0 r)
exp(j0 t)
4r

cos(0 t k0 r)
4r

from which it is evident that the phase velocity is

Vph = 0 /k0 = Vs > 0
As another well known example of LSI system, let the frequency go to zero in (3.3), so that
the Helmholtz equation becomes Poisson equation. This, for example, relates the electric potential
V (r) to a charge distribution (r), which acts as its source:
2 V (r) =

(r)

The transfer function associated to this equation is (from (3.5))

H(k) =

1
k 2

and the corresponding Greens function (from (3.6))

G(r) =

1
4r

We recognize immediately this expression as the potential generated by a point charge q = 1 C in

a dielectric with permittivity .
We are ready now for Maxwells equations (3.1), which are still more complicated because in
addition to being multidimensional and multiple input/output, they are vector equations: this
means that the output is a vector that is not necessarily parallel to the input. This implies that
each of the four Greens function is not a scalar but a linear operator (a tensor ), which, in a basis,
is represented by a 3 3 matrix. This means that, differently from the case of sound waves, the
Greens function is not directly the field radiated by a point source. The source is really a point
but is also a vector, which can have all possible orientations. From a certain point of view, we can
say that the Greens tensor yields the field radiated in a given point by a point source in the origin
with all possible orientations. This concept will be better clarified in Section 3.2.
In coordinate-free language
E(r) = j0

GEJ (r r ) J(r )dr

GEM (r r ) M(r )dr

(3.7)

H(r) =

GHJ (r r ) J(r )dr j0

GHM (r r ) M(r )dr

To check the dimensions of the various Greens functions, it is useful to note that
0 = k0 Z0

0 = k0 Y0 = k0 /Z0

Hence we recognize that GEJ and GHM are measured in m1 , GEM and GHJ in m2 . The
explicit expressions of the various dyadic Greens functions can be obtained by applying the Fourier

PRELIMINARY VERSION

33

transform technique to (3.1). The most appropriate coordinate system that can be used to show
the result is the spherical one, because the source is a point. It can be shown that the matrices
representing the Greens functions are:

A 0 0
exp(jk0 r)
GEJ (r,,) = GHM (r,,) 0 B 0
4r
0 0 B

(3.8)
0 0
0
exp(jk
r)
0
GEM (r,,) = GHJ (r,,) jk0 0 0 C
4r
0 C
0

,
where the wavenumber is k0 = 0 0 . Note that the row and column indices are
r, ,

respectively. In dyadic form

exp(jk0 r)

+ B
GEJ (r,,) = GHM (r,,) = A
r
r + B

4r
exp(jk
0 r)

GEM (r,,) = GHJ (r,,) = jk0 C

4r
where
1
1
+
k0 r (k0 r)2
1
1
1
B =1 A=1j

2
k0 r (k0 r)2
1
C =1j
k0 r
A=2 j

Consistently with the fact that the source is a point, the Greens function does not depend on the
angular variables.
The behavior of the three coefficients at large distance from the source, k0 r (i.e. r

is
A=O

1
k0 r

B1
C1
In this region, usually called far field region, the expressions of the Greens functions simplify and
become
exp(jk0 r)
4r
exp(jk0 r)
GEM (r,,) = GHJ (r,,) jk0
r Itr
4r
GEJ (r,,) = GHM (r,,) Itr

(3.9)

where Itr is the transverse to r identity dyadic, see Appendix. This operator, when applied to an
arbitrary vector, produces as a result the projection of the vector on the plane perpendicular to r.
The operator
r Itr adds a further 90 counterclockwise turn around r.

PRELIMINARY VERSION

34

Conversely, in the near field region, k0 r 0 (i.e. r

), the coefficients become

2
(k0 r)2
1
B
(k0 r)2
1
C j
k0 r
A

3.2

Elementary dipoles

As discussed previously, the Greens function is the basic tool for the computation of the field
radiated by any source by means of eq. (3.7). However, it is convenient to start with the simplest
one, i.e. a point source, and this will help in understanding the properties of the Greens functions.
Consider first an elementary source of electric type located at the origin of the coordinate system,
modeled by the current distribution
J(r) = Me (r)
The vector Me is called the electric dipole moment of the current distribution and is measured
in Am (recall that the dimensions of the three dimensional function are m3 ). An arbitrary
current distribution can be characterized by its moments. This concept is used also in the theory
of probability: if is a random variable with density function W (x), moments of all orders can
be defined by
mn = E{ n } =

xn W (x)dx

where E{ } denotes the expectation value. In the case of the current distribution, the role of
W (x) is played by J(r), but the situation is more complicated because its vector nature implies
that the moments beyond the first are tensors. The first moment (dipole moment) is a vector,
defined by
Me =

J(r)dr

(3.10)

In the case of the point source introduced above, thanks to the properties of the delta function, the
previous equation becomes an identity and we understand the reason for the name of the coefficient.
From a practical point of view, we can imagine to obtain this source by a limiting process, starting
from a rectilinear current I, whose length l is progressively reduced without changing the aspect
ratio (diameter/length of the wire), while, at the same time, the current is increased, so that the
value of the integral (the dipole moment Il) remains constant.
Introducing the dipole current into (3.7), we find that the radiated fields are given by
E(r) = j0

GEJ (r r ) Me (r )dr = j0 GEJ (r) Me

(3.11)

H(r) =

GHJ (r r ) Me (r )dr = GHJ (r) Me

Since we know the expressions of the matrices representing the Greens functions in the spherical
basis, it is necessary to express the vector Me in the same basis. Let us assume that the polar
axis of the coordinate system is in the direction of Me , i.e. assume Me = Me
z. Note that
this step is allowed because the Greens function does not depend on the angular variables, as a

PRELIMINARY VERSION

35

consequence of the isotropy of free space. It is to be remarked, as a general rule, that the Greens
function depends only on the structure and, hence, shares its symmetries. This choice guarantees
the simplest description of the radiated field. Since the radiated field must have the direction of
Me as a symmetry axis, orienting the polar axis of the coordinate system in this direction allows
the expressions to be independent on .

+ (Me )
Me = (Me
r)
r + (Me )

+ Me (
= Me (
z
r)
r + Me (
z )
z )

(3.12)

= Me cos
r Me sin
where we have exploited (A.12). Now recalling the expression of the Greens function (3.8), we
obtain

A 0 0
cos
exp(jk0 r)
0 B 0 sin Me
E(r) = j0
4r
0 0 B
0
(3.13)
Z0 Me exp(jk0 r)

= j
A cos
r B sin
2r
where use has been made of

2
Z0

and Z0 is the wave impedance. Concerning the meaning of (3.12), it is to be remarked that the
matrix (3.8) represents the Greens function in the spherical basis consisting of the unit vectors

r, ,
defined in the observation point r. Hence, even if the source is located in the origin, its
components are evaluated in the basis associated to the point r.
We can proceed similarly for the magnetic field:

0 0
0
cos
exp(jk0 r)
0 0 C sin
H(r) = jk0
4r
0 C
0
0
(3.14)
Me exp(jk0 r)
=j
C sin

2r
0 = k0 Z0 =

In conclusion, the electromagnetic field radiated by an electric dipole has the following expression
E(r) = j

Z0 Me exp(jk0 r)
1
1
2 j
+
2r
k0 r (k0 r)2

cos
r 1j

1
1

k0 r (k0 r)2

sin
(3.15)

H(r) = j

Me exp(jk0 r)
2r

1j

1
k0 r

sin

This wave has two components of electric field and only one of magnetic field. Imagine a geographical system of coordinates such that the direction of the dipole moment defines the direction
of the earth axis. The angle is the colatitude (= 90 latitude), the angle is the longitude.
Then the electric field is contained in the meridian planes and the magnetic field is tangent to the
parallels. This type of wave is called TM (Transverse Magnetic) since the magnetic field has no
radial component. We recognize also that the radial component of the electric field is dominant
close to the source, but negligible with respect to the others at large distance. Here the wave is
essentially TEM, since neither field has a (significant) radial component.

PRELIMINARY VERSION

36

Let us compute the energy budget by means of the Poynting theorem. The Poynting vector is
S = E H =

Z0 Me2

BC sin2
r + AC sin cos
4r2 2

Compute the components

1
1
1
1
1
1
1
1
1+j
=1j

+j
+
j
k0 r (k0 r)2
k0 r
k0 r (k0 r)2
k0 r (k0 r)2
(k0 r)3
1
=1j
(k0 r)3
2
2
1
2
2
2
2
AC = j
+
1+j
=j
+

+j
k0 r (k0 r)2
k0 r
k0 r (k0 r)2
(k0 r)2
(k0 r)3
2
2
=j
+
k0 r (k0 r)3

BC = 1 j

Substituting in the previous equation we get

S=

Z0 Me2
4r2 2

1j

1
(k0 r)3

sin2
r+j

2
2
+
k0 r (k0 r)3

sin cos

According to Poynting theorem, the surface density of active power flow is

dP
1
= R{S
}
d
2

(3.16)

where
is the normal to the surface element. In order to compute the total radiated active power,
we have to evaluate the flux of the Poynting vector across a closed surface surrounding the source.
For maximum simplicity we choose a sphere of radius r:
1
1 2
R S
d =
2
2 0

2
1 Z0 Me
2
sin3 d
=
2 42
0
1 Z0 Me2 2
=
2 32

Prad =

Z0 Me2
sin2 r2 sin dd
4r2 2
(3.17)

Here we have used the following facts

the normal to the spherical surface is
=
r
the area element in spherical coordinates is d = r2 sin dd
the integrand does not depend on , so the integration yields the 2 factor
the integration yields

sin3 d =

4
3

The factor 1/2 has been left explicit to make it clear that Me is a peak value, that is the time
domain dipole moment is Me (t) = Me cos(0 t). If, on the contrary Me is an effective value, the
factor 1/2 has to be dropped.

PRELIMINARY VERSION

37

We notice that the total radiated power does not depend on the radius of the sphere chosen
to compute it. Algebraically this is the result of the cancellation between the r2 factor in the
denominator of the Poynting vector and the one in the area element d. To get a more physical
explanation, consider the fluxes through two concentric spheres of different radii: if they were
different, power would be lost or generated in the shell, which is impossible by conservation of
energy in a lossless medium.
In (3.17) we took the real part of the integral. The imaginary part is the reactive power
1
1
I S
d =
2
2
2
1
1 Z0 Me
2
=
2 4r2 2 (k0 r)3
1 Z0 Me2 2 1
=
2 32 (k0 r)3

Q=

Z0 Me2 1
sin2 r2 sin dd
4r2 2 (k0 r)3

sin3 d

It is reasonable that Q depends on r: the reactive power is the energy that twice per period is
exchanged between generator and load, hence crosses the spherical surface of radius r. Moreover,
it may be remarked that in the direction, the structure is closed as a (virtual) cavity. This
explains why the component of the Poynting vector is pure imaginary. Indeed, a real part of S
would imply a steady energy flow in that direction; however, this is impossible, since the angular
domain is finite (0 /2) and the dielectric is lossless.
It is useful to explicitly indicate the dominant components close to the source and far from it.
In the far field region, r

E(r) j

Z0 Me exp(jk0 r)

sin
2r
(3.18)

H(r) j

Me exp(jk0 r)
sin

2r

We see that the fields tend to be linearly polarized, orthogonal and proportional to each other and
also orthogonal to the radial direction. These properties are summarized in the impedance relation
1

r E(r)
Z0
E(r) Z0 H(r)
r

H(r)

The radiated field is a spherical wave (because of the factor exp(jk0 r)), diverging from the origin

with phase velocity c = /k0 = 1/ 0 0 . Indeed, the time varying electric field is given by
E(r,t) = R{E(r)ejt }
Z0 Me

sin sin(0 t k0 r)
2r
Note that there is no contradiction between the statement that the radiated field is a spherical
wave and the fact that this field has the sin dependence: the former refers to the constant phase
surfaces, the latter to a magnitude factor.
The field amplitudes decay as 1/r. This behavior is strictly connected with the principle of
conservation of energy. Indeed, it is easy to see that the total radiated active power computed in

PRELIMINARY VERSION

38

(3.17) is due only to these 1/r components.

The active power density per unit surface dP/d, at any distance, (d orthogonal to
r) is given,
according to (3.16), by
dP
1 Z0 Me2
=
sin2
(3.19)
d
2 4r2 2
Thus, in the far field,
dP
1 |E|2

d
2 Z0
exactly as in the case of a plane wave, see (2.13). This fact should not be surprising, since a spherical
wave can be locally approximated by a plane wave with direction of propagation coincident with
the radial direction. In the near field region r

E(r) j

Z0 Me
2
1

cos
r+
sin
2
2r (k0 r)
(k0 r)2
(3.20)

H(r)

Me 1
sin

2r k0 r

which may be reduced to

E(r) j

Z0 Me

2 cos
r + sin
8 2 r3
(3.21)

H(r)

Me
sin

4r2

We notice that the exponential has been dropped, since its value, for r
is essentially 1. Clearly,
very close to the source, the propagation delay is negligible. We can recognize that the electric
field coincides with that of a static dipole, whose moment, however, is a sinusoidal function of
time. Likewise, the magnetic field coincides with that created, in accordance with the Biot-Savart
law, by an infinitesimal current element, where the current is a sinusoidal function of time. This
regime is called quasi-static.
Fig. 3.6 shows a sketch of the electric field lines in a plane = const; close to the source the
field line behavior is similar to that of a static dipole, further away, entering in the radiation region,
they are completely different. Note that they are closed: we are accustomed to magnetic field line
being closed because B = 0, i.e. B is solenoidal. Actually in this case also the electric field is
(almost) everywhere solenoidal. Recall the divergence equation
E=

=0

for r = 0

because the source is point like and located in the origin.

Note that the image shown in the figure refers to a specific time; as time passes the contours move
outward radially. Fig. 3.7 shows the traditional way to plot the component of the electric field in
the far field region, in a plane =const. It is a polar plot and the curves have equation = sin
(with denoting the length of OP ), hence they are circles.
The other elementary source that we describe now is the dual of the electric dipole, i.e. a
magnetic dipole. We consider then formally a point source of magnetic current,
M(r) = Mm (r)

PRELIMINARY VERSION

39

Figure 3.6.

Dipole electric field lines

Figure 3.7. Dipole far field. Normalized polar plot of E and H at large distance. The maximum
field value is reached on the equator = /2 and is normalized to 1. For any the length of the
segment OP is proportional to the value of E

The vector Mm is called the magnetic dipole moment of the current distribution and is defined as
Mm =

M(r)dr

Its units are Vm. It is well known that magnetic currents do not exist, as flow of magnetic charges,
but can be introduced formally to describe electric currents circulating in closed loops. Indeed,
this elementary source can be imagined to be obtained with a limiting procedure from a loop,
enclosing an area S, on which the current I is circulating. The equivalent magnetic current has
a direction perpendicular the plane of the loop as indicated by the thumb of the right hand when
the fingers are aligned along the electric current flow. The point source is obtained by shrinking
the loop and increasing the current so that the magnetic dipole moment remains constant. It can

PRELIMINARY VERSION

40

be shown that the magnetic dipole moment of a small loop is

Mm = jSI
Notice that since = k0 Z0 , this quantity is measured in /m and Mm turns out with the right
dimensions. In order to compute the fields radiated by such a source, we substitute its expression
in (3.7) and obtain
E(r) =

GEM (r r ) Mm (r )dr = GEM (r) Mm

(3.22)

H(r) = j

GHM (r r ) Mm (r )dr = GHM (r) Mm

As in the case of the electric dipole, we choose a spherical coordinate system with the polar axis
aligned along Mm , which then takes the form Mm = Mm
z. Recalling (3.8), (3.12) and computing
the scalar products we obtain
E(r) = j

Mm exp(jk0 r)
2r

1j

1
k0 r

sin

(3.23)

Y0 Mm exp(jk0 r)
1
1
H(r) = j
2 j
+
2r
k0 r (k0 r)2

1
1
cos
r 1j

k0 r (k0 r)2

sin

If we compare these expressions with those of the fields radiated by an electric dipole (3.15), we see
that they (obviously!) satisfy the principle of duality, in the sense that the expressions of electric
and magnetic fields are exchanged in the two cases. This wave is of T E type since the electric field
has no radial component and tends to become T EM in the far field region.
The Poynting vector associated to these fields is
S = E H =

2
Y0 Mm

B C sin2
r + A C sin cos
2
4r 2

which is just the complex conjugate of the corresponding expression for the electric dipole. Hence
the result is written down by inspection
S=

2
Y0 Mm
2
4r 2

1+j

1
(k0 r)3

sin2
rj

2
2
+
k0 r (k0 r)3

sin cos

The total radiated active power is given by the real part of the flux of S through a sphere concentric
with the source and is
2
1 Y0 Mm
2
1
d =
(3.24)
Prad = R S
2
2 32
For completeness we report the expression of the reactive power
Q=

1
I
2

S
d =

2
1 Y0 Mm
2 1
2
2 3
(k0 r)3

which has the opposite sign with respect to the case of the electric dipole, since this is a magnetic
source. This must not be surprising: also in circuit theory the reactive power in inductors and
capacitors has opposite signs.

PRELIMINARY VERSION

41

It is useful to explicitly indicate the dominant components close to the source and far from it.
In the far field region, r

E(r) j

Mm exp(jk0 r)
sin

2r
(3.25)

H(r) j

Y0 Mm exp(jk0 r)

sin
2r

The impedance relation is the same as the one for the electric dipole fields:
1

r E(r)
Z0
E(r) Z0 H(r)
r

H(r)

In the near field region r

E(r)

Mm 1
sin

2r k0 r
(3.26)

Y0 Mm
2
1

H(r) j
cos
r+
sin
2
2r
(k0 r)
(k0 r)2
which may be reduced to
E(r)

Mm
sin

4r2
(3.27)

H(r) j

Y0 Mm

2 cos
r + sin
8 2 r3

Obviously, Figs. 3.6 and Fig. 3.7 can be used also for magnetic dipoles.

3.3

Radiation of generic sources

After the analysis of the two elementary sources, we can discuss the radiation properties of arbitrary
distributions of electric and magnetic currents. Actually, the solution of this problem has already
been given in (3.7): the integrals are generally to be evaluated by numerical techniques and the
task is absolutely non trivial, since the integrands are highly oscillatory due to the presence of
the exponential exp(jk0 r). However, when the point where the field is to be computed is far
from the source, it is possible to carry out a number of approximations that lead to a closed form
evaluation of the radiation integrals in a number of interesting cases. In this section we discuss
this approximation and find its domain of validity, obtaining a generalization of the concept of far
field region already introduced in the previous section.
Let us make reference to Fig. 3.8, which shows the geometry of the problem. The field radiated
in the observation point P (often called field point) is given by (3.7) that we repeat here for ease

PRELIMINARY VERSION

42
d

P
r'

O
J, M
Figure 3.8. Radiation of an arbitrary source. Geometry for the approximate evaluation of the
radiation integral. P is the field point, r denotes a generic point of the current distribution and
the vector there represents the elementary dipole J(r )dV

of reference
E(r) = j0
H(r) =

GEJ (r r ) J(r )dr

GHJ (r r ) J(r )dr j0

GEM (r r ) M(r )dr

(3.28)

GHM (r r ) M(r )dr

The meaning of these equations is very clear: the given source distribution is described as a
collection of elementary electric (J(r )dV ) and magnetic (M(r )dV ) dipoles located in the source
point r , the field radiated by each of them is computed, as in the previous section, and the partial
results are summed. In other words, it is just an application of the superposition principle implied
by the linearity of Maxwells equations. The position of P is specified by its spherical coordinates
r, , with respect to a (global) coordinate system with origin in O. Let us introduce the relative
position of P with respect to the source point r
d=rr
and assume that k0 d
be used:

1, so that the asymptotic expressions of the Greens functions (3.9) can

exp(jk0 d)
4d
(3.29)
exp(jk0 d)

GEM (d,d ,d ) = GHJ (d,d ,d ) jk0 d Itd

4d
where d, d , d denote the spherical coordinates of P with respect to the source point r . Now
suppose, as it is usual in a telecommunication application, that the field point is so far away
that the angle subtended by the source, when observed from the field point, is negligible. This
means that the vectors d and r can be considered to be parallel, so that the various local spherical
coordinate systems can be collapsed in the global one with origin in O, so that d and d .
Of course d is different from r and can be approximated as follows.
GEJ (d,d ,d ) = GHM (d,d ,d ) Itd

d=

dd=

(r r ) (r r ) =

r2 2r r + r 2 = r

The observation point P is assumed to be at large distance, so r /r

be expanded in Taylor series (binomial theorem)

1
1
1 + x = 1 + x x2 + . . .
2
8

1+

r
r

rr
r

1 and the square root can

PRELIMINARY VERSION

43

By letting
x=

r
r

rr
r

we obtain, by keeping only the terms up to the second degree in r /r

d=r 1

rr
1
+
r
2

r
r

1
+ (
r r )
2

r
r

+ . . .]

where the last term is the lowest degree one coming from the term (1/8)x2 . Factoring common
terms yields
r2
d = r
rr +
1 (
r r )2 + . . .
(3.30)
2r
This infinite expansion can be truncated at any point: more terms retained means more accurate
results but also greater difficulty in the evaluation of the radiation integral. The relative distance
d appears in two places in the Greens function, in the denominator and in the phase exponential.
To appreciate that the accuracy required in the two cases is different, let D be the diameter of the
smallest sphere containing the source. Then, as the source point sweeps the source volume
|d r|

D
,
2

D
d
1
r
2r

i.e.

