Professional Documents
Culture Documents
2. Uniqueness Theorem
A solution is said to be unique if it is the only one possible among a given
class of solutions. The EM field in a given region V[ S ] is unique if
- all sources are given;
- either the tangential Etan components or the tangential H tan components
are specified at the boundary S. 1