(3.31)

Then substituting d with r (i.e. truncating the Taylor expansion at the first term) in the denominator produces an error that can be made as small as one wishes by assuming r large enough.
Concerning the exponential, (3.31) implies that
d=r+

D
2

with

|| 1

but then
exp[jk0 d] = exp[jk0 r] exp[jk0

D
]
2

and approximating d with r would imply an error that does not depend on r but only on the
electrical size of the source, so that it could not be made small by increasing r. In conclusion, the
phase term requires that in (3.30) at least the first two terms are used. The denominator is not
critical and we can take d r there. In conclusion, we use:
d r
rr
dr

in the exponential
in the denominator

In this case we say that the radiation integral is computed in the Fraunhofer approximation. If
also the quadratic term in (3.30) is used in the exponential, we have the Fresnel approximation ,
which clearly is more accurate or it has a wider domain of applicability.
Consider first the Fraunhofer approximation of which we establish the domain of applicability.
A sufficient condition for the validity of the approximation is that the phase error produced by
neglecting the quadratic term in (3.30) is small. In the worst case (
r r = 0) the phase error is
= k0

r2
2r

PRELIMINARY VERSION

44

and reaches the maximum value

max = k0

D2
D2
=
8r
4r

for r = rmax = D/2. A this point one could imagine that the requirement
1 is enforced.
However, the distance rmin that would result would be very large and traditionally the condition
max

is assumed instead. This condition yields for the domain of applicability of the Fraunhofer approximation
D2
r rmin = 2

The procedure seems to be a bit strange, but one must remember that the integrand is an oscillatory
function so that the sufficient condition
1 is not necessary. The distance rmin is called far
field distance or Fresnel distance: indeed it is the boundary between the far field region (also called
Fraunhofer region or radiation region) and the Fresnel region, which is closer to the source even if
not adjacent to it.
Assuming the field point P is in the far field region of the source, the expressions of the Greens
functions (3.29) become
exp(jk0 r)
exp(jk0 r
r)
4r
exp(jk0 r)
GEM (d,,) = GHJ (d,,) jk0
r Itr
exp(jk0 r
r)
4r
GEJ (d,,) = GHM (d,,) Itr

(3.32)

Remember that here two approximations have been performed: first r

and then r 2D2 /
and, clearly, both must be satisfied. If these expressions are substituted in (3.28) we obtain the
far field expressions to be used for an arbitrary source:
exp(jk0 r)
Itr
4r
exp(jk0 r)
+ jk0

r Itr
4r

E(r) j0

J(r ) exp(jk0 r
r)dr +
M(r ) exp(jk0 r
r)dr

exp(jk0 r)

r Itr J(r ) exp(jk0 r

r)dr +
4r
exp(jk0 r)
j0
Itr M(r ) exp(jk0 r
r)dr
4r

H(r) jk0

These bulky expressions can be rewritten in a simpler form, very similar to the one in (3.18)
pertaining to an elementary dipole:
E(r) j

Z0 exp(jk0 r)
Pe (
r)
2r
(3.33)

H(r) j

exp(jk0 r)

r Pe (
r)
2r

PRELIMINARY VERSION

45

where the generalized electric dipole moment Pe (

r) is defined by
Pe (
r) = Itr

J(r ) exp(jk0 r
r)dr Y0
r Itr

M(r ) exp(jk0 r
r)dr

(3.34)

The units of Pe (
r) are Am and, on comparing this expression with (3.10), we realize that the
generalization consists in the fact that
it depends on integrals of both electric and magnetic currents
the integrals contain also the exponential factor, so that the generalized electric dipole moment depends on
r. Moreover, due to the transverse identity Itr , Pe (
r)
r = 0, i.e. the
generalized dipole moment is always orthogonal to
r.
In the case of an electric dipole
J(r) = Me
z(r)
we find (from (3.18) more easily than from the definition) that

Pe (
r) = Me sin
and in the case of a magnetic dipole
M(r) = Mm
z(r)
(from (3.25) rather than from the definition)
Pe (
r) = Y0 Mm sin

It is interesting to note that, on recalling the definition of the triple Fourier transform of a
function of three space variables (3.4), the definition of the generalized electric dipole moment
(3.34) can be rewritten as
Pe (
r) = Itr F3 {J(r)}|k=k0r Y0
r Itr F3 {M(r)}|k=k0r

(3.35)

from which we see that the radiated field is related to the Fourier transform of the source distribution. In particular, the spectral component that contributes in a given observation direction
r
is
2
k=

This observation is important because it allows to derive some general properties of the radiated
field by basic theorems of the Fourier transform: for instance, a consequence of the uncertainty
principle of Fourier transforms is that a directive antenna must be large in terms of wavelengths.
From (3.33) we can derive some general properties of the field radiated by an arbitrary source
in the far field region:
whatever the source, the radiated field is a spherical wave, as indicated by the presence of
the exponential exp(jk0 r). It can be shown that close to the source the wavefronts have a
shape that more or less matches that of the source, but the propagation phenomenon modifies
them so that they transform smoothly into spheres.
the magnitude of electric and magnetic field decays as 1/r

PRELIMINARY VERSION

46

Pe ( , )

Pe ( , )

Figure 3.9. Polar plot of Pe (

r) for a fairly directive antenna. For the particular direction shown,
the spherical components of Pe are shown. Obviously Per 0

.
Pe (
r) for any source with a finite size in terms of wavelength is a function of , , which is
more rapidly varying than the sin typical of a point dipole. Indeed, the transform of the
Dirac delta function is a constant and the sin originates from the dot product Itr
z. Fig. 3.9
shows a polar plot of Pe (
r) for a fairly directive antenna. The plot is a surface constructed
as follows. For each direction specified by spherical angles , , as the one shown, a point
of the surface is identified: its distance r from the origin is proportional to the magnitude
of Pe (,). For the particular direction shown, also the spherical components Pe , Pe are
shown. Obviously the radial component is identically zero. It must be clear that the two
vectors shown are not tangent to the surface shown.
The plot shows that the magnitude of Pe (
r) (and then also of the radiated electric field) is
maximum in the direction of the z axis, which appears to be a symmetry axis of the antenna:
|Pe | does not depend on . One of the most striking characteristics of the plot is that it
shows that there are many directions for which |Pe | vanishes, so that the antenna does not
radiate any field in those directions. The explanation is simple. As (3.34) shows, Pe (
r) is
given by an integral, that is by a sum of many infinitesimal contributions (J(r )dV ), each
one weighted by a phase factor (exp(jk0 r
r)) that depends on the the observation direction.
For some directions (as
z in the figure) the various contribution sum in phase and the result
is large (constructive interference). For other directions ( = const. in the figure), the
contributions are out of phase and the end result is zero because of destructive interference.

PRELIMINARY VERSION

47

In other words, the zeros arise always because of cancellation. It is to be borne in mind
that this characteristic behavior is possible only because we assume that the source is time
harmonic, with a well defined frequency. In the case of a thermal source of light, i.e. a lamp
or a flame, the source is absolutely not time harmonic, but can be better described as a white
noise source: hence no constructive or destructive interference is possible and the plot is in
the form of a kind of sphere. Indeed, consider a certain direction: even if for one frequency
the various contributions give rise to cancellation, for other frequencies they do not and the
total field is never zero. This is the reason why two lamps on a table produce a uniform
brighter illumination of it. If the two lamps were ideally single frequency, characteristic dark
fringes (interference fringes) would appear
the Poynting vector has the following expression
S(r) = E(r) H (r)

Z0
Pe (
r) (
r Pe (
r))
4r2 2

Z0
|Pe (
r)|2
r (
r Pe (
r)) Pe (
r)
4r2 2
Z0
r)|2
r
2 2 |Pe (
4r

(3.36)

because the generalized dipole moment is transverse to

r. Since S is real and its field lines
are radial, active power flows in the radial direction. If we want to compute the total
radiated power, we have to compute the flux of the Poynting vector across any closed surface
surrounding the source. It is obviously convenient to choose a sphere with large enough
radius r, so that the previous expression can be used
Prad =

1
R
2

S
r d =

1 Z0
2 42

d
0

|Pe (, )|2 sin d

(3.37)

where we have used d = r2 sin dd. As in the case of an elementary dipole, the total power
leaving the sphere must be independent of its radius, so that we can simply talk about the
total power radiated by the source. Recalling the definition of solid angle d = sin dd, we
can introduce the concept of power density per unit solid angle (W/steradiant) or radiation
intensity
dPrad
1 Z0
=
|Pe (, )|2
(3.38)
d
2 42
Alternatively, we can introduce the power density per unit surface (oriented as the tangent
plane to the sphere of radius r through the observation point) by
1 Z0
dPrad
=
|Pe (, )|2
d
2 42 r2

(3.39)

If we compare this expression with (3.33), we realize that

dPrad
1 |E|2
=
d
2 Z0

(3.40)

which is the same expression as (2.13) for plane waves, as we pointed out in the case of the
dipole fields. Then Pe (
r) contains the information about the radiation pattern, i.e. the way
in which radiated power is distributed according to the direction. Moreover, it determines

PRELIMINARY VERSION

48

also the polarization. Indeed, it can be expressed in the spherical basis relative to any
observation direction as:
+ Pe (, )
Pe (, ) = Pe (, )

as shown in Fig. 3.9

If Pe and Pe have the same phase, the polarization is linear.
If Pe and Pe have the same magnitude and are in phase quadrature, the polarization
is circular.
In all other cases, the polarization is elliptical
It is finally to be remarked that any source that is significantly smaller than the wavelength
behaves as an elementary dipole, independently of the details of its actual shape. Indeed, in these
conditions, for all the points of the source the exponential in (3.34) is essentially equal to one and

Pe (
r) = Itr Me = Me sin
as for an elementary dipole. As a consequence, we cannot deduce the shape of a small antenna
from measurements of the radiated field carried out in the far field region. Alternatively, small
details of a large source cannot be inferred from measurements of the radiated field. This property
explains why no microscope can show details of an object that are smaller than a wavelength.

Chapter 4

Antennas
Antennas are fundamental components in wireless telecommunication systems. Information transfer between transmitter and receiver is carried out by free space propagation and antennas provide
the necessary coupling between circuits and open space. Even if at sufficiently high frequency any
circuit radiates electromagnetic energy, antennas are designed to do it in a particularly efficient
way.
Depending on the frequency range and on the applications, antennas may have very different
forms. If the frequency of operation is very small (and consequently the wavelength is very large)
ordinary antennas are forced to be electrically small for size limitation reasons. Usually they
are in the form of dipoles or loops constructed with metal wires or rods. They radiate energy
all around them and are in general inefficient. At intermediate frequencies, antennas can have
a size comparable with the wavelength: widespread is the use of /2 dipoles or /4 monopoles;
loops with one wavelength circumference are also used. These antennas are still not directive
but their efficiency is larger than that of small antennas. Directive antennas, i.e. such that can
radiate electromagnetic energy essentially in one direction, must be electrically large, hence are
practicable only at very high frequency. Typical examples are reflector antennas, which exploit
the same working principles of optical telescopes. Antennas for satellite telecommunication belong
generally to this class. Similar properties have antenna arrays: they consist of electrically large
periodic arrangements of small antennas (half-wave dipoles for example).
In this chapter the main types of antennas will be described, but first the parameters necessary
to quantify their performance must be introduced

4.1

Antenna parameters

Antennas are used in transmit and in receive mode. Even if their behaviors in the two operating
modes are strictly related, they are characterized by different parameters.
Transmitting antennas behave as loads for their feeding transmission line, hence are characterized by an input impedance or a reflection coefficient. Moreover the power density radiated in the
various directions is different: parameters are necessary to describe this property.
Receiving antennas behave as generators as seen by the relevant receivers. Hence they can be
characterized by an internal impedance and by an open circuit voltage in a Thevenin equivalent
49

PRELIMINARY VERSION

50

circuit. Also, their sensitivity to waves incident from various directions is different and has to be
quantified.

4.1.1

Input impedance

Consider an antenna with a couple of terminals, such as the one on the left in Fig. 4.1 The input

Ia
Va
in

Figure 4.1. Left: wire antenna (dipole) with a couple of terminals at which voltage and current
can be defined. Right: horn antenna connected to a waveguide; in this case only the reflection
coefficient for the fundamental mode can be defined

impedance is obviously defined as

Zin = Rin + jXin =

Va
Ia

The presence of a real part means that the active power (Pin ) is absorbed from the feeding line:
part of this is radiated away (Prad ) and the rest is dissipated into heat due to the limited metal
conductivity (Ploss ). The imaginary part of the input impedance is related to the energy stored in
the reactive fields in close proximity of the antenna structure. We can define the antenna radiation
efficiency by
Prad
=
(4.1)
Pin
Obviously 1. Microwave antennas have very high efficiency,
100kHz it can be as low as = 0.1.

1, but at a frequency of

Since Pin = 12 Rin |Ia |2 , it is customary to introduce the antenna radiation resistance as
Rrad =

2Prad
|Ia |2

and the efficiency can also be written as

=

Rrad
Rin

Clearly the radiation resistance is an equivalent resistance and has nothing to do with the Joule
effect. Notice that there is no contradiction in the fact that radiation in a lossless medium is

PRELIMINARY VERSION

51

described by a resistance: indeed radiated power is irreversibly transferred from the generator to
infinity, just as electrical power flowing in a copper conductor is irreversibly transformed from
electrical to thermal nature. Notice also that there is no radiation reactance, because this is
a contradiction in terms. Indeed, radiation is related to a steady (one directional) energy flow,
whereas the input reactance is associated to the energy exchanged twice per period back and forth
between the antenna and the surrounding space. It turns out that the antenna behaves as a circuit
with many resonance frequencies, in correspondence of which the input reactance vanishes.
It is to be remarked that there is a relationship between the generalized dipole moment Pe (
r)
and the radiation resistance Rrad , which originates from (3.37):
Rrad =

2Z0
42 |Ia |2

d
0

|Pe (, )|2 sin d

(4.2)

In the case of the horn antenna shown in the right part of Fig. 4.1 the definition of an input
impedance is conventional, whereas the input reflection coefficient (for the fundamental mode) is
well defined and similar considerations can be carried out.

4.1.2

Radiation pattern, Directivity and Gain

Antennas radiate the power they get from the transmitter with different density per unit solid
angle in the various directions . The radiation pattern is a polar plot of this density and has the
form of a surface as shown in Fig. 4.2: for each direction , , a point is defined at a distance r
from the origin proportional to dPrad /d. The set of all these points has the characteristic shape
shown in the figure. The antenna is moderately directive and exhibits a major lobe (or main
lobe) and a number of minor lobes (side lobes) Associated to the antenna is a spherical coordinate

Figure 4.2.

Coordinate system for antenna analysis

PRELIMINARY VERSION

52

system that is the most appropriate to describe the field radiated in the Fraunhofer region (far
field region). A sphere is shown with radius greater than the Fresnel distance and an elementary
patch on it d. Since d = r2 d, we can say that the radiation pattern is a plot of dPrad /d
as the patch d sweeps the spherical surface. The figure shows also the three basis vectors in a
particular observation point.
Often two-dimensional patterns are used to describe the characteristics of an antenna, instead
of the three-dimensional one. An example is shown in Fig. 4.3. The patterns are always normalized

Figure 4.3. Normalized two-dimensional patterns. (a) is a polar plot of the field magnitude in
linear scale. (b) is a polar plot of the power density in linear scale. (c) is a polar plot of the power
density in dB. In each plot the Half Power Beam Width (HPBW) is indicated. The lobes are marked
alternatively by a plus and a minus to indicate that the field has a phase 0 and rad, respectively

PRELIMINARY VERSION

53

to the maximum. Sometimes field patterns instead of power patterns are used: to appreciate the
relationship between the two it is to be remembered that the power density is proportional to the
square of the field. Thus in a power pattern using linear scales, sidelobes become less visible. On
the contrary, a dB scale makes sidelobes more visible. It is clear that the normalized field pattern
coincides, apart for a scale factor, with the polar plot of Pe (, ), shown in Fig. 3.9.
An important characteristic of a pattern is the width of the main lobe. Usually the Full
Half Power Beam Width (FHPBW) is used for this purpose. The main lobe may be rotationally
symmetric or not. In the latter case two FHPBW must be specified, relative to two orthogonal
planes. If the field radiated along the axis of the main beam is linearly polarized, these planes are:
the E-plane, defined by the main lobe axis and the E vector
the H-plane, defined by the main lobe axis and the H vector
The directivity of an antenna is a function of , defined as the ratio between the radiation
intensity for that direction and the average of the radiation intensity over all directions. The
average is simply obtained as the ratio between the total radiated power and the total solid angle.
Hence:
dPrad
d (,)
d(,) =
(4.3)
Prad
4
The denominator can also be interpreted as the radiation intensity that would be produced by
a (hypothetical) isotropic radiator, radiating the same total power as the actual antenna. The
directivity is a pure number and its polar plot coincides with the power pattern, apart for a scale
factor. The maximum value of the directivity function, i.e. the one corresponding to the main
lobe direction, is called simply the antenna directivity, typically expressed in dB:
D = 10 log10 max d(,)
A property of the directivity function is that its average value over all directions is one. Indeed
1
4

d(
r)d =

1 1
4 Prad
4

dPrad
1 1
d =
Prad = 1
d
4 Prad
4

A small antenna has in general low directivity. By definition, an isotropic radiator, has unit
directivity. It can be proved that such an antenna is not physically realizable. Using (3.19) we find
that elementary dipoles have directivity D = 3/2. Because of the previous property, an antenna
with large directivity has necessarily a narrow main lobe and viceversa.
Closely related to the directivity is the gain of an antenna. The definition is
dPrad
d (,)
g(,) =
Pin
4
This means that the radiation intensity in a given direction is compared with the radiation intensity
that would be produced by a lossless isotropic radiator fed with the same input power as the actual
antenna. Hence the plots of both the directivity and gain functions are normalized power patterns,

PRELIMINARY VERSION

54

the difference being just the normalization factor. The directivity is a purely geometrical parameter,
whereas gain takes also antenna efficiency into account. Recalling (4.1), we find that
g(,) = d(,)
Sometimes gain and directivity are defined as ratios of power densities per unit surface, produced
at the same distance r by the actual antenna and by the isotropic radiator:
dPrad
d (,)
g(,) =
(4.4)
Pin
4r2
As for the directivity, when a single number is indicated as the gain of an antenna, it is meant to
be the maximum gain, usually expressed in dB:
G = 10 log10 max g(,)
A useful approximate formula allows to estimate the maximum directivity. If the antenna is very
directive, and the side lobes can be neglected,
D

32400
1d 2d

where 1d , 2d are the FHPBW (in degrees) in two orthogonal planes

4.1.3

Effective area, effective height

An antenna in receive mode behaves as a generator for the receiver, as shown in Fig. 4.4. The wave
incident on it is generally a spherical wave, but since it is produced by a very far away transmitting
antenna, it can be approximated by a plane wave in the neighborhood of the receiving antenna.
It turns out that if the polarization of this plane wave is changed while the incidence direction is
kept fixed, the received power changes. In other words, the sensitivity of the receiving antenna
depends also on the polarization of the incident field. This concept may seem obvious in the case
of the wire antenna. Since the current on it can flow only in the direction of the wire and since this
current is produced by the incident electric field that applies forces on the electrons in the metal
and sets them in oscillation, it is to be expected that the antenna response is maximum when the
electric field is parallel to the wire. The aperture antenna is apparently symmetrical, hence this
property is not so clear. The explanation is that even though the feeding waveguide is circular,
only one well defined polarization is used, which is fixed by the waveguide-coaxial cable transition,
not shown in the figure.
The internal impedance of the equivalent generator is denoted by Zg . It can be proved, by
means of the theorem of reciprocity, that Zg coincides with the input impedance Zin of the same
antenna when used in transmit mode. The generator itself can be characterized by means of
the available power or by means of the open circuit voltage Vg . As a consequence, two antenna
parameters can be defined.
The effective area (or equivalent area) in a particular direction is the ratio of the available power
at the output terminals to the power flux density (per unit surface) of a plane wave incident from
that direction, the incident wave being polarization matched to the antenna:
Aeq (,) =

max Pavail
dPinc
d (,)

(4.5)

PRELIMINARY VERSION

55

Ei
+

( , )

Vg

Zg

(a)
(c)

Ei

( , )

(b )
Figure 4.4. Receiving antennas and their equivalent circuit. (a) Wire antenna with an incident
plane wave coming from direction (,). (b) Horn, as an example of an aperture antenna, with an
incident plane wave. (c) Thevenin equivalent circuit of both types of antennas

where it is understood that the maximization regards only the polarization of the incident wave.
The effective length (or effective height) in a particular direction is the ratio of the open circuit
voltage at the output terminals to the magnitude of the electric field of a plane wave incident from
that direction, the incident wave being polarization matched to the antenna:
hef f (,) =

max Vg
|Ei ||(,)

Note that the effective height is a vector quantity: its direction is that of the incident electric field
that is polarization matched to the antenna. Hence the effective height defines the polarization of
the antenna in receive mode.
As a matter of principle, both parameters can be used to describe both types of antennas.
However, it is clear the equivalent area is concept is more appropriate for aperture antennas and
the effective height is more appropriate to wire antennas. Actually it is customary to define an
aperture efficiency
max Aeq (,)
a =
Ag
where Ag is the area of the geometrical aperture and the maximization is performed with respect
to the direction of the incoming plane wave. Likewise the length efficiency of a wire antenna is
defined by
max hef f (,)
l =
g

where

is the length of the antenna.

PRELIMINARY VERSION

56

In general the theorem of reciprocity allows to establish a relationship between the receive
mode and transmit mode properties of a same antenna. It can be shown that the effective area is
connected to the gain by
2
Aeq (,) =
g(,)
(4.6)
4
and that the effective height is connected to the generalized electric dipole moment by
hef f (,) =

1
Pe (,)
Ia

(4.7)

From this we learn that the polarization of an antenna in receive mode, relative to a particular
direction, coincides with the polarization of the electric field radiated in that direction by the same
antenna in transmit mode.
From an application point of view, it is useful to compute the equivalent generator in the case
the incident wave is not polarization matched. If p
T X , p
RX are the polarization unit vectors of
the incident wave and of the receiving antenna, respectively, it can be shown that
Vg = hef f (,)|Ei |
pT X p
RX
and
Pavail = Aeq (,)

dPinc
d

|
pT X p
RX |

(4.8)

(4.9)

(,)

Notice that the polarization of the incident wave has been denoted by p
T X , because this is exactly
the definition of the polarization of the transmitting antenna that produces this wave.
Several relationships exist among the various parameters introduced up to now.
Radiation resistance
From (4.2)
2Z0
42

Rrad =

d
0

|hef f (, )|2 sin d

(4.10)

Effective height and gain.

Recall the definition of gain (4.4) and (3.40)
dPrad
|E|2
d (,)
2Z0 (,)
g(,) =
=
Pin
Pin
2
4r
4r2

(4.11)

Moreover, from (3.39) and (4.7)

1 Z0 |Ia |2
dPrad
=
|hef f (, )|2
d
2 42 r2
and Pin = 12 Rin |Ia |2 . Substituting we get
g(,) =

1 Z0
|hef f (, )|2
2 Rin

(4.12)

PRELIMINARY VERSION

57

Equivalent area and effective height.

From (4.9) the available power is
Pavail = Aeq (,)

1 |Ei |2
2
|
pT X p
RX |
2 Z0

From the equivalent circuit of Fig. 4.4, we have

Pavail =

1 |Vg |2
1 |hef f (,)|2 |Ei |2 |
pT X p
RX |2
=
2 4Rg
2
4Rin

From the comparison between the two, the desired relationship follows
Aeq (,) =

Z0
|hef f (,)|2
4Rin

This relationship could also be derived by direct application of (4.6)on (4.12).

Directivity and effective height.
From (4.3), (3.38) and (4.7), we get
dPrad
d
d(
r) =
Prad
4

1 Z0 |Ia |2
|hef f (
r)|2
2
2
4
=
1 Z0 |Ia |2 1
|hef f (
r)|2 d
2 42 4

and cancelling common factors

d(
r) =

|hef f (
r)|2
1
4

|hef f (
r)|2 d

Radiated electric field.

From (3.33)
E(r) j

Z0 Ia exp(jk0 r)
|hef f (
r)| p
T X
2r

Now invert (4.12)

E(r) j

Z0 Ia exp(jk0 r)

2r

Rin
g(
r) p
T X
Z0

and use again Pin = 12 Rin |Ia |2 to find

E(r) j

4.2

Z0 exp(jk0 r)
2r

2Pin
g(
r) p
T X
Z0

Friis transmission formula

When designing a radio link the fundamental problem is to compute the received power knowing
the transmitted power, the characteristics of the antennas and their distance. The solution is given
by Friis formula, that we derive.

PRELIMINARY VERSION

58

Pin

Pavail

GTX p TX

GRX p RX

R
Figure 4.5.

Geometry for the Friis transmission formula

Consider a transmitting antenna with gain GT X in the direction of the link and polarization
unit vector p
T X , fed with the power Pin , as shown in Fig. 4.5. Let GRX , p
RX , the corresponding
parameters of the receiving antenna, placed at a distance R from the transmitting one. This
distance must be large enough so that the two antennas are one in the far field region of the other.
This condition is always satisfied in all common situations. In the cases where the relative distance
is small, Friis formula must be substituted by a more complex one.
From the definition of equivalent area (4.5) we can relate the available power at the output port
of the receiving antenna to the incident power flux density
Pavail = Aeq

dPinc
2
|
pT X p
RX |
d

Now the incident power flux density is created by the transmitting antenna and we can relate it
to the input power to the transmitting antenna via the definition of gain (4.4)
dPinc
dPrad
Pin
=
= GT X
d
d
4R2
hence

Pin
2
|
pT X p
RX |
4R2
It is convenient to express the effective area of the receiving antenna in terms of its gain by (4.6),
so that the Friis formula reads
Pavail = Aeq GT X

Pavail = Pin

GT X GRX
4R

|
pT X p
RX |

(4.13)

This result is remarkably simple and one could be perhaps kind of astonished, after having seen
the complicated equations yielding the field radiated by an antenna, e.g. (3.33) and (3.34). The
explanation, however, is straightforward: all the complications are hidden in the gains of the
antennas (see (4.12)), whose definition was exactly conceived with a view to a simple form of the
Friis transmission formula!
However, it must be remarked that the gain of an antenna can be measured fairly easily as
suggested by (4.13). First two identical antennas are placed one in front of the other at a convenient
distance. Then the received power is maximized by changing the polarization of the transmitting
antenna: in the case of linearly polarization, this requires just a rotation of the antenna around its
axis. Finally, from a measurement of the input power to the T X antenna and of the received power,
it is simple matter to obtain the maximum gain of each antenna. If this operation is repeated for
all frequencies in a band, a calibrated reference antenna has been obtained. This can now be
used to measure another antenna, whatever its characteristics, by the same technique. Also, if the
reference antenna is kept fixed while the antenna under test is rotated, as shown in Fig. 4.6, its

PRELIMINARY VERSION

59

Figure 4.6. Measurement of the radiation pattern of an antenna under test by means of
a calibrated reference antenna

gain function is measured for any direction , . Of course, the spherical coordinate system is
attached to the rotating antenna, so that the line joining the two antennas has the direction ,
in this reference.

4.3

Examples of simple antennas

We describe in the following a few types of antennas, commonly used in the applications. The
structures we describe are made of metal, whose conductivity is so high that, as far as the computation of the radiation pattern is concerned, can be considered infinite. Wire antennas have a
pair of terminals that are connected to a transmission line. When the antenna is in operation,
it absorbs a current Ia from the transmission line, which then gives rise to a current distribution
J on the surface of all the parts of the antenna. This current is the source of the radiated field
and Eq.(3.34) allows us to compute it. We face two problems here, however. When (3.34) was
derived, it was assumed that the current distribution J(r) was radiating in free space and not
in presence of the metal structure. Indeed, the Greens function we have found in section 3.1 is
that of free space. The second problem is that the distribution J(r) is not known, because it is
not under our direct control. When we solved the radiation problem, we assumed to know the
current distribution and, on the basis of that, we were able to compute the radiated field. In the
antenna case, the only thing that we can control is Ia but nothing else. Hence the question arises:
according to what rule does the current distribute on the parts of the antenna? The answer lies
in the boundary conditions that we introduced in (1.16): the current distribution has such a form
that the radiated electric field has zero tangential component on the antenna surface. In this way
all the electromagnetic field components in the perfect conductor are zero, as they should be. It
is clear that this prescription does not help us very much in finding the expression of J(r), since
mathematically it entails the solution of a complicated integral equation! However, we have the
answer to the first question we asked above. Indeed, if the electromagnetic field inside a volume
is zero, we can freely change the constitutive relations of the material in that volume. A similar
situation exists in circuit theory. Suppose that in a circuit the element ZL is connected between
nodes A and B and suppose it is known that both the current IA and the voltage VAB are zero.
Ohms law VAB = ZL IA becomes the identity 0 = ZL 0, which is satisfied for any ZL . Hence the
value of this impedance can be changed and chosen at will, without affecting the rest of the circuit.
Similarly, in the electromagnetic case, if both the electric and magnetic fields in the metal are zero,
the metal can be substituted with any other dielectric, without any consequences in the rest of
the structure. The only sensible choice in this case is to substitute the metal with free space, so
that the medium is altogether homogeneous and the free space Greens functions (3.8) can be used
safely to compute the radiated field everywhere and in particular in the far field region.
To better appreciate what we have found, consider another case, where a given source J(r), for
instance an electric dipole, is supposed to radiate in presence of a metal sphere, placed at a certain

PRELIMINARY VERSION

60

distance. In this case the metal cannot be substituted with free space, because the given source is
arbitrary and does not produce zero field in the region occupied by the metal. Nevertheless, if the
metal is a perfect conductor, the field there is zero, but then it is the Greens function that takes
care of that. In conclusion the Greens function to be used in this case is not the free space one
and
E(r) = j G(r,r ) J(r )dr
Note that this is not a convolution integral, because the Greens function G(r,r ) depends on the
coordinates of both the source r and the field point r separately and not only on their difference
r r : this is an obvious consequence of the fact that the system is no longer space invariant, i.e.
free space containing the metal sphere is no longer homogeneous.
Another way to look at this problem is to fix the attention on the current distribution J that is
induced on the surface of the sphere. The form of it is such that the total electric field radiated by
J(r) and J has zero tangential component on the PEC sphere. Then, as before, since the total E
field in the metal sphere is zero due to the current distribution, we can substitute the metal with
free space and use the free space Greens function. In conclusion we have two alternatives:
use only the independent current distribution, but use a complicated Greens function (generally not known explicitely) to enforce the vanishing E field in the metal
introduce also the unknown induced current distribution, but use the free space Greens
function to compute the radiated field
In the following applications we will always use the second approach.
These simple considerations are formulated in a more rigorous form by the equivalence theorem.

4.3.1

Wire antennas

Consider an antenna made of a metal wire of small diameter and with a shape described by the
parametric equation r = r (s), where s is the arc length parameter. Let I(s) be the electric current
distribution on it (measured in A). The electric current density can be written

I (s)

P
r ( s)

r
O
Figure 4.7.

Wire antenna of shape r = r (s). P is the field point

J(r) = I(s)(r r (s))

s
where
s is the unit vector tangent to the curve, is a 2D Dirac delta function with support on
the curve r = r (s). Its role is to express mathematically the statement that the current flows on
a wire of negligible diameter. The magnetic current for this source is obviously zero. This is the

PRELIMINARY VERSION

61
z

P
r

Ia
y

Va
x

(a)
Figure 4.8.

(b )

Linear antenna. (a) Actual shape. (b) Computational model

current that was called J above. Rigorously, its expression should be found by solving an integral
equation, but sometimes it can be estimated with reasonable accuracy.
If we want to compute the electromagnetic field radiated in the far field region, it is enough to
evaluate the generalized electric dipole moment of this source. According to the definition (3.34):
sb

Pe (
r) = Itr

I(s)(r r (s))
s exp(jk0
r r )dr = Itr

I(s)
s exp(jk0
r r (s))ds

(4.14)

sa

The integration limits define the beginning and the end of the wire. This is a very general formula,
which can be applied for any shape of the wire and any current distribution on it. In general the
integral has to be evaluated by numerical quadrature, but there are some cases in which a closed
form result is obtained.
Short dipole
As a first example we consider a rectilinear antenna of length , as shown in Fig. 4.8. The current
distribution is I(z) can be assumed to be constant if
, say /10. This simple antenna
is called short dipole or Hertzian dipole. Sometimes two small spheres or plates are added to the
wire ends to help creating a constant current distribution. In this case
s=z

s=
z

Hence

/2

Pe (
r) = Itr
z

r r (s) = z cos

/2

sin Ia
Ia ejk0 z cos dz =

/2

Here we have used

I(z) = Ia

ejk0 z cos dz

/2

+
sin
Itr
z = (
)

z =

recalling (A.4). The integral is

/2
/2

ejz dz =

ejz
j

/2

=
/2

ej

/2

ej
j

/2

2j sin( /2))
=
j

sin( /2))
( /2)

PRELIMINARY VERSION

62
Fe(x)

1
0.8
0.6
0.4
0.2
0
0.2
0.4
0

10

x
Figure 4.9.

Plot of the function FE (x)

It is useful to introduce the function FE (x), shown in Fig. 4.9:

FE (x) =

sin x
x

(4.15)

Note that FE (x) is even and vanishes for x = n, n = 1,2, . . .. Moreover the side lobes are about
13.4 dB below the maximum.
In conclusion, the required dipole moment is
sin Ia
Pe (
r) =

sin(k0 cos /2))

sin Me FE (k0 cos /2)
=
(k0 cos /2)

where we have introduced the dipole moment Me = Ia . The maximum value of the argument of
FE is /, which is very small, so that FE
1 and the Hertzian dipole has the same pattern
as the elementary dipole. Notice that the previous formula can be used for any value of , even
greater than . However, if the antenna is long, the current I(s) cannot be assumed to be constant
and the previous formulas are no longer applicable. We list here the main properties:
The effective height (4.7) is

hef f (,) = sin

The radiation resistance is, see (4.10) and (3.17)

Rrad =
For example, if
large current Ia

2
2Prad
= Z0
2
|Ia |
3

= /20, the radiation resistance is Rrad

2. This means that a fairly
10 is necessary to radiate a power Prad = 100W.

PRELIMINARY VERSION

63

If we want to estimate the efficiency of short dipoles we have to compute the dissipated
power. It can be proved that the resistance per unit length of a wire with radius rw , length
and conductivity is
1 1
Rdiss =
2rw
where the denominator of the second fraction is an estimate of the area of the annulus with
radii rw and rw , with denoting the skin depth (see(2.20)), in the assumption of well
developed skin effect /rw
1. Consider for example a dipole with length = 1 m for
frequencies in the range between 100kHz and 30MHz, for which / varies between 3.3 104
and 0.1, so that the dipole is always short. Fig. 4.10 shows plots of radiation resistance Rrad
and of wire resistance Rdiss in the case the wire radius is rw = 1mm. Notice that in this
frequency range, the skin depth varies between 0.2 and 0.01 mm. Fig. 4.11 shows a plot
Radiation and loss resistance []
2.5

2
(a)
1.5

0.5

(b)

0
0.02

0.04

0.06

0.08

0.1

L/

Figure 4.10.

Plot of Rrad (a) and Rdiss (b) in the case of a short dipole with radius rw = 1 mm

of the efficiency for the same short dipole, when the wire radius is rw = 1, 3, 5 mm. We
see clearly that the the efficiency of very short dipoles can be very low, but is almost 1 for
dipoles about /10.
The directivity is the same as that of the elementary dipole: recalling (4.3),(3.19), (3.17), we
find
dPrad
Z0 Me2
sin2
d (,)
2
3
4
d(,) =
= sin2
=
2
Prad
2
1 Z0 Me 2 1
4
2 32 4
Hence the maximum directivity is D = 1.5 and the maximum gain G = 1.5.
Half-wavelength dipole
We consider now a dipole whose length L is comparable with the wavelength. In this case we
cannot suppose that the current I(s) is constant. It can be shown that a good approximation of

PRELIMINARY VERSION

64

efficiency
1
(c)
0.8

0.6

(b)

(a)

0.4

0.2

0
0

Figure 4.11.

0.02

0.04
0.06
L/lambda

0.08

0.1

Plot of the short dipole efficiency. (a) rw = 1 mm, (b) rw = 3 mm, (c) rw = 5 mm

the current distribution is

I(z) =

Ia
sin k0
sin k0 L/2

L
|z|
2

for

L
L
z
2
2

This current distribution is shown in Fig. 4.12 for two lengths of the dipole, i.e. for L = /2 and
3/2: if the geometrical length of the wire is the same in the two cases, the frequency in the second
case is three times the one in the first case. Note that the current is zero at the dipole ends, since
1

0.5

0.5

1
0.5

0
z/L

0.5

Figure 4.12. Plot of the current distribution on a dipole /2 and 3/2 long. A sketch of the dipole
is superimposed on the current plot

no plates (with the goal of providing capacitance) are present there, as in the case of the Hertzian
dipole. It can be shown that the generalized dipole moment, computed by (4.14) is
Pe (,) = Ia

cos(k0 L cos /2) cos(k0 L/2)

sin

PRELIMINARY VERSION

65
FH(x)

1.2
1
0.8
0.6
0.4
0.2
0
0.2
0

10

x
Figure 4.13.

Plot of the function FH (x)

The case of particular interest in the applications is that of L = /2, i.e. the half wave dipole.
The relevant generalized dipole moment is
Pe (,) = Ia sin

cos( cos /2)

sin2

that we write as

Pe (,) = Ia sin FH
cos

2
where the function FH (x) is defined by
FH (x) =

cos x
1

2x

(4.16)

and its plot is shown in Fig. 4.13 Note that FH (x) is even and vanishes for x =
n = 1,2, . . .. Also, the side lobes are about 23 dB below the maximum.

(2n + 1)
,
2

The maximum value of the argument of FH is /2, reached for = 0. The pattern turns out
to be slightly more directive than that of the elementary dipole, as shown in Fig. 4.14 The main
properties of the half-wavelength dipole are
Maximum directivity D = 1.643, to be compared with that of the elementary dipole D = 1.5.
Full Half Power Beamwidth =78 , to be compared with 90 of the elementary dipole.
Radiation resistance Rrad
73 . This is the most important property of half-wavelength
dipole antennas. Indeed, matching to transmission lines that typically have characteristic

PRELIMINARY VERSION

66

90

1.5
60

120

1
30

150
0.5

180

210

330

240

300
270

Figure 4.14. Directivity pattern of the /2 dipole (solid line) and of the elementary
dipole (dashed line). The maximum directivity is D = 1.643 for the /2 dipole and
D = 1.5 for the elementary dipole.

impedance Z = 50 or Z = 75 is very simple. Moreover, to radiate a power Prad =

100 W requires only a current Ia = 1.65 A instead of 10 A, as in the case of the /20 dipole
previously examined. Hence, the efficiency is higher. Finally, note that if the radiation
resistance of the half-wavelength dipole is computed according to the equation derived for
short dipoles, the value Rrad = 197 is obtained. The difference is due to the different
current distribution I(z) assumed in the two cases.
It is to be remarked that the dipole antenna requires a balanced feeding transmission line, such
as a twin-lead one (two-wire transmission line). A coaxial cable is an inherently unbalanced line:
if the inner conductor is connected to a dipole arm and the outer to the other, the result is that an
unwanted current flows to ground on the outer surface of the outer conductor, with a degradation
of the antenna performance. To avoid this phenomenon, a balun (balanced to unbalanced) must
be placed between the coax and the dipole.

L=

Figure 4.15. Half-wavelength dipole realized by a /4 monopole on a metal ground plane and fed
by a coaxial cable, whose outer conductor is bonded to the plane.

PRELIMINARY VERSION

67

Sometimes a half-wavelength dipole is realized in the form of a /4 long monopole, mounted

perpendicularly to a metal ground plane, as sketched in Fig 4.15. If this ground plane were infinite,
image theorem would guarantee the perfect equivalence. In practice some degradations occur.
Loop antenna
Loop antennas are generally classified into two groups, electrically small or large depending on
whether the circumference is smaller than /10 or larger, up to one wavelength. If the loop is
small, its actual shape plays no role: circular, square or elliptical loops have the same radiation characteristic of an elementary magnetic dipole, directed perpendicularly to the loop plane.
Moreover, radiation is maximum in the loop plane. As the circumference is increased toward one
wavelength, the maximum of the pattern shifts from the plane of the loop to the direction of the
axis.
Small loops are very poor radiators, because of their very low radiation resistance. Radiation
resistance can be increased up to a value comparable to that of typical transmission lines by
increasing the number of turns or by winding the loops on a ferrite core.
Concerning the equivalence between small loops and magnetic dipoles, it can be shown that
the magnetic dipole moment of a loop embracing a surface S, on which the current distribution is
constant and equal to Ia , is
2
Mm = jSIa =
Z0 SIa

Then the radiated fields can be computed by means of (3.23). As it was explained in detail there,
the radiation pattern of the magnetic dipole is the same as that of the electric one. The difference
is in the fields: the magnetic dipole radiates a magnetic field with (dominant in far field) and r
components, and an electric field with only a component, parallel to the loop plane.
The radiation resistance is computed, according to the definition, by means of (3.24):
Rrad =

2Prad
2
= Z0
2
|Ia |
3

kS

= 20 2

If the loop antenna is made with N turns, the previous radiation resistance has to be multiplied
by N 2 . Indeed, the N -turns coil with a current Ia is equivalent to a single coil with current N Ia ,
which radiates a power N 2 larger than the single turn one.
As an example, an antenna made of N = 100 turns with radius a = 1 cm at f = 300 MHz has
a radiation resistance Rrad = 30.7 , whereas the single coil one has only Rrad = 30.7 104 !
The maximum directivity of the small loop is the same as that of the elementary dipole, i.e.
D = 1.5.

4.3.2

Aperture antennas

Aperture antennas are completely different from wire antennas. They are still made of metal
parts, and we can consider the currents flowing on them as responsible for the radiated fields.
However, these antennas are characterized by the presence of apertures, as in the case of horns
and paraboloids, as shown in Fig. 4.16 and Fig. 4.17. Actually, a paraboloid is just a mirror and
typically is used in conjunction with a horn, placed in the focus, that acts as a feed, as sketched
in Fig.4.17. In the case of aperture antennas, one can estimate the fields on the aperture with

PRELIMINARY VERSION

Figure 4.16.

68

Examples of horn antennas, connected to rectangular and circular waveguides.

&

Figure 4.17. Front fed paraboloid. The aperture is the circle defined by the dish rim and the plane
is to be considered in contact with it.

reasonable accuracy, while the currents on the metal parts can be very difficult to guess. By means
of the equivalence theorem it is possible to represent the aperture fields in terms of equivalent
sources, which are responsible for the radiated fields.
Rectangular horn antenna
Horns are always fed by a waveguide of convenient cross section, generally rectangular or circular.
Waveguides are used invariably in single mode operation, so that their cross section has a size less
than one wavelength. For this reason a truncated waveguide is a very poor radiator:
the mode of propagation approaching the waveguide end suffers a fairly large reflection coefficient
the radiation pattern is very wide

PRELIMINARY VERSION

69

In practice a horn carries out a slow transformation of the waveguide cross section into a larger
one, so that both problems are solved at the same time. In the case of a pyramidal horn fed by a
rectangular waveguide the aperture electric and magnetic fields are assumed to be
E(x,y) = E0 cos

x
y

1
E0
x

z E(x,y) =
cos
x

Z0
Z0
A
A
B
B
A
and
y
x
2
2
2
2

H(x,y) =
for

(4.17)

where the aperture is rectangular with sides A and B, as shown in Fig. 4.18. This field distribution
is real because the phase error has been neglected for simplicity. This phase error takes into account
the fact that the wavefront propagating in the horn is spherical instead of plane, as assumed here.
The electric field distribution is that of the fundamental mode of the feed rectangular waveguide,
mapped smoothly onto the aperture. The magnetic field would be also that of the fundamental
mode if the wave impedance were substituted by the modal impedance. The use of Z0 is more
accurate for apertures with size greater than 2-3 . A plot of the normalized field distribution
is shown in Fig. 4.19 It can be shown that the generalized dipole moment of this horn has the

y
B

x
A
Figure 4.18.

Rectangular horn aperture, with sides A and B.

expression (valid only in the front half space z 0, i.e. 0 /2))

Pe (,) = 2Y0 E0 AB cos2

2
FH (x ) FE (y ) p

with
A
sin cos

B
y =
sin sin

x =

where the functions FE and FH were introduced in (4.15) and (4.16) and the polarization unit
vector p
is given by
+ cos
p
= sin

Fig. 4.20 helps in understanding the definition of p

. The rectangle is the horn aperture, seen
from a point on the z axis. The circle indicates a cone of directions with the same (assumed
to be very small) and different : in each point the local p
is shown. It appears to be parallel
to y
, but actually is tangent to the sphere through the observation point. figura diagramma horn
polarizzzione 3D bis This is more clearly indicated in Figs. 4.21,4.22 from two different views.

PRELIMINARY VERSION

70

1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.5
0.5
0
0

y/B

Figure 4.19.

0.5

0.5

x/A

Horn normalized aperture field distribution.

= const
Figure 4.20.

Polarization vector p
for small .

The plane x,z ( = 0,) is called H plane because it contains the H field; similarly, the plane
y,z ( = /2,3/2) is called E plane because it contains the E field.
An example of 3D field pattern for a horn with size A = 2, B = is shown in Figs. 4.23. For
better clarity a cartesian diagram is used, hence the z coordinate (vertical axis) shows |Pe (,)|.
The angles , define a unit vector
s and its x and y components are used as coordinates in the
horizontal plane: sx = sin cos and sy = sin sin . The polar coordinates in this plane are
s =

s2x + s2y = sin and

As a consequence, the limiting circle with unit radius in the horizontal plane corresponds to
= /2.
Fig. 4.24 shows the field pattern of the same horn in the standard polar diagram. The aperture
is also shown.
An example of a more directive horn, with A = 4 and B = 4 is shown in
Figs. 4.25, 4.26.

PRELIMINARY VERSION

Figure 4.21.

Polarization vector field p

drawn on the half sphere of directions.

Figure 4.22.

Polarization vector field p

drawn on the half sphere of directions.

71

Two dimensional plots are easier to use. Figs. 4.27, 4.28 show the plots of the functions FE (y ),
FH (x ) in dB scale. The corresponding plots in linear scale are shown in Figs. 4.29, 4.30.
The E plane pattern has a number of side-lobes that increases with B/: indeed, the maximum
value of y is B/. The first side-lobe is near y = 1.5 and its value is -13.3 dB.
As for the H plane, the maximum value of x is A/. The first side-lobe is near x = 2 and has
a value of -23 dB.
The reason why the H plane side-lobes are lower than the E plane ones is due to the fact that
the field distribution is tapered in the x direction but not in the y direction, see Fig. 4.19. As

PRELIMINARY VERSION

72

Rectangular horn: A/=2, B/=1

E plane =/2

H plane =0

0.8

0.6

0.4

0.2

0
1
1

0.5
0.5

0
0
0.5

0.5
1

sin sin

Figure 4.23.

1
sin cos

Rectangular horn field pattern, cartesian diagram, A = 2 and B = .

Rectangular horn: A/=2, B/=1

E plane
H plane

B/
A/

Figure 4.24.

Rectangular horn field pattern, standard polar diagram, A = 2 and B = .

a consequence, the beamwidth is larger in the H plane than in the E plane. If the main lobe
beamwidth (between the zeros) is to be the same in the E and H planes in order to have a
rotationally symmetrical main lobe, the condition B = (2/3)A must be enforced.

PRELIMINARY VERSION

73

Rectangular horn: A/=4, B/=4

E plane =/2

H plane =0

0.8

0.6

0.4

0.2

0
1
1

0.5
0.5

0
0
0.5

0.5
1

sin sin

Figure 4.25.

1
sin cos

Rectangular horn field pattern, cartesian diagram, A = 4 and B = 4.

Rectangular horn: A/=4, B/=4

E plane
H plane

B/
A/

Figure 4.26.

Rectangular horn field pattern, standard polar diagram, A = 4 and B = 4.

Finally, it can be shown that the aperture efficiency of horns is

a =

Aeq
= 0.8
Ag

PRELIMINARY VERSION

74

FE(y)
0

10

dB

15

20

25

30

35

40

Figure 4.27.

0.5

1.5

2
y/

2.5

3.5

Horn E plane field pattern: plot of the function FE (y ).

FH(x)

10

dB

15

20

25

30

35

40

Figure 4.28.

0.5

1.5

2
x/

2.5

3.5

Horn H plane field pattern: plot of the function FH (x ).

In this way, if the size A, B of the horn is known, the maximum gain can be easily computed from
Aeq = a AB

G=

4
Aeq
2

PRELIMINARY VERSION

75

F ( )
E

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.5

1.5

2.5

3.5

y/

Figure 4.29.

Horn E plane field pattern: plot of the function FE (y ).

F ( )
H x

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

Figure 4.30.

0.5

1.5

x/

2.5

3.5

Horn H plane field pattern: plot of the function FH (x ).

Also, losses are very small and gain and directivity coincide.

Chapter 5

Waveguides
Electromagnetic waves have the natural tendency to propagate in all directions away from the
point where they are generated. The fields created by a dipole in free space are a good example.
For some applications, it is desirable that waves propagate in only one direction, say z, while the
field is appreciable only in a neighborhood of it. This behavior can be obtained by generating the
electromagnetic wave inside a metal tube, called waveguide. The side walls have very high (if not
infinite) conductivity and act as mirrors that prevent propagation in the transverse direction. The
cross section of the tube is arbitrary, but generally it is rectangular, circular or in the form of a
ridge, as shown in Fig. 5.1. Sometimes the cross section is simply connected, sometimes doubly or
multiply connected: the coaxial cable belongs to this latter class.

Figure 5.1. Various waveguide types. From left to right: rectangular, cylindrical,
coaxial and ridge waveguide

Waveguides are not always in the form of hollow tubes. The microstrip is a different kind
of waveguide, which appears to be open, without sidewalls. the field confinement is due to the
dielectric inhomogeneity of the cross section. Optical fibers are another example of waveguides
without sidewalls. The confinement here is related to the total reflection at the interface between
two different dielectrics.
In these notes we focus on the hollow tube waveguides.
76

PRELIMINARY VERSION

5.1

77

Waveguide modes

Fig. 5.2 shows an example of rectangular waveguide containing a source, in the form of a small
dipole, connected to the inner conductor of a coaxial cable through a hole in the wide side of the
waveguide. The problem that we want to solve is to compute the fields radiated by this source
in the waveguide. For simplicity, we assume that the walls are perfect conductors and that the
waveguide is filled with a homogeneous dielectric with permittivity and permeability . The
mathematical formulation of the problem is the same as (2.1), that we repeat here for convenience

y
x

Figure 5.2. Waveguide with rectangular cross section. The field in the waveguide is radiated by a
kind of small dipole, connected to a coaxial cable through a hole in the wide side of the waveguide

E
H

Etg = 0

=
=

jH M
jE + J
on the metal

(5.1)

As we already did in Chapter 2, we solve first the problem without sources, in order to find the
possible fields that can exist in the waveguide. They must satisfy
Maxwells equations
the boundary conditions
that is

E
H

Etg = 0

= jH
= jE
on the metal

(5.2)

The problem is homogeneous and its solutions are called modes of propagation. When they are
known, we solve the radiation problem by computing the excitation coefficient of each of them due
to the dipole antenna.
The z axis is a symmetry axis of the system: the system is LSz I, i.e. Linear z-invariant. In
other words the coefficients of the equations do not depend on z, hence we expect the solutions

PRELIMINARY VERSION

78

to depend on z by means of exponentials that we denote by exp(jkz z). This means that the
solution of (5.2) can be written as
Et () = V (z)e()
Ez () = I(z)Zt ez ()
Ht () = I(z)h()
Hz () = V (z)Yt hz ()

(5.3)

where denotes the transverse plane coordinates: a generic point P has coordinates (, z). The
functions V (z), I(z) are
V (z) = V0+ ejkz z + V0 ejkz z
I(z) = Yt V0+ ejkz z Yt V0 ejkz z

(5.4)

We recognize that the z dependence of the field is indeed in the form of complex exponentials that
represent traveling waves. Moreover the form of these equations is the same as the ones describing
the electrical state of a transmission line. In this case, however, there is no real transmission line
but only an equivalent one: the z dependence of each propagation mode can be studied by means
of a modal transmission line with suitable propagation constant kz and characteristic (or modal)
impedance Zt , where the t subscript stands for transverse (to z). Actually, we notice that the
presence of the z symmetry axis makes it convenient to split the fields into transverse (Et , Ht )
and longitudinal components (Ez , Hz ):
E = Et + Ez
z
H = Ht + Hz
z

(5.5)

Both the transverse and the longitudinal fields are written in the form of products of normalized
functions of the transverse coordinates, called modal functions, and of functions of z (modal voltage
and current) that can be considered as coefficients.
It is to be remarked that the transverse components play a major role: indeed the power flow
budget is evaluated, as always, by means of the flux of the Poyntings vector S. Obviously, power
can flow only in the z direction, so that only the z component of S has to be computed: it is easy
to realize that only the transverse field components yield a contribution
P =

1
R
2

(Et Ht )
z d

A mode of propagation is characterized by the set of modal functions {e, ez , h, hz } and a

number kt called transverse wavenumber. The modal functions are not independent, however. For
this type of waveguide (PEC walls and transversally homogeneous dielectric) it turns out that the
modes of propagation belong two sets: T E modes, identified by a vanishing longitudinal electric
field, ez () 0 over the complete cross section and T M modes, with identically zero longitudinal
magnetic field hz () 0. Each family consists of an infinite number of elements, that we label by
an index i.
In particular, T M modes, denoted by a prime have the following characteristics
hiz () 0,
e i () =

eiz () =

t i ()
kti

kti
i ()
jkzi

PRELIMINARY VERSION

79

h i () =
z e i ()
kti
T E modes are denoted by a double prime and have the properties
eiz () 0,
h i () =

hiz () =

kti
i ()
jkzi

t i ()
kti

z
e i () = h i ()
kti
When no prime or double prime is used, it is understood that the statement is applicable to both
T M and T E modes.
We see that all the T M mode functions are derived from the couple {i (), kti } and all T E
ones are derived from the couple {i (), kti }, where the generating functions i (), i () are
proportional to the longitudinal field components. The T M generating functions are solutions of
the two-dimensional Dirichlet problem for the Helmholtz equation:
(2t + kti2 )i () = 0
i ( ) = 0

(5.6)

We recognize the functions i () to be eigenfunctions of the Dirichlet transverse laplacian with

eigenvalue kti2 .
Similarly, the T E generating functions are solutions of the two-dimensional Neumann problem
for the Helmholtz equation:
(2t + kti2 )i () = 0
i ()

=0

(5.7)

It is interesting to note that the generating functions i () can be interpreted as the oscillation
modes of a drum, i.e. of an elastic membrane with the rim fixed on rigid ring with the same shape
as the waveguide cross section. Analogously, the generating functions i () can be interpreted
as the oscillation modes of the surface of a liquid in a box with the same cross section as the
waveguide.
If the waveguide cross section is not simply connected, as in the case of the coaxial cable, it can
be shown that T M modes exist with zero transverse wavenumber. Such TM modes are special,
because beyond having hz () 0, they have also ez () 0, so that they are called T EM modes.
This means that a coaxial cable has a T EM mode and an infinite number of T M and T E modes.
When one describes a coaxial cable as a transmission line, only the T EM mode is studied.
Indeed, it is always stated that a coax can be modeled as a transmission line up to a certain
maximum frequency that depends on the size of the cross section. Indeed, for higher frequencies,
higher order modes are no longer cut-off and start propagating (see next section).
The transverse wavenumbers kti , kti are real and positive and form an unbounded sequence.

PRELIMINARY VERSION

80

It can be shown that the generating functions are orthogonal to each other. Defining the inner
product of two functions by
f ()g ()d

< f,g >=

(5.8)

the orthonormality conditions are

< i (),j () >= ij
and
< i (),j () >= ij
where ij is the Kronecker delta. As a consequence, also the vector modal functions are orthonormal
< ei ,ej >=< ei ,hj
z >=
z >=
< ei ,ej >=< ei ,hj
< ei ,ej >=< ei ,hj
z >=

ei hj
z d = ij
ei hj
z d = ij

ei hj
z d = 0

A direct consequence of these orthonormality properties is that the total power carried by a number
of modes is the sum of the individual powers carried by each mode:
P =
=
=

1
R
2
1
R
2
1
R
2

(Et Ht )
z d

V i ei +
i

V i ei

Ij hj d

Ij hj +
j

(Vi Ii + Vi Ii )
i

Notice that this property is absolutely not trivial and does not depend on the linearity of Maxwells
equations, since power depends quadratically on the fields. We can view this relation as a statement
of a Pythagoras theorem in a Hilbert space with an infinite number of dimensions.
In conclusion, propagation modes in a uniform waveguide are completely independent. If
many of them are simultaneously present in a waveguide, each one has a completely autonomous
evolution, also form the energy point of view. As soon as an obstacle or a discontinuity is introduced
in the waveguide, such as a screw, a post, an iris, or a change of cross section, new boundary
conditions must be satisfied on the obstacle and this causes mode coupling.

5.2

Equivalent transmission lines

In this section we focus on the evolution of modes along the waveguide: in other words, we study
the modal voltages and currents. It was said above that each propagation mode is characterized
by a number of modal functions and a constant kti , kti called transverse wavenumber. This
wavenumber determines the longitudinal evolution of the mode, since it can be shown that the
longitudinal propagation constant kzi is given by
kzi =

2
k 2 kti

(5.9)

PRELIMINARY VERSION

81

for both T M and T E modes, where k 2 = 2 is the wave number in the medium that fills the
waveguide. Moreover, the modal impedances can be shown to be
Zti =

kzi

for T M modes

Zti =

kzi

for T E modes

Depending on the frequency of operation, the evolution of modes can be totally different. Consider
a specific mode i. It is clear from (5.9) that at very high frequency kzi is real, whereas a very low
frequencies it is imaginary. In particular, we define the critical frequency (or cut-off frequency) of
each mode
kti
fci =
(5.10)

2
In terms of it, for both T M and T E modes
if f fci the mode is above cut-off : kzi and Zti are real
if f fci the mode is below cut-off : kzi and Zti are pure imaginary
At the cut-off frequency, kzi = 0 and Zti = 0 and Zti .
The mode with the smallest critical frequency is called fundamental mode.
Let fc0i be the critical frequency of mode i in an empty waveguide. As it is clear from the definition,
the presence of the dielectric makes the critical frequency smaller:
fc0i
fci =
r r

Figure 5.3.

Infinitely long waveguide with a source that excites a mode above cut-off

In order to clearly understand the characteristics of the two regimes, suppose that the waveguide
is infinitely long, as shown in Fig. 5.3, so that only the forward wave is excited by the source:
Vi (z) = Vi0+ ejkzi z
Ii (z) = Yti Vi0+ ejkzi z
for z 0

PRELIMINARY VERSION

82

1. Suppose f > fci . Let us compute the time evolution of this wave.
vi (z,t) = R{Vi (z)ejt } = |Vi0+ | cos(t kzi z + arg(Vi0+ ))
ii (z,t) = R{Ii (z)ejt } = Yti |Vi0+ | cos(t kzi z + arg(Vi0+ ))
Since kzi is real, the mode is propagating with phase velocity

c/ r r

=
vph =
=
2
kzi
kti
fci
k 1
1
k
f

where we have used (5.10) and the fact that

c
= =
=v
k

r r
Note that this last quantity is the phase velocity v of a plane wave in the medium that fills
the waveguide. Note also that if the waveguide is empty, r = 1 and the phase velocity of the
mode is greater than the speed of light c. However, the theory of relativity is not violated:
indeed, only geometrical points (wave nodes or wave crests) move at the velocity vph and no
mass or energy or information. See the analogous discussion concerning plane waves (2.12).
The previous formula can also be written in a slightly different for in terms of the critical
frequency of the mode in an empty waveguide:

c/ r r
c
vph =
=
2
2
fci
fc0i
1
r r
f
f
We remark that the mode phase velocity depends on frequency, hence propagation in a
waveguide is dispersive. When the field in the waveguide is not monochromatic, the various
frequency components move at different velocity and the signal suffers distortions. If however
the signal is narrow band, as in the case of an amplitude modulated pulse of much longer
duration than the period of the carrier, the distortion manifests itself in the fact that the
envelope appears to travel at a different velocity than the carrier: the former travels at the
group velocity (vg ), the latter at the phase velocity (vph ). The group velocity can be shown
to be given by
1
vg =
d
R{kzi }
d
Carrying out the derivative, we find
vg =

c
r r

fci
f

c
r r

fc0i
f

Note that since the envelope does not changes shape, the difference in velocities is conventionally not considered a real distortion.
It can also be shown that the energy of the mode travels at the group velocity. We see that
the group velocity is always smaller than c and this is very important for the requirements
of the theory of relativity.

PRELIMINARY VERSION

83

Finally, it is to be remarked that a simple relationship exists between phase and group
velocity:
c2
vg vph =
r r
The wavelength on the equivalent transmission line is called guided wavelength and is defined
as
2

gi =
=
2
kzi
fci
1
f
where
=

c/ r r
0
2
=
=
k
f
r r

is the plane wave wavelength in the dielectric that fills the waveguide and 0 the corresponding
one in empty space. Obviously gi > . Also this formula can be written in terms of fc0i :

gi =
1

fci
f

fc0i
f

r r

The modal impedance of T E modes is given by

Zti =
=
kzi

Z0

kti
k

=
1

It is useful to recall that

Z=

r
r
2

fci
f

= Z0

Z0 r

r r

fc0i
f

r
r

is the plane wave impedance in the dielectric that fills the waveguide.
The modal impedance of T M modes is given by

Zti =

kzi
=

kti
k

= Z0

r
r

fci
f

Z0
r

r r

fc0i
f

Since the characteristic impedance Zti is real, the wave is characterized by the power flow
P =

1 |Vi0+ |2
2 Zti

2. Suppose f < fci . In this case kzi = j|kzi | is imaginary. The negative sign of the propagation
constant is related, as always, to our time convention exp(j0 t) for phasors. The time
evolution of the modal voltage and current wave is
vi (z,t) = R{Vi (z)ejt } = |Vi0+ |e|kzi z | cos(t + arg(Vi0+ ))
ii (z,t) = R{Ii (z)ejt } = |Yti ||Vi0+ ||e|kzi z | cos(t + arg(Vi0+ ) 90 )

PRELIMINARY VERSION

84

Figure 5.4.

Infinitely long waveguide with a source that excites a mode below cut-off

Also the modal admittance is imaginary, but its sign depends on the polarization: negative
for T E modes and positive for T M . Fig. 5.4 shows clearly that the voltage and current
waves are evanescent. Since the phase of the wave does not depend on z, phase velocity,
group velocity and guided wavelength are not defined. The decay constant, i.e. the length
necessary for the amplitude to reach the level 1/e of the starting value, is
1
0.9
0.8
0.7

1/e

0.6
0.5
0.4
0.3
0.2
0.1
0
0

Figure 5.5.

Li
2

10

12

14

Decay constant of a mode below cut-off

Li =

1
|kzi |

and is shown in Fig. 5.5. We see also that voltage and current are in phase quadrature,
hence no active power flow is associated to a purely forward evanescent wave. The situation
is different when both forward and backward wave are present in the same region, as we will
explain later. Nevertheless, an evanescent wave plays an important role in the energy budget,
since it stores energy. Since the wave does not move, but remains attached to the source, it
does not give rise to any power flow. The wave oscillates in time and exchanges its energy
with the source twice per period.
If we look at the equations derived above, we see that the various modal parameters have two
different frequency behaviors. In particular
kzi
vg
Z
=
= ti =
k
v
Z
and

vph
Z
g
=
= ti =

v
Z

fci
f

fci
f

1
1

PRELIMINARY VERSION

85

The symbols , v, Z, denote the values of wavelength, phase velocity and wave impedance in the
dielectric that fills the waveguide. The two behaviors are plotted in Fig. 5.6 It is evident that the
2

1.5
g/=vph/v=Zt/Z
1
kz/k=vg/v=Zt/Z
0.5

0
0

f/fc

Figure 5.6. Dispersion curves of various parameters of a mode above cut-off. The two vertical dash
dotted lines indicate the standard operating band of a rectangular waveguide in the fundamental
mode. Note that the first higher order mode goes above cut-off at f /fc = 2.

waveguide is very dispersive close to cut-off. However, it is not possible to use it in the frequency
band where the curves are almost flat, since the guide is not single mode there. Actually, assuming
that the curves refer to the fundamental mode of a rectangular waveguide, the vertical lines are
the limits of the standard operating range. Indeed, the first higher order mode goes above cut-off
at f /fc = 2.
In general a source excites all the modes of a waveguide, the amount of excitation depending
on the geometry of the source. At a given frequency, a finite number of excited modes are above
cut-off and, by carrying energy away from the source, is responsible for the radiation phenomenon.
The value of the radiation resistance of the dipole is related to them. At the same time, an infinite
number of excited modes are below cut-off: their effect is important only in the neighborhood of
the source, where they describe the reactive field. Moreover, evanescent modes are responsible for
the imaginary part of the input impedance of the dipole.
A waveguide is said to be single mode if at the operation frequency only the fundamental mode
is above cut-off. At microwave frequency, waveguides are essentially always single mode, in order
to avoid interference effects between the various modes above cut-off.
The sources present in the waveguide can be represented in circuit terms by means of voltage
and current generators to be inserted on the modal equivalent lines, as shown in Fig. 5.7. If
the source is described by the distributions J(r), M(r), it can be shown that the corresponding
generators are
vi (z) =< Mt ,hi > + < Jz ,Zti ezi >
ii (z) =< Jt ,ei > + < Mz ,Yti hzi >

(5.11)

The inner products, defined in (5.8), are integrals over the waveguide cross section, hence, the
generators are distributed on the line since their strength is a function of z.

PRELIMINARY VERSION

5.3

86

Rectangular waveguide

The simplest waveguide to analyze is the one with rectangular cross section, see Fig. 5.8 The cross
section is simply connected, hence the modes of this waveguide are T M and T E but not T EM . As
for T M modes, the solution of (5.6), obtained by the classical method of separation of variables, is
n
2
m
sin
mn (x,y) = sin
a
b
ab
m 2
n 2
ktmn =
+
a
b

m, n = 1,2,3, . . .
(5.12)

This is the generating function of T M modes and is proportional to the longitudinal component of
the electric field ez . We see that the index i used in the previous section to label the mode functions
is actually a double index i (m, n). Fig. 5.9 and Fig. 5.10 show the plots of two generating
functions that help in understanding the meaning of the integer labels: m is the number of hills
and valleys along the x direction, whereas n refers to the y direction. The contour lines plotted
on the bottom plane serve two purposes: first, they are constant height lines and help in the
comprehension of the 3D plot, second they are the field lines of the transverse magnetic field hmn .
Indeed, according to the equations of Section 5.1, e i () is proportional to the transverse gradient
of (), hence it is orthogonal to the contours. However, h i () is orthogonal to e i (), hence it
is tangent to the contour lines. Note that the mode is T M , hence hzi () 0 and h i () is the
total magnetic field. It is no surprise that the contour lines are closed: this is a consequence of the
fact that the magnetic field is solenoidal. The field lines of e i () are not drawn, but can be easily
imagined as the orthogonal trajectories to the contour lines shown. See, however, Fig. 5.13.
Concerning T E modes, the solution of (5.7) is
mn (x,y) =
ktmn =
where

m n

ab
m
a

cos
2

m
n
cos
a
b

n
+
b

m, n = 0,1,2, . . .

but not m = n = 0
(5.13)

is the Neumann symbol :

0
m

=1
=2

for m = 0

Fig. 5.11 and Fig. 5.12 show the plots of the generating functions of two T E modes. Again, we
see clearly that the indices m, n count the hills and valleys in the x and y directions, respectively.

vi ( z )

k zi
ii ( z ) Z
ti

Figure 5.7. Distributed generators on the modal transmission line corresponding to mode i,
characterized by propagation constant kzi and characteristic impedance Zti .

PRELIMINARY VERSION

87

y
b

z
Figure 5.8.

Rectangular waveguide with sides a, b, a b.

1
0.5
0
0.5

3
2

1
1
0.5
y
Figure 5.9.

1
0

Rectangular waveguide (a = 3, b = 1): generating function 11 (x,y) of mode T M11

When one of the labels is zero, the function is constant in the corresponding direction, as in the
case of Fig. 5.11, referring to the fundamental mode. The contour lines are the field lines of e i ().
Indeed, h i () is proportional to the gradient of () and is then orthogonal to the level lines.
Moreover, e i () is orthogonal to h i (), hence tangent to the contour lines. The contours are
closed curves because the transverse electric field (that coincides with the total one, since the mode
is T E) is solenoidal: indeed no charges are present since modes are solutions in absence of sources.
The field lines of h i () are not shown here, but it is easy to imagine them, orthogonal to the level
lines, see also Fig. 5.14.

The mode functions ei () and hi () can be obtained by computing the derivatives of the
generating functions, according to the general formulas of the previous section. We limit ourselves
here to write the mode functions of the fundamental mode T E10 :

PRELIMINARY VERSION

88

1
0.5
0
0.5
1

1.5
1y
0.5

0 0
Figure 5.10.

Rectangular waveguide (a = 3, b = 1): generating function 32 (x,y) of mode T M32

1
0.5
0
0.5
3

1
1
y

2
x

0.5

1
0 0

Figure 5.11.

Rectangular waveguide (a = 3, b = 1): generating function 10 (x,y) of mode T E10

2
x
sin
y

ab
a
2
x
h 10 (x,y) =
sin
x

ab
a
j
2
x
hz10 (x,y) =
cos
kz10 a ab
a
e

10 (x,y)

(5.14)

PRELIMINARY VERSION

89

1
0.5
0
0.5
1

1.5
1
y0.5

1
0 0

Figure 5.12.

Rectangular waveguide (a = 3, b = 1): generating function 32 (x,y) of mode T E32

Notice that neither the mode functions nor the transverse wavenumbers depend on the dielectric
that fills the waveguide(provided the dielectric is homogeneous). Fig. 5.13 shows the field lines of
the first T M modes. Note that the in the cross section the electric field lines are orthogonal to
the magnetic ones. Moreover, on the lateral surface, the magnetic field lines are orthogonal to the
field lines of the induced current, according to (1.19). Fig. 5.14 shows, in a similar manner, plots
of the field lines of the first T E modes.
Concerning the transverse wavenumbers, we see that they are given by the same equation for
T M and T E modes: the difference is just in the allowed values of the indices. There is a useful
geometrical representation of transverse wavenumbers that allows to determine graphically the
modes that are below or above cut-off. Looking at the defining equations, one has the idea to
arrange the points representing modes in a regular lattice as shown in Fig. 5.15, which is based
on the use of Pythagoras theorem. The distance of the representative point of a given mode
from the origin is just its transverse wavenumber. The plane in which the points are drawn is
a spectral plane of spatial frequencies, i.e. wavenumbers, conjugate to x and conjugate to
y. Now, suppose that we want to find which modes are above cut-off at a given frequency. It is

enough to draw a circle with center in the origin and radius equal to k = : the modes whose
representative points are inside the circle have real kzi and then are above cutoff.
As explained in the previous section, below cut-off modes are characterized by evanescent fields
that remain attached to their sources and do not provide active power transfer (excluding tunnel
effects to be discussed later). Their importance is lower and lower as the magnitude of their
imaginary propagation constant kzi increases. Then it is useful to arrange modes in order of
increasing kti . Fig. 5.15 shows clearly that the points closest to the origin are those corresponding
to
T E10 , T E20 , T E01 if b < a/2

PRELIMINARY VERSION

90

Figure 5.13. Field lines of the first T M mode functions. The symbol Emn is equivalent to T Mmn .
1. is the cross-sectional view, 2. the longitudinal view and 3. the (opened) surface view. The
critical wavelengths are also indicated

T E10 , T E01 , T E20 if b > a/2

Only T E modes need be considered since T M points are not on the coordinate axes. The first mode

PRELIMINARY VERSION

91

Figure 5.14. Field lines of the first T E mode functions. The symbol Hmn is equivalent to T Emn .
1. is the cross-sectional view, 2. the longitudinal view and 3. the (opened) surface view. The
critical wavelengths are also indicated

is T E10 that is then the fundamental mode of the rectangular waveguide. Different waveguides have
different fundamental mode: T E11 for the circular waveguide and T EM for the coaxial cable.
The first higher order mode is different according to the aspect ratio of the cross section. The

PRELIMINARY VERSION

92

critical frequencies of these modes in an empty waveguide, computed by (5.10) are

fc10 =

c
c
c
, fc20 = , fc01 =
2a
a
2b

It is clear that the bandwidth over which the guide is single mode is maximum if the mode sequence
is T E10 , T E20 , T E01 ; in this case it equals an octave. If b > a/2 the single mode bandwidth reduces,
down to zero in the case of square waveguides: in this extreme case, the T E10 and T E01 modes
are degenerate, i.e. they have the same critical frequency. Mode degeneracy is a very common
phenomenon in rectangular waveguides, due to their high degree of symmetry.
It was said before that waveguides for microwave applications are almost invariably single mode.
If, on the contrary, the sides are very large with respect to , counting the modes above cut-off
can be not straightforward. However, it is not difficult to estimate this number Nabove . Indeed,
the area of the quarter of circle with radius k is k 2 /4; the density of points is 2 (T E and T M )
for each rectangle with sides /a /b, ignoring that T M modes have labels different from zero.
Then
k 2 2
k 2 ab
(ka)(kb)
ab
Nabove = 2
/
=2
=
= 2
2
4 ab
4
2

5.3.1

Design of a single mode rectangular waveguide

A typical design problem is that of finding the dimensions of a rectangular waveguide so that it
is single mode over the bandwidth [fmin ,fmax ], with fmax 2fmin , of course. The specifications
are generally completed by asking that the first higher order mode has an attenuation greater
than a minimum value lim over the complete operative band and that the waveguide can carry a
maximum power Plim without dielectric breakdown.
The operating conditions are indicated in Fig. 5.16:

2
b

( 3,1)

kt 31
k
(1,0 )

2
a

3
a

4
a

Figure 5.15. Mode scheme of the rectangular waveguide. The transverse wavenumber
of modes T M32 and T E32 is displayed. At the indicate frequency, only the T E10 mode
(fundamental mode) is above cut-off

PRELIMINARY VERSION

93

f min

f c10

f max

f c 20

Figure 5.16. Operating conditions for a single mode waveguide. The grey region between fmin
and fmax is the required single-mode band of operation. The square parentheses denote the
available single-mode band of the waveguide. If a is modified, the position of the parentheses
changes, with the constant ratio fc20 = 2fc10

c
2a
c
=
a

fmin fc10 =
fmax fc20

from which we find the range of values of a that satisfy the specification
c
2fmin

(5.15)

fmax

If a is chosen close to amin , the single-mode band of the waveguide, denoted by the square parentheses in the figure, shifts to the right. In these conditions, the mode T E20 has the maximum
attenuation over [fmin ,fmax ], but in the first part of this band the propagation is highly dispersive, as Fig. 5.6 clearly indicates.
If, on the contrary, a is chosen close to amax , the single-mode band of the waveguide shifts to the
left. In these conditions, the dispersion effects are minimized, but the attenuation of mode T E20 is
insufficient, in particular for frequencies close to fmax . Actually, there is a specification concerning
this point. Compute the attenuation of mode T E20 :
20 =

fc20
f

2 k2 = k
kt20

It is minimum at fmax
20min =

2fmax
c

fc20
fmax

By requiring 20min lim , we find a constraint on a (contained in fc20 )

c

a
fmax

clim
2fmax

(5.16)

+1

The third specification, that dispersion effects are minimized, forces the sign to be substituted
by an equal in the previous equation. Assuming that the problem proposed has a solution, i.e.
that this value lies in the range (5.15), the width of the waveguide has been found.
The fourth specification, concerning the power at the breakdown limit, affects only the height b.
The active power flow on the T E10 modal line is
P =

1 |Vmax |2 1
1 |V + |2
=
2 Zt10
2 Zt10 S

PRELIMINARY VERSION

94

assuming that the modal line is mismatched, the VSWR has the value S, and Vmax is the maximum
voltage along the line.
By recalling (5.14), the maximum value of the electric field is reached at x = a/2, so that
Emax =

2
Vmax
ab

Substituting in the previous expression, we get

P =

2
1 ab Emax
1
2 2 Zt10 S

(5.17)

This is the power in the waveguide for a given value of Emax . Obviously, S = 1 if the line is
matched. If we solve the previous equation with respect to Emax for a given power Plim , we see
that the maximum field is proportional to the modal impedance. Hence, recalling the plot of
Fig. 5.6, we conclude that the condition Emax R (R denoting the dielectric rigidity) has to be
enforced at f = fmin , where the impedance is maximum. The value of a to be used in the previous
equation is that found by (5.16). Moreover, it is to be remarked that the condition on the power
at the breakdown limit must be consistent with b a/2, on which the determination of a is based.
Alternatively, we can compute the power at the discharge limit Pdisch by substituting in (5.17)
Emax with the dielectric rigidity of air R = 20 kV/cm. Then Pdisch must be enforced to be greater
than the desired power Plim . Pdisch depends on frequency through Zt10 : the minimum value of
Pdisch is that at f = fmin , where Zt10 is maximum. Hence the condition to be guaranteed is
Pdisch (fmin ) Plim
In order to make an example with numerical values, we can consider the WR90 standard rectangular
waveguide for X band, whose data are
a = 0.9 inches=2.286 cm; b = 0.4 inches=1.016 cm
fmin = 8.0 GHz, fmax = 12.4 GHz
lim = 7.8 dB/cm
Plim = 201.6 kW, in matched conditions, S = 1.
We can add that the attenuation of the T E20 mode is = 18.9 dB/cm at f = fmin .

5.3.2

Tunneling effects

It was shown in section 5.2 that the modes below cut-off do not carry active power. Actually,
this is not true if in a waveguide region both a forward and a backward below cut-off mode exist.
This phenomenon is called tunneling because of the analogy with the tunnel effect in quantum
mechanics.
Let us consider a rectangular waveguide filled completely with dielectric apart from the region

AB, as shown in Fig.5.17(a). Let fc0 and fcd = fc0 / r be the critical frequencies of the T E10

PRELIMINARY VERSION

(a)

95

TE10

( b)

A+

Z0
kz 0

Z d k zd

A+

B+

Z d k zd

B+

Figure 5.17. Tunnel effect in an inhomogeneously filled rectangular waveguide. The mode T E10 is
above cut-off in the dielectric and below cut-off in the air region. (a) Actual structure. (b) Modal
equivalent transmission line circuit

mode in the empty and in the filled waveguide, respectively. Assume that the the frequency f of
the incident mode is
fcd < f < fc0
so that the mode is above cut-off in the filled waveguide and below in the empty one. The problem
is to compute the field everywhere. First of all, the point to address what happens at the interface
in A. Two are the alternatives: either the interface gives rise to mode coupling or not. The
conditions to be enforced, according to (1.17), are the continuity of the tangential components of
electric and magnetic field across the interface. Since the interface is a plane perpendicular to
z,
the tangential components coincide with the transverse ones. Hence, the condition that must hold
is
Et (,zA ) = Et (,zA+ )
Ht (,zA ) = Ht (,zA+ )
The field in a waveguide is a superposition of modes of propagation of the type (5.3), then
Vi (zA )ei () =
i

Vi (zA+ )ei ()
i

Ii (zA )hi () =
i

Ii (zA+ )hi ()
i

Eqs.(5.12), (5.13) show that the mode functions do not depend on the material that possibly fills
the waveguide, then the previous equations lead to
Vi (zA ) = Vi (zA+ )
Ii (zA ) = Ii (zA+ ) i

PRELIMINARY VERSION

96

This equation shows clearly that the dielectric interface does not produce mode coupling: each
mode can be analyzed independently from the others. Notice that if the interface is not planar
or not orthogonal
z, it produces mode coupling. This means that even if only the T E10 mode is
incident on it, all the other modes (in general) are excited, with the right amplitudes, so that the
appropriate continuity conditions are satisfied. In this case, since only the T E10 mode is incident,
no other mode is generated and the modal equivalent circuit is the one shown in the lower part of
Fig. 5.17.
The parameters of the modal lines are
kz0 = k0

fc0
f

kzd = k0

Z0

Z0 =
1

fc0
f

fc0
f

Z0

Zd =

fc0
f

Notice that the mode is below cut-off in the empty waveguide, hence
kz0 = j

Z0 = jX0

with 0, as required by the usual convention, and X0 0 because Zt = /kz . If the

incident electric field is known, we can compute the incident voltage in A, VA+ . Next we compute
the reflection coefficient in A . Start with
ZB +
Zd
= j
Z0
X0
z 1
= B
with |B | = 1
zB + 1
= B exp(2jkz0 LAB ) = B exp(2LAB )

zB =
B
A+

The trajectory on the Smith chart of the reflection coefficient when the observation point moves on
a transmission line with imaginary propagation constant is a radial line instead of a circumference.
Hence A+ < 1. Next
1 + A+
zA+ =
= rA+ jxA+
1 A+
with rA+ > 0 and xA+ > 0 strictly. Compute zA
zA = zA+

Z0
X0
X0
X0
= (rA+ jxA+ )j
= xA+
+ jrA+
Zd
Zd
Zd
Zd

Note that the real part of zA (as well as the imaginary part) is strictly positive, hence the point
A is strictly inside the unit circle, A < 1. As a consequence, the active power flowing beyond
the point A is
1 |VA+ |2
PA =
1 |A |2 > 0
2 Zd
Now, this means that there is an active power flow through the part AB of the waveguide, although
the T E10 mode is below cut-off. The paradox is easily solved if one takes into account that in the
part AB both the forward and the backward waves exist (because of the discontinuity in B). The

PRELIMINARY VERSION

97

formula to be used for the computation of the active power on a line with imaginary characteristic
impedance Z0 = jX0 can be shown to be
P (z) =

1
1
R{V (z)I (z)} =
Im{V + (z)V (z)}
2
X0

(5.18)

We see that although a single evanescent wave does not give rise to any active power flow, the
presence of a forward and a backward one does: one can say that the two evanescent waves
cooperate in order to produce the power flow.
It is interesting to examine what happens when the thickness LAB , so that the structure
contains only the interface in A. From the previous equations we see easily that, in this limit, it
is A+ = 0, so that rA+ = 1, xA+ = 0 and rA = 0 with the consequence that |A | = 1 and
PA = 0. This agrees with (5.18): if the interface in B is absent, the backward wave is identically
zero.
We can see numerical data about this phenomenon in Fig. 5.18, which shows an explanation
of the tunnel effect in terms of phasor diagrams. The waveguide is the standard WR90 guide,
0.4

V+

0.3

V
0.2

0.1

0.1

V
0.2

0.3

0.4

I+

0.5

0.2

0.2

I
0.4

0.6

0.8

Figure 5.18. Phasor diagram illustration of tunnel effect at the frequency f = 5 GHz.
The vectors represent, in the complex plane, the phasors of voltage and current in a point
halfway between A and B. Note that the phase difference between V and I is less than
/2, implying an active power flow.

with a = 0.9 in = 2.286 cm and b = 0.4 in = 1.016 cm. The dielectric has a relative permittivity
r = 4, the thickness of the empty guide is LAB = 2 cm. The critical frequencies of the T E10 mode
are fc0 = 6.56 GHz and fcd = 3.28 GHz and the frequency of the incident wave is f = 5 GHz,
so that the mode is below cut-off in the empty guide. Voltages and currents are evaluated in
a point M halfway between A and B. We see that the forward voltage is in phase quadrature
with the forward current, consistently with the fact that the characteristic impedance is positive
imaginary: indeed, the voltage is leading. Also, the backward voltage is in phase quadrature with
the backward current, but the current leads now. As a result, the total voltage and current are
not in phase quadrature, implying an active power flow. Since the phase difference between V and
I is less than /2, the power flow is positive. Obviously, this phase difference is the phase of the

PRELIMINARY VERSION

98

0.2

V+

0.15
0.1
0.05

0
0.05

0.1
0.15
0.2

I+

0.25
0.3

0.1 0.05

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

Figure 5.19. Phasor diagram illustration of tunnel effect at the frequency f = 5 GHz.
The vectors represent, in the complex plane, the phasors of voltage and current in B .
Note that VB and IB are in phase.

+
local impedance in M and the phase difference between VM
and VM
is the phase of the reflection
coefficient M .
Fig. 5.19 is a similar plot that refers to the point B . We notice in this case that the total V and

S11
1

0.8

0.6

0.4

0.2

0
2

fcd 4

6 fc0 8
10
frequency (GHz)

12

14

16

Figure 5.20. Tunnel effect in an inhomogeneously filled rectangular waveguide. The vertical dashed
lines denote the critical frequencies in the empty and in the filled waveguide. Plot of the magnitude
of the scattering parameter S11 (f ) vs. frequency

I are even in phase. The reason is that they coincide with VB + and IB + , which are certainly in

PRELIMINARY VERSION

99
S21

0.8

0.6

0.4

0.2

0
2

fcd 4

6 fc0 8
10
frequency (GHz)

12

14

16

Figure 5.21. Tunnel effect in an inhomogeneously filled rectangular waveguide. The vertical dashed
lines denote the critical frequencies in the empty and in the filled waveguide. Plot of the magnitude
of the scattering parameter S21 (f ) vs. frequency

phase because the wave there is purely forward and above cut-off. See also Fig. 5.22 and Fig. 5.23.
The frequency response of the structure is illustrated in Fig. 5.20, and Fig. 5.21, which show
plots of the scattering parameters S11 and S21 versus frequency at the reference planes A and
B + . We note first of all that the two curves are complementary: since the structure is lossless, the
scattering matrix is unitary, so that
|S11 (f )|2 + |S21 (f )|2 = 1
As long as fcd < f < fc0 , the T E10 mode is below cut-off in the region AB and the reflection
coefficient is large but not of unit magnitude. Correspondingly, the transmission coefficient S21
is not zero, indicating that a power flow across the region AB exists. Next, as the frequency
is increased, the T E10 mode is everywhere above cut-off. The transmission coefficient increases
steadily up to unit magnitude at a frequency of about 10 GHz, while the reflection coefficient is
zero at the same frequency. We can check that at this frequency the thickness LAB is exactly half
of the guided wavelength. This means that A+ = B , a complete turn on the Smith chart, and
ZA = ZB + . The consequence is obviously a zero reflection coefficient.
When the frequency is still increased, the reflection coefficient increases, again reaching a maximum
at the frequency for which LAB = (3/4)g0 . This behavior continues forever, with S11 = 0 at the
frequencies for which LAB = ng0 /2 for any integer n and local maxima at the frequencies for
which LAB = (2n + 1)g0 /4. The local maxima are different but tend to have the same height for
large frequency. This is due to the fact that the modal impedances are functions of frequency but
for large frequency, f
fc10
Z0
Z0 Z0
Z0
r
It is useful to examine the plots of the T E10 modal voltage and current. Fig. 5.22 and Fig. 5.23
refer to a frequency f = 5 GHz, at which the mode is below cut-off in the empty waveguide.

PRELIMINARY VERSION

100

Both voltage and current are evanescent between A and B: even if it is not so evident, note that
the curve is a combination of a decaying (the forward wave) and a growing (the backward wave)
exponential. The VSWR to the left of A is very large, because of the high reflection coefficient.
This causes the small minima. The plot is flat to the right of B because the transmission line is
matched and only the forward wave exists there.
If the frequency increases beyond fc0 , the T E10 mode goes above cut-off also in the empty region.
Fig. 5.24 and Fig. 5.25 show the voltage and current plots for f = 9 GHz. The reduction in the
value of S11 shows up in the reduction of the VSWR to the left of A and in the increase of the
transmitted wave to the right of B. It is also evident that the mode is above cut-off in the AB
region since the curve is oscillatory there.
Fig. 5.26 and Fig. 5.27 show the voltage and current plots for f = fr = 9.965 GHz, the frequency
for which the length AB is exactly half of the guided wavelength g0 . The reflection coefficient in
A is zero, so that the VSWR is zero to the left of A and the plot is flat. Moreover, the magnitude
of the transmission coefficient is one. The empty waveguide behaves as a resonator with resonance
frequency fr , i.e. as band-pass filter, as it is evident from the |S21 | plot in Fig.5.21
The phenomenon that we have described is called tunnel effect because it is similar (and is
described by the same mathematics) to the one of quantum mechanics concerning the transmission
of electrons through potential barriers.
|V(z)|
2

1.5

0.5

0
0.04

0.02

0
0.02
A z (m) B

0.04

0.06

Figure 5.22. Tunnel effect in an inhomogeneously filled rectangular waveguide. The vertical dashed
lines denote the interfaces in A and B between the empty and the filled waveguide. Plot of the
magnitude of the T E10 modal voltage vs. z at the frequency f = 5 GHz.

5.3.3

Irises and waveguide discontinuities

Modes in a perfect, uniform waveguide are independent: each of them can exist alone in the
waveguide since it satisfies Maxwells equations and all boundary conditions. If an obstacle is
introduced, or a change in the cross section, new boundary conditions arise, which were not taken
into account in the definition of the modes. In general then, all waveguide modes are excited

PRELIMINARY VERSION

101
Z d|I(z)|

1.5

0.5

0
0.04

0.02

0
0.02
A z (m) B

0.04

0.06

Figure 5.23. Tunnel effect in an inhomogeneously filled rectangular waveguide. The vertical dashed
lines denote the interfaces in A and B between the empty and the filled waveguide. Plot of the
magnitude of the T E10 modal current vs. z at the frequency f = 5 GHz. The current is multiplied
by the modal impedance for viewing convenience
|V(z)|
2.5

1.5

0.5

0
0.02

0.01

0
A

0.01
z (m)

0.02
B

0.03

0.04

Figure 5.24. Tunnel effect in an inhomogeneously filled rectangular waveguide. The vertical dashed
lines denote the interfaces in A and B between the empty and the filled waveguide. Plot of the
magnitude of the T E10 modal voltage vs. z at the frequency f = 9 GHz

by the discontinuity with suitable coefficients, such that the total electric and magnetic field,
consisting of incident and scattered waves for all the modes, satisfy the new boundary conditions.
As a example, consider an iris in a rectangular waveguide, as shown in Fig. 5.28. Irises are metal
windows, perpendicular to the guide axis. The thickness of irises is often small and can be neglected

PRELIMINARY VERSION

102
Z d|I(z)|

2.5

1.5

0.5

0
0.02

0.01

0
A

0.01
z (m)

0.02
B

0.03

0.04

Figure 5.25. Tunnel effect in an inhomogeneously filled rectangular waveguide. The vertical dashed
lines denote the interfaces in A and B between the empty and the filled waveguide. Plot of the
magnitude of the T E10 modal current vs. z at the frequency f = 9 GHz. The current is multiplied
by the modal impedance for viewing convenience
|V(z)|
3
2.5
2
1.5
1
0.5
0
0.02

0.01

0
A

0.01
z (m)

0.02
B

0.03

0.04

Figure 5.26. Tunnel effect in an inhomogeneously filled rectangular waveguide. The vertical dashed
lines denote the interfaces in A and B between the empty and the filled waveguide. Plot of the
magnitude of the T E10 modal voltage vs. z at the frequency for which S11 = 0.

as a first approximation. Suppose that the fundamental mode T E10 is incident from the left. The
boundary conditions to be satisfied on the iris are that the tangential electric field is zero on the
metal parts of the iris and different from zero on the aperture. In this case the tangential field
is the transverse field. If we recall the electric field distribution of T E10 , shown in Fig. 5.14, we

PRELIMINARY VERSION

103
Z d|I(z)|

2.5

1.5

0.5

0
0.02

0.01

0
A

0.01
z (m)

0.02
B

0.03

0.04

Figure 5.27. Tunnel effect in an inhomogeneously filled rectangular waveguide. The vertical dashed
lines denote the interfaces in A and B between the empty and the filled waveguide. Plot of the
magnitude of the T E10 modal current vs. z at the frequency for which S11 = 0. The current is
multiplied by the modal impedance for viewing convenience

A
kz 0 , Z 0

jB

TE10

A
Figure 5.28. Example of an iris in a rectangular waveguide. Upper left: longitudinal view. Upper right: cross section view. Lower left: equivalent circuit for the
fundamental mode (B > 0, capacitive iris).

realize that it is impossible to satisfy these boundary conditions with just the incident reflected
and transmitted waves of this mode. Indeed, if the total Ey field must be zero on the the metal,
certainly Eyr = Eyi , that is the reflection coefficient must be = 1. But the modal fields do not
depend on y (the second subscript is n = 0), hence the total Ey field is zero also on the aperture
and the transmitted field should be zero. This would imply that the iris behaves as a short circuit
plate and this is contrary to experimental evidence: indeed, although the reflection coefficient can
be high when the aperture is small, power is always transmitted through the hole. In conclusion,
the behavior is more complicated and an infinite number of higher order modes are excited in order
to satisfy the boundary conditions. In this way the reflected wave of T E10 need not be opposite
to the incident wave and the reflection coefficient is smaller than one.

PRELIMINARY VERSION

104

If the iris thickness is negligible, its fundamental mode equivalent circuit is an admittance connected in parallel on the T E10 modal line. Indeed, the transverse electric field must be continuous
on the complete cross section: equal to zero on the metal and different from zero (but continuous)
on the aperture. Hence the modal voltage is continuous, V (A ) = V (A+ ), and the equivalent circuit is a load in parallel. Note that this load takes into account only the effect the iris has on higher
order modes. Indeed, this load is connected on the T E10 modal line and the iris effects on the
propagation of this mode are taken into account directly by transmission line theory. Generally the
waveguide is operated in single mode conditions, so that all higher order modes are below cut-off.
As well known, they do not contribute active power propagation but store electromagnetic energy.
The admittance that represents the iris on the T E10 modal line must describe this behavior, hence
it is a pure susceptance jB.
As an example of this procedure, we compute the scattering matrix of the iris for the fundamental
mode, assuming that the susceptance B is known. Assume that the reference planes are in A
and in A+ and the reference impedances at both ports coincide with the modal impedance. Let
b = BZ0 be the normalized susceptance (not to be confused with the small side of the waveguide!!). To compute S11 and S21 , close port 2 on the reference impedance and compute reflection
and transmission coefficients. They are
YA = jB + Y0
yA = jb + 1
S11 = A =

yA 1
jb
=
yA + 1
2 + jb

S21 = 1 + A = 1 +

jb
2
=
2 + jb
2 + jb

Then S12 = S21 because of reciprocity and S22 = S11 because of the symmetry of the structure.
The problem is the determination of the susceptance, which is a difficult one and requires the solution of an integral equation. Fig. 5.29 shows plots of the normalized susceptance of a symmetrical
iris. It can be shown that when the edges are perpendicular to the electric field lines, B > 0 and
the iris is said to be capacitive. Two sets of curves are shown for better reading accuracy: the
right set uses the vertical axis on the left and is to be used for large apertures (d/b 1), the left
one uses the axis on the right that measures essentially the inverse of the normalized susceptance.
Note that when the aperture is small, it resembles a short circuit, so the susceptance is large. If
the edges of the iris are parallel to the electric field lines, the iris is said to be inductive, because
its susceptance is negative, B < 0. Fig. 5.30 shows the normalized reactance of a symmetrical
inductive iris. Again, if the aperture is small, the reactance is small, i.e. the susceptance is large
as that of a short circuit.
It is interesting to compare the susceptance values of a capacitive and an inductive iris with the
same aperture: the former is much smaller than the latter. As a numerical example, consider a
WR90 (a = 2.286 cm, b = 1.016 cm) waveguide at f = 12 GHz. An inductive iris with an aperture
of d = 0.3 cm has a normalized susceptance B/Y 32, whereas a capacitive iris with the same
aperture has a susceptance B/Y 1.16.
Alternatively, consider a capacitive iris with a very small aperture, d = 0.5 mm, which has a susceptance B/Y = 4.3. An inductive iris with B/Y = 4.3 has a much wider aperture, d = 0.68 cm.
Such a susceptance produces
|S11 | = 0.9067

|S21 | = 0.4217

PRELIMINARY VERSION

105

Figure 5.29. Normalized susceptance of a symmetrical capacitive iris.

wavelength and Y0 is the modal admittance.

g is the guided

A possible explanation of this phenomenon is related to the induced currents on the iris plates,
which are parallel to the electric field since they are perpendicular to H (see (1.19)), which in
turn is perpendicular to E. These induced currents radiate the scattered field but are disturbed

PRELIMINARY VERSION

106

Figure 5.30. Normalized reactance of a symmetrical inductive iris. g is the guided wavelength and Z0 is the modal impedance. The curves are parametrized by a/ where = c/f
is the free space wavelength.

(interrupted) by the aperture in the case of capacitive iris so that their effect is small. In the case
of inductive iris they are not interrupted and are then very efficient.

PRELIMINARY VERSION

107

Irises are reactive circuit elements and can be used to build matching devices: in practice, they
substitute stubs in waveguide technology. Moreover they can be used to build filters. A simple
example consists of two irises separated by a distance of about g /2. They form a cavity, coupled
to the waveguide by means of the apertures: For the center frequency the structure is transparent,
otherwise it is highly reflecting, so that it behaves as a bandpass filter. Cavities are the equivalent
of LC resonators in waveguide technology.

Appendix A

Mathematical Basics
A.1

Coordinate systems and algebra of vector fields

The position of a point in space is specified by means of a coordinate system, that is a one to
one correspondence between points in space and triples of real numbers. The most common are
cartesian, cylindrical and spherical ones. It is to be remarked that all coordinate systems are
equivalent: a specific one is chosen according to the geometry of the problem at hand. If the
domain has a symmetry, it is convenient that the coordinate system has the same symmetry. For
example, to describe the position of points on the surface of the Earth, a spherical system is more
convenient than a cartesian one. To describe the magnetic field created by a straight wire carrying
an electric current, a cylindrical system is better suited than a cartesian one. To describe the
position of objects in a room, a cartesian system is more appropriate than a spherical one. Fig.
A.1 shows the main characteristics of the cylindrical reference system. Fig. A.2 describes the
spherical reference system.
The relationship between cylindrical (, , z) and cartesian coordinates is
x = cos
y = sin

(A.1)

and that between spherical (r, , ) and cartesian coordinates is

x = r sin cos
y = r sin sin
z = r cos

(A.2)

Every coordinate system has a set of three fundamental unit vectors, that will be denoted by
carets. The expressions of the cylindrical unit vectors in the cartesian basis are
= cos
x + sin
y

= sin
x + cos
y

(A.3)

Clearly the unit vectors are different in each point. The origin is a singular point of the system
of coordinates, since the unit vectors are not defined there. The corresponding expressions for the
1

PRELIMINARY VERSION

Figure A.1. Cylindrical coordinate reference system. (a) The three mutually perpendicular surfaces defining the position of a point. (b) The three unit vectors in the point P ; in the text ,
,
z
are used in place of a , a , az . (c) The elementary volume dV at P : the sides are d, d, dz.

PRELIMINARY VERSION

Figure A.2. Spherical coordinate reference system. (a) Definition of the spherical coordinates.
(b) The three mutually perpendicular surfaces defining the position of a point. (c) The three unit

vectors in the point P ; in the text
r, ,
are used in place of ar , a , a . (d ) The elementary
volume dV at P : the sides are dr, rd, r sin d.

PRELIMINARY VERSION

spherical system are

r = sin cos
x + sin sin
y + cos
z
= cos cos

x + cos sin
y sin
z

(A.4)

= sin
x + cos
y
and again in each point the basis of the unit vectors is different. As in the previous case, the
origin is a singular point. The cartesian system is different in this respect: indeed, the basis of
unit vectors is the same in each point, x
,
y,
z. Moreover, no point is singular.
Suppose that a certain vector field E(r) is given in a certain region of space. This means that
in every point r the vector E(r) is defined: we can imagine it as an arrow with the tail in the point
r. This is the abstract picture (coordinate-free). According to the system of coordinates, in the
point r a basis of unit vectors is defined and three numbers are associated to the vector, i.e. its
components:
+ E
E(r) = Ex x
+ Ey y
+ Ez
z = E + E
+ Ez z = Er r + E

(A.5)

Components can also be arranged in a 3 1 matrix (column vector) so that the previous equations
can also be written as

Ex
E
Er
E(r) Ey
E(r) E
E(r) E
(A.6)
Ez
Ez
E
Eq.(A.5) can be conveniently written as

Ex
y

z Ey =

z
E(r) = x
Ez

E
E =
Ez

Er
E
E

(A.7)

Notice that the row vector contains the unit vectors instead of numbers.
Also eq.(A.3) can be written in matrix form

y

z

cos
sin
0

sin
cos
0

0
0
1

(A.8)

As for the spherical system

y

z

sin cos
sin sin
cos

cos cos
cos sin
sin

sin
cos
0

(A.9)

In this way it is a simple matter to find the relationship between cylindrical and cartesian components of a vector. From (A.7) and (A.8)

E
Ex

z E
y

z Ey =
E(r) = x
Ez
Ez

(A.10)
cos sin 0
E
y

z sin cos 0 E
= x
0
0
1
Ez

PRELIMINARY VERSION

from which we get, on comparing the second and the fourth term of this chain of equalities

E
Ex
Ey = P
E
CaCy
Ez
Ez
where the change of basis matrix is

P CaCy
The inverse transformation is

cos
= sin
0

sin
cos
0

0
0
1

E
Ex
E = P
Ey
CyCa
Ez
Ez

where

P CyCa = P 1
= P TCaCy
CaCy

sin 0
cos 0
0
1

cos
= sin
0

Since both the cartesian and the cylindrical basis are orthonormal, the change of basis matrix is
unitary, hence the inverse of it coincides with its transpose.
The transformation of the components of a vector between the spherical and cartesian bases is
obtained analogously. From (A.7) and (A.9)

Ex
Er

y

z Ey =
E(r) = x
r
E
Ez
E

(A.11)
sin cos cos cos sin
Er
y

z sin sin cos sin cos E
= x
cos
sin
0
E
from which we get, on comparing the second and the fourth term of this chain of equalities

Ex
Er
Ey = P
E
CaS
Ez
E
where the change of basis matrix is

P CaS
The inverse transformation is

sin cos cos cos

= sin sin cos sin
cos
sin

sin
cos
0

Er
Ex
E = P
Ey
SCa
E
Ez

where

P SCa = P 1
= P TCaS
CaS

sin cos
= cos cos
sin

sin sin
cos sin
cos

cos
sin
0

(A.12)

PRELIMINARY VERSION

Vectors can be multiplied by a scalar, i.e. a number: the result is a vector with the same direction.
Two vectors can be multiplied in two ways. The scalar product of two arbitrary vectors is a scalar,
the vector product is a vector. In the former case
A B = |A||B| cos
= Ax Bx + Ay By + Az Bz = A B + A B + Az Bz = Ar Br + A B + A B

(A.13)

where is the angle between the two vectors. The scalar product of orthogonal vectors is obviously
zero. The first definition is coordinate-free. The other expressions are so simple because the bases
are orthonormal. In matrix form

Bx
A B = Ax Ay Az By
Bz
Indeed, recalling (A.7)

AB=

Ax

Ax

Ay

Ay

Az

Az

y
x
y

z

x
x
x
y
x

z
y
x
y
y
y

zx

zy

z
z

Bx
By
Bz

Bx
By
Bz

(A.14)

and the central 3 3 matrix is the identity matrix since the cartesian basis is orthonormal.
Two vectors can also be multiplied so that the result is a vector A B (or, more appropriately,
A B) :
|A B| = |A||B| sin
(A.15)
and the direction is orthogonal to the plane on which the two vectors lie and points in the direction
of the right-hand-rule. The vector product of parallel vectors is obviously zero. The magnitude
of the vector product has the geometrical meaning of surface of the parallelogram with sides A and
B. The easiest way to introduce it is by specifying the vector products of unit basis vectors:

x
x
x
y
x

z
0

z
y
y
x
y

z = y
x
y
y
y

z =
z 0
x

(A.16)

zx

zy

z
z
y

x 0
from which we see very clearly that the vector product is anticommutative. Recalling (A.7) again

Bx
x
y

z By
A B = Ax Ay Az y

z
Bz

z
y
Bx
z 0
x
By
(A.17)
= Ax Ay Az
y

x 0
Bz

z
= det Ax Ay Az
Bz By Bz

PRELIMINARY VERSION

The last expression is the traditional one (although only mnemonic!), where the determinant must
be expanded using the elements of the first line.
Similar expressions hold for the cylindrical and
cylindrical system

z =

z
z

and for the spherical one

r



r

spherical coordinates. In particular, for the

z
0

z =
z 0

z
z

(A.18)

r
r

r 0

(A.19)

r
r

=
r

These definitions can be easily remembered by the method of Fig. A.3 A useful identity to know

Figure A.3.

Correct order of unit vectors

is the one that allows to expand a double vector product

A (B C) = (A C) B (A B) C
Vector and scalar product can be combined in the mixed product A (B C). This number is
the volume of the (skewed) parallelepiped with edges defined by A, B and C; it is clearly zero if
the three vectors lie in the same plane.
An important concept associated to vector fields is that of field line. A field line is a line such
that its tangent at any point r is parallel to the vector field E(r) in that point. They are useful
tools for visualizing vector fields. If the vector field is regular, for each point only a field line passes.
Sometimes in a region two vector fields are given and there is functional dependence of one
on the other. The simplest relationship that can exist is a linear one, so that the principle of
superposition holds. An example in electromagnetism is the relationship existing between the
electric field and the electric induction in a dielectric. Or the one between an electric current and
the electric and magnetic fields it produces. Referring to the former example, we can write in
abstract form
D = L{E}
However just as in the case of vectors, it is generally useful to introduce a basis to carry out
computations. Using a cartesian basis, the previous equation becomes, because of linearity
D = Dx x
+ Dy y
+ Dz
z = Ex L{
x} + Ey L{
y} + Ez L{
z}
In matrix form

Dx
Dy =
Dz

L{
x}

L{
y} L{
z}

Ex
Ey
Ez

PRELIMINARY VERSION

Now L{
x} is a (abstract) vector with three components along the three axes that can be collected
in a column vector. The same can be done for the other two elements so that a 3 3 matrix arises
and the previous equation can be written, with a more physical notation

Dx
xx xy xz
Ex
Dy = yx yy yz Ey
(A.20)
Dz
zx zy zz
Ez
where

xx
L{
x} = yx
zx

xy
L{
y} = yy
zy

xz
L{
z} = yz
zz

The row index denotes the component, the column index the unit vector. The matrix represents
the abstract linear operator L in the cartesian basis:

xx xy xz
L yx yy yz =
zx zy zz
and eq.(A.20) can be rephrased as
D = E
Following in the reverse direction the steps leading from (A.5) to (A.6), this equation can be written
L = xx x
x
+ xy x
y
+ xz x

z + yx y
x
+ yy y
y
+ yz y

z + zx
zx
+ zy
zy
+ zz
z
z

(A.21)

Symbols such as x
y
are called (unit) dyads and play just the role of placeholders to denote a
specific row and column. However, it is simple to give rules to operate with scalar products on
dyads so as to reproduce the results that could be obtained by the matrix formalism. The rules
are the following
def

Ex
y
= (E x
) y
= Ex y

(A.22)

def

x
y
E = x
(
y E) = Ey x

Eq.(A.21) is called the dyadic form of the linear operator L.

A dyad must not necessarily be formed by two unit vectors. If A, B are arbitrary vectors, we
form the dyad AB with the rule of operation
def

E AB = (E A) B

(A.23)

def

AB E = A (B E)
If we want to obtain the matrix that represents the abstract operator L = AB in the cartesian
basis, we can use a distributive law of the side-by-side-placement of two vectors
L = AB = (Ax x
+ Ay y
+ Az
z) (Bx x
+ By y
+ Bz
z)
= Ax B x x
x
+ Ax By x
y
+ Ax Bz x

z + Ay Bx y
x
+ . . . + Az B z
z
z
The matrix form can be obtained immediately

Ax Bx Ax By Ax Bz
Ax
L Ay Bx Ay By Ay Bz = Ay
Az Bx Az By Az Bz
Az

Bx

By

Bz

(A.24)

PRELIMINARY VERSION

It is possible to introduce the vector product of dyads and vectors: it is just enough to extend
(A.23):
def

E AB = (E A) B

(A.25)

def

AB E = A (B E)
Useful linear operators are
the identity I, such that
IA=AI=A
for any vector A. Its dyadic and matrix form in cartesian

1 0
I=x
x
+y
y
+
z
z 0 1
0 0
in cylindrical

1
I = +

+
z
z 0
0
and in spherical

0
I=
r
r + +

coordinates are

0
0
1

0
1
0

0
0
1

0
1
0

0
0
1

the identity transverse to

z, i.e. the projection operator onto the x,y plane, such that
Itz A = A Az
z

Itz

1 0
=x
x
+y
y
0 1
0 0

0
0
0

the identity transverse to

r, i.e. the projection operator onto the tangent plane to the sphere
r =const., such that Itr A = A Ar
r

Itr

+
=

0
0

0
1
0

0
0
1

Consider again the vector product of two vectors A B = C. Since the vector C depends linearly
on B, we can say that it is the result of the application of a linear operator on B. This operator
can be found by recalling the second line of (A.17)

z
y
Bx
z 0
x
By
A B = Ax Ay Az
y

x 0
Bz

0
Az Ay
Bx
0
Ax By
z + Az y
Ax
z Az x
Ax y
+ Ay x
= Az
= Ay
Ay Ax
0
Bz
(A.26)

PRELIMINARY VERSION

10

First we have carried out the multiplication of the row vector times the (formal) matrix, obtaining
a row vector whose elements are abstract vectors. Their components can be interpreted as columns
of the successive matrix. Alternatively,
AB=AIB
so that the matrix we are looking for is the one representing A I, i.e.
A I = (A x
) x
+ (A y
) y
+ (A
z)
z
Then
(A x
) = (Ax x
+ Ay y
+ Az
z) x
= Ay
z + Az y

(A y
) = (Ax x
+ Ay y
+ Az
z) x
= Ax
z Az x

(A
z) = (Ax x
+ Ay y
+ Az
z)
z = Ax y
+ Ay x

(A.27)

Substituting in the previous equation we get

A I = (Ay
z + Az y
)
x + (Ax
z Az x
)
y + (Ax y
+ Ay x
)
z
= Ay
zx
+ Az y
x
+ Ax
zy
Az x
y
Ax y

z + Ay x

z
In conclusion

0
A I Az
Ay

Az
0
Ax

Ay
Ax
0

As a direct application of this result, we consider another useful operator to be added to the
previous list. It is
rI=
r Itr :

0 0 0

r I 0 0 1

0 1 0
where we have used the spherical basis and clearly
r (1 0 0)T in this basis.

A.2

Calculus of vector fields

In this section we review the basic concepts of calculus, i.e. derivatives and integrals, applied to
vector fields. Let us start with a scalar field h(x,y), where h is a smooth function of only two
variables for simplicity. The partial derivatives
h
x

h
y

evaluated in the point P (x0 ,y0 ) have the meaning of local rate of change of h for small
increments of the coordinates dx, dy in the x, y directions around point P , respectively. A quantity
of interest is the directional derivative of h in the direction
s, which has the meaning of local rate
of change of h for small displacements in the
s direction. According to the chain rule,
h dx h dy
h
=
+
s
x ds
y ds

PRELIMINARY VERSION

11

sy

dy

ds

sx

dx

Figure A.4.

Direction along which the directional derivative is computed

With reference to Fig. A.4, we find

dx
= cos = sx
ds
so that

dy
= sin = sy
ds

h
h
h
=
sx +
sy
s
x
y

This formula can be considered as a scalar product of the vector

gradh = h =

h
h
x
+
y

x
y

called gradient of h, times the unit vector

s. The gradient of a scalar function is a vector field.
The magnitude of the gradient is the maximum rate of change of h when the direction
s is allowed
to vary. The direction in which his happens is the direction of the gradient. If h is the height
of the ground above the sea level, so that h(x,y) can be interpreted as the local height of a hill,
the direction of the gradient is that of the steepest slope. Note, however, that the gradient is a
vector belonging to the x, y plane and it is not tangent to the surface. When
s is orthogonal to the
gradient, the directional derivative is zero: in that direction h is not changing, hence the gradient
is always orthogonal to contour lines (lines on which h= const). Fig. A.5 shows an example of
a scalar field h(x,y). Fig. A.6 shows a contour plot of the same function with the vector field
gradh, computed numerically by finite differences. Note that the gradient is zero in the extrema
of the graph of h, i.e. maxima, minima and saddle points. Also, the arrows are orthogonal to the
contours.
If the scalar field h(x,y,z) depends on the three space coordinates, the gradient gradh is defined
as
gradh = h =

h
h
h
x
+
y
+

z
x
y
z

(A.28)

If the scalar field is given in a non cartesian system of coordinates the expression of the gradient
are more complicated, essentially because the unit vectors are not constant. It can be proved that
in cylindrical coordinates
gradh(,,z) = h =

h
1 h
h
+

(A.29)

PRELIMINARY VERSION

12

z=h(x,y)

10
8
6
4
2
0
2
4
6
8
3

z
y

2
0

1
0

2
3

Figure A.5.

Scalar field z = h(x,y)

grad h(x,y) and contours

2

1.5

0.5

0.5

1.5

2
2

1.5

Figure A.6.

0.5

0
x

0.5

1.5

Contours of z = h(x,y) and vector field gradh(x,y)

PRELIMINARY VERSION

Figure A.7.

13

Closed surface for the definition of the flux of a vector field.

in spherical coordinates
gradh(r,,) = h =

h
1 h
1 h

r+
+

r
r
r sin

(A.30)

We turn now to a vector field A(r) and define its flux across a surface, which may be either open
or closed. Fig. A.7 shows a closed surface enclosing the volume V , with the outward normal
unit vector
. The outward flux of A through is defined by the scalar
(A) =

A
d

Recalling Gauss Theorem

AdV =
V

A
d

(A.31)

we can interpret this equation as defining an average divergence

average{ A} =

1
V

AdV =
V

1
V

A
d

In the limit of the volume reducing to a point we obtain the divergence as flux per unit volume
across a small closed surface surrounding the point. It can be shown that the explicit expressions
for the divergence of a vector field are
in cartesian coordinates
Ax
Ay
Az
+
+
x
y
z

(A.32)

1
1 A
Az
(A ) +
+


z

(A.33)

div A = A =
in cylindrical coordinates
div A = A =

PRELIMINARY VERSION

14

in spherical coordinates
div A = A =

1 2
1

1 A
(r Ar ) +
(sin A ) +
r2 r
r sin
r sin

(A.34)

In conclusion, the divergence of a vector field is a particular combination of derivatives of the field
components, whose meaning is clarified by Gauss theorem. Consider a flow of charges across a
closed surface so that A = J = v where v is the velocity field and the volume charge density.
The units of J are A/m2 , hence it is a current density. The flux (J) has the units of A and
it yields the amount of charge coming out every second from the total volume V , i.e. the total
current. The source of this outflow is the divergence of J, which tells us how much charge per
unit volume is coming out every second from each point of V . The flux of J provides a global
information, the divergence of J a local one.
Finally it is useful to write Gauss theorem (A.31) for the one dimensional domain [a,b], with
boundary consisting of the points x = a (with outward normal
x) and x = b (with outward
normal x
)
b
Ax
dx = Ax (b) Ax (a)
a x
Clearly, this is just the fundamental formula of integral calculus.
Another type of integral that we can form with a vector field is a line integral
A ds

where the curve may be either open or closed. Assume it is closed, so that the integral is called
the circulation of A. Stokes theorem states that
A
do =
o

A ds

(A.35)

where o is an open surface with as boundary; the orientation of the tangent vector is related

Figure A.8.

Geometry for the application of Stokes theorem

to that of
by the right-hand-rule, as shown in Fig. A.8. For the purpose of the definition, it
is convenient to limit the generality of the theorem and consider a planar curve so that o can

PRELIMINARY VERSION

15

be taken as the part of plane inside , with constant normal vector

. This equation allows us to
define the
component of an average curl
average{ A}
=

1
o

A
do =
o

1
o

A ds

Obviously, by choosing three loops with linearly independent normals we can define completely
the total amount of rotation or average curl. By letting the size of the loops go to zero, we can
define the curl in a point as circulation per unit area in the neighborhood of a single point. By
carrying out this prescription in different systems of coordinates, it can be proved that the explicit
expressions of the curl of a vector field are
in cartesian coordinates
curl A = A =
Ay
Az

y
z

x
+

Az
Ax

z
x

y
+

Ax
Ay

x
y

(A.36)

in cylindrical coordinates
curl A = A =
A
1 Az

A
Az

1
1 A
(A )

(A.37)

in spherical coordinates
curl A = A =
1

A
(A sin )
r sin

1
Ar
(rA )

r r

r+

1
r

1 Ar

(rA ) +
sin
r

(A.38)

If we examine the expressions of gradient, divergence and curl in cartesian coordinates we can
identify a kind of formal vector defined by
=x

+y

+
z
x
y
z

(A.39)

such that the various differential operators can be imagined to be formed by means of standard
product, scalar product and vector product of times a scalar field or a vector field. It is to
be remarked that such an interpretation (due to the American physicist W. Gibbs 1839-1903) is
possible only in cartesian coordinates. Hence, for instance in spherical coordinates, even if the
curl of A is written usually as A, this expression has to be taken as a single symbol (whose
meaning is given by (A.2)) and not interpreted as the product of two factors.
First order differential operators can combined to form second order operators.
The gradient of a scalar field is a vector, so that we can compute the curl and divergence of
it. However
h = 0
identically for any smooth h(r). To remember the property, we can note that the vector
product of two equal vectors is zero.

PRELIMINARY VERSION

16

The divergence of a gradient produces the Laplace operator or laplacian

h = 2 h =

We recall only the expression of the laplacian in cartesian coordinates

2 h =

2h 2h 2h
+ 2 + 2
x2
y
z

The divergence of a vector field A(r) is a scalar, hence its gradient can be computed
( A)
The curl of a vector field A(r) is a vector, hence its divergence and curl can be computed.
However
( A) = 0
identically for any smooth A(r). To remember the property, we can note that the triple
mixed product with two equal factors is zero.
As for the curl of the curl, the following identity is to be noted
( A(r)) = ( A) 2 A
Often multidimensional integrals have to be computed. It is useful to remember from Fig. A.2
that the elementary volume in spherical coordinates is a cube with edges of size dr (along
r), rd
and r sin d (along ),
(along )
thus
dV = r2 sin drdd
Concerning surface integrals in spherical coordinates, the elementary patch is a square with sides
and r sin d (along ),
rd (along )
thus
d = r2 sin dd
Connected with this is the concept of solid angle, displayed in Fig. A.9
The natural measurement unit of a plane angle is radian (rad). A one radian angle is the one
with the vertex at the center of a circle of radius r that subtends an arc whose length is equal to
the radius r. Thus the radian measure of an angle is equal to the ratio between the length of the
subtended arc and the radius: for instance, the measure of a full round angle is 2 radians.
Similarly, the solid angle, , is the two-dimensional angle in three-dimensional space that an
object subtends at a point; it is measured in steradians (sr). A one steradian solid angle is the one
with the vertex at the center of a sphere of radius r that subtends a patch whose area is equal to
r2 . Since the surface of a sphere is 4r2 , the measure of the total solid angle around a point is 4
steradians. As another example, the solid angle defined by x 0, y 0, z 0 has a measure of
4/8 = /2 steradians. The elementary solid angle subtended by a patch of area d is, in spherical
coordinates
d
d = 2 = sin dd
r

PRELIMINARY VERSION

Figure A.9.

A.3

17

Definition of radian for a plane angle and of steradian for a solid angle

Multidimensional Dirac delta functions

The one dimensional Dirac function is a distribution defined by the property

f (x)(x x0 )dx = f (x0 )

for any function f (x) continuous in x = x0 . If f (x) 1 identically, this becomes

(x x0 )dx = 1

The support of (x x0 ) is the point x = x0 .

Suppose now that f (x,y,z) is a function of three variables and suppose that we want to compute

I=

f (x,y,z)(x x0 )dxdydz

Integrating first with respect to x leads to

I=

f (x0 ,y,z)dydz

The support of (x x0 ) is still x = x0 , which in this case is not a single point but a plane.
Physically, such a function can describe a surface charge. The value of the integral has been

PRELIMINARY VERSION

18

obtained by evaluating the function f on the support of the function and then integrating along
it.
This concept can be generalized if we consider a function with support on a smooth surface of
R3 with parametric equation r = r (u,v). Consider the integral
I=

f (r)(r r )dr

Again its value is obtained by evaluating f on the support r of the function and then integrating
along it
r
r
dudv

I = f (r )d =
f (r (u,v))
u
v
where
d =

r
r
dudv

u
v

is the elementary area on the surface . Note that the function can be described as one dimensional here and its support is two dimensional.
Consider now a function with one dimensional support consisting of the straight line parallel
to the z axis with equation
x = x0
y = y0
Physically, it can describe a line charge. Mathematically, its expression is (x x0 )(y y0 ), hence
it can be defined as two dimensional, because it is the product of two one dimensional functions.
We can be interested in computing the integral

I=

f (x,y,z)(x x0 )(y y0 )dxdydz

Its value is obviously the integral of f along that line

I=

f (x0 ,y0 ,z)dz

A generalization of this result is obtained if the support of the function is a smooth curve R3
with parametric equation r = r (s). Suppose we must evaluate the integral
I=

f (r)(r r )dr

Once again, its value is obtained by evaluating f on and integrating along it

I=

f (r )d =

where
d =

f (r (s))

dr
ds
ds

dr
ds
ds

is the line element on .

The last example to consider is that of a three-dimensional function with support on the
point
x = x0
y = y0
z = z0

PRELIMINARY VERSION

19

which can represent a point charge. Next, consider the integral

I=

f (r)(r r0 )dr

Its value is
I = f (r0 )
We can say that this result is obtained by evaluating the function f on the support of the function
and then integrating along it. However, since this support is zero-dimensional, this last integral
disappears.
In conclusion, all the integrals can be evaluated by the same rule: evaluate the integrand on
the support of the function and integrate along it. Notice that the sum of the dimensionality of
the function and of its support have constant sum 3, the dimensionality of ambient space.

Appendix B

Solved Exercises
B.1

Polarization and Phasors

Exercise n. 1
A time harmonic electric field is E(t) = cos 0 t x
+ sin 0 t y
. Determine the type of polarization,
draw the polarization curve defined by this vector and write the expression of the phasor E.
Solution Write
E(t) = R x
ej0 t + R j
y ej0 t
= R (
x j
y) ej0 t
Hence the phasor is E = x
j
y. Compute
E E =x
(
y) =
z=0
E E =x
(
y) = 0

and

|E | = |E |(= 1)

The polarization is circular counterclockwise. The plot of the polarization curve is shown in Fig.
B.1.
Alternatively, the polarization is circular because
Ex = 1

Ey = j = ej/2

= /2

|Ex | = |Ey | = 1

Exercise n. 2

A time harmonic electric field is E(t) = 2 cos(0 t + 45 ) (

x+y
) + 2 sin(0 t + 45 ) (
xy
).
Determine the type of polarization, draw the polarization curve defined by this vector and write
the expression of the phasor E.
20

PRELIMINARY VERSION

21

y
2

1.5

0.5

0
x
0.5

1.5

2
2

1.5

Figure B.1.

0.5

0.5

1.5

Polarization curve defined by E(t) above

Solution Write

E(t) = R (
x+y
) 2ej45 ej0 t + R j(
xy
) 2ej45 ej0 t
= R {(1 + j)(
x+y
) j(1 + j)(
xy
) ej0 t }
= R {(1 + j)(
x+y
) + (1 j)(
xy
) ej0 t }
Hence the phasor is E = 2
x + j2
y. Compute
E E = 2
x (2
y) = 4
z=0
E E = 2
x (2
y) = 0

and

|E | = |E |(= 2)

The polarization is circular clockwise. The plot of the polarization curve is shown in Fig. B.2
Alternatively, the polarization is circular because
Ex = 2

Ey = j = ej/2

= /2

|Ex | = |Ey | = 2

Exercise n. 3
The phasor of a magnetic field is H = (1 + j)
x + 2(1 + j)
y. Determine the type of polarization,
write the expression of the time varying field H(t) and draw the polarization curve defined by this
vector.
Solution Find real and imaginary part of the phasor
H =x
+ 2
y

H =x
+ 2
y

PRELIMINARY VERSION

22

y
4

1
x
0

4
4

Figure B.2.

Polarization curve defined by E(t) above

y

1
x
0

4
4

Figure B.3.

Polarization curve defined by H(t) above

Compute
H H = (
x + 2
y) (
x + 2
y) = 0
The polarization is linear.
The time varying field is shown in Fig. B.3
H(t) = (
x + 2
y)(cos 0 t sin 0 t)

= (
x + 2
y) 2 cos(0 t + 45 )
The plot is shown in Fig. B.3

PRELIMINARY VERSION

B.2

23

Plane Waves

Exercise n. 1
Consider a plane wave in a dielectric with r = 4, r = 1, at the frequency f = 10GHz. The
electric field in the origin is
E(0) = 2
x j
y+
z V/m
1. Compute the phase velocity vph , the wavelength , the wave impedance Z, the wavenumber
k in deg/cm, the power density dP/d.
2. Find the polarization of the electric field E0 . Find the direction of propagation
s knowing
that the phase of the wave decreases in the z direction.

3. Compute the magnetic field in the point P : (2,2,2)T 5 cm at the time t = T /4

Use the approximate values c = 3 108 m/s and Z0 = 377.
Solution
Phase velocity:

1
c
c
vph = =
= = 1.5 108 m/s

r r
2

Wavelength

vph
1.5 108
=
= 1.5 102 m
f
1.0 1010

=
Wave impedance:
Z=

= Z0

r
Z0
=
= 188.5
r
2

Wavenumber:
2
6.2832
=
= 4.1888 rad/cm

1.5 102
360
=
= 240.0 deg/cm

k=

Power density
dP
1 |E0 |2
1 |E0x |2 + |E0y |2 + |E0z |2
1 (4 + 1 + 1)
2
=
=
=
= 0.01591 W/m
d
2 Z
2
188.5
2 188.5
Polarization:
The real and imaginary part of E(0) are
E0 = 2
x+
z

E0 =
y

They are not parallel, hence the polarization is not linear. Moreover
E0 E0 = 0

but

|E0 | = |E0 |

hence the polarization is not circular. In conclusion, it is elliptical.

Direction of propagation:
Since the electric field is perpendicular to
s, we have

s=

2
z+x

E0 E0
(2
x+
z) (
y)
=
=
|E0 E0 |
|(2
x+
z) (
y)|
5

PRELIMINARY VERSION

24

The phase factor is

exp[j(kx x + ky y + kz z)]
If the phase is decreasing in the z direction, then kz = ksz > 0, hence

s=

2
zx

Phasor of magnetic field in the origin

2
zx

H0 = Y
s E0 = Y

(2
x j
y+
z) = Y

2
5
1
j x
+ y
+j
z
5
5
5

Phasor of magnetic field in P :

H(P ) = H0 exp j

= H0 exp j

2
2

s rP = H0 exp j

2
zx

(2
x + 2
y + 2
z) 5 =

2 1
2 2 cm
(2 2 2) 5 = H0 exp j
= H0 exp [j 120 ]
5
1.5 cm

Magnetic field in P at t = T /4
H(rP ,t) = R H(rP )ej0 t

t=T /4

= R H0 ej(90

= H0 cos(30 ) H0 sin(30 ) = Y
1
=
188.5

1
x
+
5

120 )

2
1
5 3
y
+ x
+
z
2 5
2 5
2 5

15
1
y
+
z
2
2 5

Exercise n. 2
Consider a plane wave propagating in the z direction in a dielectric with r = 4,r = 1 and
= 0.01 S/m at the frequency f = 1.0 GHz and E0 = x
.
1. Compute the wavenumber k, the phase velocity vph , the wavelength
2. Compute the wave impedance Z, the active power density in the origin dP/d, the attenuation dB in dB/m
Use c = 2.99792458 108 m/s, 0 = 8.854 1012 F/m.
Solution
Wavenumber:
k=

0 r j

0 = 0 0

r j

= k0
0

r j

PRELIMINARY VERSION

25

The free space wavelength and wavenumber are

c
= 0.29979 m
f

0 =

k0 =

2
= 20.9585 rad/m

Then
k = 20.9585

4 j 0.1798 = 20.9585(2.0005 j 0.0449) = 41.9275 j 0.9416 m1

The real and imaginary part are

= 41.9275 rad/m
Phase velocity:
vph =

= 0.9416 Np/m

= 1.4986 108 m/s

Wavelength
=

vph
= 0.1499 m
f

Wave impedance:
Z=

=
0 r j/

0
0

r
= Z0 (0.4996 + j 0.0112) = 188.2227 + j 4.2270
r j/(0 )

Wave admittance
Y =

1
1
=
= (5.3102 j 0.1192) 103 S
Z
188.2227 + j 4.2270

Notice that
R{Y } =

1
1
=
= 5.3129 103 S
R{Z}
188.2227

In this case the difference is small because the phase of Z is small but becomes enormous when
this phase approaches /2.
Active power density in the origin
dP
1
1
2
= R{Y }|E0 |2 = 5.3102 103 (|E0x |2 + |E0y |2 + |E0z |2 ) = 2.6551 103 W/m
d
2
2
Attenuation:
= Im{k} = 0.9416 Np/m
dB = 20 log10 e = 8.68589 = 8.1785 dB/m

Exercise n. 3
Consider a plane wave propagating in free space at the frequency f = 5 GHz. The electric field in
the points of the plane z = 0 has the value
E(x,y,z)|z=0 = E0 exp[j(x + y)]

x, y

PRELIMINARY VERSION

26

with = k0 /5 and = k0 /2 and it is known that this wave is a TE field, i.e. E0 has no zcomponent and that it carries the power density dP/d > 0 through the plane z = 0.
Compute the direction of propagation
s and the spherical angles that define this direction. Compute
also the fields in the origin E0 , H0 .

Solution
The propagation factor of a plane wave is
exp[jk r] = exp[j(kx x + ky y + kz z)]
hence, by inspection, we find kx = and ky = . From the dispersion relation
kx2 + ky2 + kz2 = 2 0 0 = k02
it follows that
kz =

k02

kx2

ky2

k02

= k0

71
1
1
1
= k0
25 4
10

The sign of the square root is taken to be positive because power is flowing toward the region
z > 0, due to the fact that dP/d > 0 through a d belonging to the plane z = 0. From this

k
71
1
1

s=
= x
+ y
+

z
k0
5
2
10
The spherical angles that identify the direction
s are found by recalling (A.2) and noting that
|
s| = 1
sx = sin cos
sy = sin sin
sz = cos
From this we find

71
= 32.5827 deg
10
sx
sy
= arccos
= arcsin
= 68.1986 deg
sin
sin
= arccos sz = arccos

The electric field in the origin, with magnitude

E0 =

2Z0

dP
d

must be perpendicular to
s because it is a plane wave and perpendicular to
z because required in
the text. Then
E0 = E 0

k
z
(
x +
y + kz
z)
z
(
x
y)

s
z
= E0
= E0
= E0
2
|
s
z|
|k
z|
|k
z|
+ 2

Thanks to the denominator, the vector multiplying E0 has unit magnitude.

PRELIMINARY VERSION

27

The magnetic field is computed by the impedance relation

Y0
Y0 E0
(
x
y)
k E0 =
(
x +
y + kz
z)
2
k0
k0
+ 2

H0 = Y0
s E0 =
=

Y0 E0
k0

2 + 2

kz (
x +
y) ( 2 + 2 )
z

Exercise n. 4
Consider a plane wave field propagating in free space. The electric field is
E(x,y,z) = E0 x
cos kz
Compute the magnetic field.

Solution
The general relation between electric and magnetic fields is the first Maxwell equation:
H=

1
E
j

In the case of a plane wave this equation simplifies in the impedance relation
H=

1
k E = Y
sE

The given electric field is not a plane wave, but is the sum of two counter-propagating ones, as it
can be seen by Eulers formula
E(x,y,z) = E0 x
cos kz =

E0 jkz E0 jkz
x
e
+
x
e
2
2

The direction of propagation of the first wave is

s=
z, that of the second
s =
z. Now we apply
the impedance relation to each wave, using the appropriate unit vector
s:
E0
E0

zx
ejkz + Y0 (
z) x
ejkz
2
2
E0 jkz
E0 jkz
e
Y0 y
e
= Y0 y
2
2
= jY0 E0 y
sin kz

H(x,y,z) = Y0

PRELIMINARY VERSION

B.3

28

Antennas

Exercise n. 1
Consider a horn antenna whose rectangular aperture has size A = 10 cm, B = 4 cm, operated at
the frequency f =20GHz. Compute
1. the maximum gain G in natural units and in dB.
2. the full width of the main lobe between the zeros in the E and H plane
3. the direction and the level of the first sidelobe in the E and H plane
4. the magnitude of the radiated electric and magnetic field on axis, at the distance R = 5 km,
when the horn has an input power Pin = 10 W.
Solution
1. The geometrical area of the aperture is Ag = AB = 40 cm2 .
2. The aperture efficiency of a rectangular horn is a = 0.8, hence the effective area is Aeq =
a Ag = 0.8 40 = 32 cm2
3. The wavelength is = c/f = 3. 108 /20 = 1.5 cm
4. The maximum gain is G = (4/2 )Aeq = 179 = 22.5 dB. The directivity is the same:
antennas with size comparable to wavelength have ohmic efficiency very close to one.
5. Compute the far field distance rf f = 2D2 /. The characteristic size D is the diagonal of the
rectangle D = 10.77 cm and rf f = 1.54 m. Since the observation distance is R
rf f the
concept of gain can be used. Hence, we compute the power density per unit surface in the
observation point:
dPrad
Pin
=G
= 5.6977 106
d
4R2
From this we compute the electric field via
|E| =

2Z0

dPrad
= 65.5 mV/m
d

and the magnetic field from the impedance relation

|H| =

|E|
= 173 A/m
Z0

Exercise n. 2
Consider a radio link between a paraboloid and a half wavelength dipole antenna. The paraboloid
has a diameter D = 1 m, is operated at the frequency f = 10 GHz, radiates a circularly polarized
field and its input power is Pin = 0.1 W. The receiving dipole has ohmic efficiency = 1, is located

PRELIMINARY VERSION
x

29

z
y

(b)

(a)
Figure B.4.

Radio link scheme.

on axis at the distance R = 20 km and perpendicular to the link direction. However it is not
vertical, but forms the angle with the vertical, as shown in Fig. B.4. Compute the received
power as a function of the angle .
Solution
The geometrical area of the paraboloid aperture is Ag = D2 /4 = 0.7854 m2 .
The aperture efficiency of a paraboloid is a = 0.6, hence the effective area is Aeq = a Ag =
0.6 0.7854 = 0.4712 m2
The wavelength is = c/f = 3. 108 /10 = 3.0 cm
The maximum gain is G = (4/2 )Aeq = 6580 = 38.2 dB. The directivity is the same:
antennas with size comparable to wavelength have ohmic efficiency very close to one.
Compute the far field distance rf f = 2D2 /. The characteristic size D is the diameter of the
aperture and rf f = 66.67 m. Since the observation distance is R
rf f the Friis equation
can be used.
The maximum gain of the dipole (equal to the directivity since the ohmic efficiency is = 1)
is G = D = 1.643
The polarization of the transmitting antenna is circular. The electric field lies in the x, y
plane and the polarization vector can be written
p
T X = c(
x j
y)
The double sign is necessary because the text does not specify whether the polarization is
clockwise or counterclockwise. The constant c has the value that makes p
T X a unit vector:
|
pT X |2 = p
T X p
T X = c(
x j
y) c (
x j
y) =
= |c|2 (
xx
j
xy
j
yx
+y
y
) = |c|2 (1 + 1) = 2|c|2
Alternatively,
|
pT X |2 = |(
pT X )x |2 + |(
pT X )y |2 = |c|2 + |c|2 = 2|c|2

PRELIMINARY VERSION

30

Hence c = 1/ 2 and

1
p
T X = (
x j
y)
2

The polarization of the receiving antenna is, by definition, the polarization of the E field
radiated in the direction of the paraboloid if the dipole were used as transmitting antenna.
The link direction coincides with the one of maximum radiation and the E field is parallel
to the antenna. Hence
p
RX = (cos
x sin
y)
Compute the polarization matching factor
1
p = |
pT X p
RX |2 = (cos j sin )
2

1
2

We see that p is independent of and has the constant value of 1/2: a 3 dB loss, with respect
to the case of polarization matching.
The available power at the output terminals of the receiving antennas is given by Friis formula
Pavail = Pin

GT X GRX
4R

|
pT X p
RX | = 0.1

6580 1.643
3

4 20 10
3 102

1
= 7.7019 1012 W
2

Exercise n. 3
Consider a radio link between two identical paraboloids. The paraboloids have clockwise circular
polarization. Compute the polarization matching factor.
Solution
Obviously, the result must be 1, since two identical antennas are polarization matched by definition.
The computation is, however, a little tricky.
Consider the transmitting antenna. Then
1
loc
p
T X = (
xloc
yTX
)
TX + j
2
With reference to Fig. B.5, we realize that the previous phasor describes really a clockwise polarization if observed from a point in the far field of the TX antenna. The axes have been identified
with a T X subscript and a loc (as local) superscript because they are to be considered as attached
to the transmitting antenna.
Now take this antenna, together with the reference system, and rotate it 180 around the xloc
T X axis
in order to make it coincident with the receiving antenna. The reference system now is different
from the original one and is indicated with the RX subscript. Clearly, the field radiated by the receiving antenna if it were connected to a generator would have still clockwise circular polarization,
as observed by a point in the far field of this RX antenna and its expression would be
1
loc
xloc
yRX
)
p
RX = (
RX + j
2

PRELIMINARY VERSION

31
loc
xTX

xloc
RX

loc
yRX

loc
yTX

Figure B.5.

ETX

loc
zTX

Radio link scheme with two identical antennas.

Note that
x
loc
loc
RX = x
TX
loc
loc
y
RX
=
yTX

zloc

zloc
RX =
TX
Then we compute the polarization matching factor
1
1
loc
xloc
yTlocX ) (
xloc
yRX
)
p = |
pT X p
RX |2 = (
T X + j
RX + j
2
2

1
(1 + j j (1)) = 1
2

as it should be.

Exercise n. 4
A Hertzian dipole of length L = 2 m is fed by a transmission line with characteristic impedance
Z = 50 . The dipole has an ohmic efficiency = 0.8 and an input capacitance C = 30 pF.
Suppose that this dipole is required to radiate a total power Prad = 100 W at the frequency
f = 10 MHz. Compute:
1. the voltage that must be applied at the dipole input
2. the reflection coefficient at the dipole input and the incident power on the line so that
Prad = 100 W. Compute also the maximum and minimum voltage on the line and the
VSWR.
Solution
1. The wavelength is = 3 108 /107 = 30 m The radiation resistance is
Rrad = Z0
The input resistance is
Rin =

2
3

= 3.5

Rrad
= 4.4

PRELIMINARY VERSION

32

The input impedance is

Zin = Rin j

1
= 4.4 j 530.5
2f C

The input current Ia is such that the radiated power is

Prad =

1
Prad |Ia |2
2

Hence
2Prad
=
Rrad

|Ia | =

2 100
= 7.55 A
3.5

The voltage Va to be applied is

|Va | = |Zin ||Ia | = 4005 V
2. The input reflection coefficient is
a =

Zin Z
= 0.9985 ej10.8
Zin + Z

The input power is Prad / = 125 W. The necessary incident power is

P+ =
On the other hand P + =

+ 2
1 |V |
2 Z ,

Pin
= 40.3 kW
1 |a |2

hence

|V + | =

2Z P + = 2008 V

The maximum voltage is then

Vmax = |V + |(1 + |a |) = 4013 V
and the minimum
Vmin = |V + |(1 |a |) = 3.1 V
The VSWR is
V SW R =

1 + |a |
= 1291
1 |a |

Obviously, using this antenna in these conditions is absolutely ridiculous and a matching
device is to be used.

Exercise n. 5
Find the sizes A, B of a rectangular horn with gain G = 19 dB at the frequency f = 12 GHz and

PRELIMINARY VERSION

33

symmetrical main lobe, as measured between the zeros, i.e. with equal beamwidth in the E and
H planes. Compute also the full beamwidth.
Solution
The first zero in the E plane is located at y = , from which

sin E
z1 =

Similarly, the first zero in the H plane is located at x = 3/2, from which
sin H
z1 =

3
2A

E
If H
z1 = z1 , then B = 2A/3.
The geometrical area is

Ag = AB =

2A2
3

The maximum effective area is

Aeq = a Ag = a

2A2
3

4
8a
Aeq =
2
3

The maximum gain is

G=

from which we get

A
=

3
G
8a

The required gain is GdB = 19 dB, i.e.

G = 10GdB /10 = 79.43
We find then A = 3.44 = 8.600 cm, since the wavelength is
=

c
3 108
=
= 2.5 102 m
f
12 109

Moreover B = 5.733 cm and

E
z1 = arcsin

= arcsin(0.436) = 25.852

The full beamwidth (between the zeros) is

H

BW = 2E
z1 = 2z1 = 51.704

Exercise n. 6
Consider a paraboloid antenna with the following radiation pattern at the frequency f = 5 GHz:

g(,) = G0 cos for

2

=0
for >
2

PRELIMINARY VERSION

34

The full half-power beam-width is FHPBW=10 .

Compute G0 and estimate the antenna diameter.
Solution
First find from the FHPBW. Set 3dB =FHPBW/2=5 :
G0
= G0 cos 3dB
2
Take logarithms

3 = 10 log10 (cos 3dB )

Hence
=

3
= 181.18
10 log10 (cos 3dB )

Next recall the normalization condition

1
4

g(,) sin dd = 1

Compute the integral, noting that g(,) does not depend on , so that the integral equals 2:
1
4

2
0

g(,) sin dd =

Set u = cos
G0
2

1
4

u d(u) =

2
0

G0
2

G0 cos sin dd =

u du =

G0 u(+1)
2 +1

=
0

G0 2
4

cos sin d

G0
2( + 1)

In conclusion, the normalization condition yields

G0 = 2( + 1) = 364.36
that is (G0 )dB = 25.61 dB. To find the diameter of the antenna, compute the maximum effective
area
2
Aeq =
G0 = 1043.8 cm2
4
since = c/f = 6 cm. The aperture efficiency of paraboloids is a = 0.6, so
Ag =

Aeq
= 1739.7 cm2
a

and the required antenna diameter is

D=

4Ag
= 47.06 cm

Fig. B.6 shows the radiation pattern in cartesian coordinates, Fig. B.7 shows the same pattern in
polar coordinates.

PRELIMINARY VERSION

35

400

350

300

250

200

150

100

50

0
100

80

60

Figure B.6.

40

20

20

40

60

80

100

Radiation pattern in cartesian coordinates

90

400
60

120
300

150

30

200

100

180

210

330

240

300
270

Figure B.7.

Polar radiation pattern

PRELIMINARY VERSION

B.4

36

Waveguides

Exercise n. 1
Find the dimensions of a rectangular waveguide with the following specifications: .
1. it is single mode in the band between fmin = 10 GHz and fmax = 15 GHz
2. the attenuation of the first higher order mode must be at least lim = 8 dB/cm for the whole
band [fmin ,fmax ]
3. dispersion must be minimized
4. the waveguide must carry up to PT = 150 kW without discharge when it is matched
If the waveguide designed above is connected to a load with reflection coefficient |L | = 0.2, what
is the maximum power that the guide can carry without discharge?
Solution
Certainly we choose b < a/2 so that the first higher order mode is T E20 . Next find a.
Enforce
c
fmin fc10 =
2a
c
fmax fc20 =
a
from which we find

c
2fmin

c
fmax

that is
1.5 a 2 cm
Compute the attenuation of mode T E20 :
20 =

2 k2 = k
kt20

fc20
f

It is minimum at fmax
20min =

2fmax
c

fc20
fmax

By requiring 20min lim , we find a constraint on a (contained in fc20 )

c

a
fmax

clim
2fmax

= 1.919 cm

+1

Note that the value of lim given in dB must be converted in Neper, by diving it by 8.6859.
Dispersion is minimized if a = 1.919 cm is chosen. This value is inside the range 1.5 a 2 cm,
hence it is acceptable. Then the maximum value for b is bmax = a/2 = 0.960 cm
Now find b on the basis of the power to be transmitted PT . The active power flow on the T E10
modal line is
1 |V + |2
Pwg =
2 Zt10

PRELIMINARY VERSION

37

because the modal line is matched.

The maximum value of the electric field is reached at x = a/2, so that
Emax =

2 +
V
ab

Substituting in the previous expression, we find that the power in the guide for a given value of
Emax is
2
1 ab Emax
Pwg =
2 2 Zt10
Now, setting Emax = R = 20 kV/cm the dielectric rigidity of air, we obtain the power at the onset
of discharge Pdisch :
1 ab R2
Pdisch =
2 2 Zt10
Note that this power is a function of frequency because Zt10 is. At f = fmin the modal impedance is
maximum, so Pdisch is minimum: this is the most critical situation. If the condition Pdisch (fmin )
PT is satisfied, Pdisch (f ) PT for all f . Hence enforce
1 ab
R2
PT
2 2 Zt10 (fmin )
from which we find

4Zt10 (fmin )PT

= 0.472 cm
aR2
In conclusion we find 0.472 b 0.960 cm; any value of b in this range is acceptable.
Suppose b = 0.472 cm is chosen. Then the discharge power is exactly Pdisch = PT = 150 kW. If
the guide is mismatched, the discharge power is reduced by the factor 1/S,
b

Pdisch =
with
S=

1 ab R2 1
2 2 Zt10 S

1 + |L |
= 1.5
1 |L |

so the discharge power is Pdisch = 150/1.5 = 100 kW. In conclusion, the guide previously designed
can carry only a maximum power of 100 kW, in mismatched conditions, if discharge is to be
avoided.

Exercise n. 2
The rectangular waveguide WR90 of Fig. B.8 is connected to a load that has a reflection coefficient
L = 0.4 exp(j45 ) for the fundamental mode T E10 , at the frequency f = 12 GHz. Design a
small piece of dielectric to be inserted at a convenient position in the guide so that matching is
obtained. Then draw the voltage plot |V (z)| with V inc = 1 V.
Solution
Compute first the parameters of the modal lines. The waveguide is WR90, with a = 0.9 in=0.9

PRELIMINARY VERSION

38

r
A

kz 0 , Z 0

k zd , Z d

L
L

kz 0 , Z 0

Figure B.8. Matching of a load by means of a piece of dielectric and equivalent modal
circuit for the T E10 mode

2.54 cm. and fc0 = c/(2a) = 6.56 GHz. Recall

kz0 = k0

fc0
f

kzd = k0

Z0

Z0 =
1

fc0
f

Zd =

fc0
f

Z0
r

fc0
f

which yield
kz0 = 2.1043 rad/cm
Z0 = 450.2789

g0 = 2.986 cm

The matching device is a /4 line. It must be positioned in such a way that ZB is real ZB = RB .
Then the characteristic impedance of the line is computed according to
Zd =

RA RB

where RA = Z0 , the desired input impedance. To find RB draw on the Smith chart the circle
through L with center in the origin. The intersections with the real axis are
rm =

1 |L |
1 + |L |

and

rM =

1 + |L |
1
=
1 |L |
rm

Which one of the two must be selected? Let r be one of the two. Then

Zd = Z0 r

PRELIMINARY VERSION

39

From the equations above, Zd Z0 , hence r 1 so the point rm must be chosen. Now we can
compute r .

Zd = Z0 r

Z0
Z0
=
rm
2
2
fc0
fc0
r
1
f
f
fc0
f

1
=
rm

r =

1
rm

fc0
f

fc0
f

fc0
f

We find rm = 0.4286 and r = 1.9347. The phase of C is 45 : on the Smith chart, we read
TL

= 0.1875

This is exactly the value of LBC /g0 , since

TL

= 0.
B+

Hence
LBC = 0.1875g0 = 0.5599 cm
Moreover gd = 1.9548 cm and LAB = gd /4 = 0.4887 cm. If we choose rM instead of rm , we
obtain r 1, which is not realizable.
In order to draw the voltage plot, i.e. the electric field plot in the waveguide, we mark some
points on the Smith chart of Fig. B.9. From L turn clockwise (Toward Generator) up to zB + =
rm = 0.4286. Then
Z0
zB = zB +
= 0.6547
Zd
Next
zA+ =

1
zB

and
zA = zA+

= 1.5275
Zd
=1
Z0

as it should be.
To draw the plot, note that zA = 1 implies that |V (z)| is flat to the left of A since there is no
backward wave. Moreover in A+ there is a maximum of voltage and in B a minimum, since the
corresponding points are on the positive and negative part of the real axis, respectively. Then
VA = 1 V. The VSWR on AB is
S=

1 + |B |
1 + |A+ |
=
= 1.5275
1 |A+ |
1 |B |

PRELIMINARY VERSION

40

zB

rM

zB+ =rm

zA+

L
Figure B.9.

zA

Smith Chart plot relative to the design of the matching device and to the voltage plot

This number, of course, coincides with zA+ . Then

VB = Vmin =

Vmax
= 0.6547
S

Of course VB = VB + . To draw the part relative to BC, note that

VC = VC+ (1 + L ) = VB++ ejkz0 LBC (1 + L )
1 + B
VB++ = VB+
1 + B +
+
+ jkzd LAB
VB = VA + e
= jVA++
1 + A
VA++ = VA+
1 + A+
We find |VC | = 1.4333 V.
Alternatively and more rapidly, we can use the conservation of energy to compute |V (z)|. Since
the structure is lossless, the net power is the same everywhere in the circuit. In general
1
1 |V inc |2
= |V (z)|2 R{Y (z)}
2 Z0
2

PRELIMINARY VERSION

41

The left hand side is the net power in A , because the reflected power is zero. In particular
|V inc |2
|V inc |2
=
Z0 R{YL }
R{yL }
inc 2
|V
|
|V inc |2
|V inc |2
|VB |2 =
=
=
Z0 R{YB }
R{yB + }
Z0 Yd R{yB }
|VC |2 =

The resulting plot is shown in Fig. B.10

|V(z)|
1.5

1.4

1.3

1.2

1.1

A
1

0.9

0.8

0.7

2.5

1.5

0.5
z (cm)

Figure B.10.

0.5

1.5

Voltage plot

Exercise n. 3
Consider the waveguide structure of Fig. B.11. It consists of a semi-infinite rectangular waveguide,
short circuited on the left and infinite on the right, excited by a dipole antenna in B. The guide is
empty up to C and then it is completely filled with a dielectric with relative permittivity r = 4.
The guide cross section has dimensions a = 5 cm, b = 2 cm. The frequency of the source is
f = 4 GHz. Moreover LAB = 2.5 cm and LBC = 2 cm. The dipole can be represented on the
fundamental mode equivalent line by a current generator is = 0.01 A.
Compute the power that the source radiates beyond C and select LAB so as to maximize it.
Solution
Compute first the parameters of the modal lines. The waveguide has a = 5 cm, hence the critical

PRELIMINARY VERSION

42

frequency of the T E10 mode is fc0 = c/(2a) = 3 GHz. Recall

kz0 = k0

fc0
f

kzd = k0

Z0

Z0 =
1

fc0
f

Z0

Zd =

fc0
f

fc0
f

which yield
kz0 = 55.4125 rad/m
g0 = 0.1134 m
kzd = 155.3245 rad/m
gd = 0.0405 m
Z0 = 569.9704
Zd = 203.3387

r
A

is
A
Figure B.11.

kz 0 , Z 0

k zd , Z d
C

Waveguide structure with a dipole source and modal circuit for the fundamental mode

VB

ZB

is

ZB

B B+
Figure B.12.

Loads seen by the current generator

Then compute the load seen by the generator. Start with that looking to the right. The line
to the right of C is infinitely long, hence ZC = Zd . Compute
zC =

Zd
= 0.3568
Z0

PRELIMINARY VERSION

43

C =

zC 1
= 0.4741
zC + 1

B + = C ej2kz0 LBC = 0.4741ej53.0039

zB + =
and finally

1 + B +
= 1.1850 + j1.1577
1 B +

Z B + = zB + Z0 = 675.43 + j659.83

These values have been obtained by means of a computer, but can be obtained, with less accuracy,
by the Smith chart. Similarly for the load looking to the left.

A = 1

B = A ej2kz0 LAB = ej21.2549

1 + B

z B =
= j5.3293

1 B

and finally

Z B =
z B Z0 = j3037.6

The original modal circuit can be lumped into the one of Fig. B.12. The voltage at the generator
terminals is
is
VB = VB + =
= 4.3934 j6.2010 V
(B.1)

Y B + Y B+
In order to compute the power radiated beyond C we can apply the principle of energy conservation
since the structure is lossless. Then
PC = PB + =

1
|V + |2 R{ Y B + } = 0.0219 W
2 B

since Y B + = (7.5756 j7.4007) 104 S.

If we want a more detailed computation, that yields also the phase of voltage and current, we can
proceed step by step across the discontinuities:
VB++ =

VB +
= 4.4530 j3.5127 V
1 + B +

VC+ = VB++ ejkz0 LBC = 5.1307 + j2.4176 V

VC++ = VC+

1 + C
= 2.6982 + j1.2714 V
1 + C +

where C + = 0, of course. We can check that

PC =

1 |VC++ |
= 0.0219 W
2 Zd

exactly as found above by the energy method.

If we compute the power radiated to the left, we should find zero, since the waveguide is shorted
in A. Indeed

1
P B = |VB |R{ Y B } = 0
2

PRELIMINARY VERSION

44

because Y B is pure imaginary.

If we want to maximize the radiated power, it is necessary to maximize the magnitude of VB .
Since the numerator of (B.1) is fixed, we must minimize the magnitude of the denominator. Note

that Y B + is fixed and that Y B is pure imaginary and its value depends on the length LAB .

Clearly, in these conditions, the denominator has minimum amplitude if it is real. Now,
y B+ =

0.4318 j0.4218, so that we must enforce y B = j0.4218, which requires LAB /g0 = 0.3135 and
LAB = 3.5551 cm.
In ordinary applications, the waveguide is matched when looking to the right of the source, so

that
y B + = 1 and the optimum situation is reached if
y B = 0, which implies LAB = g0 /4.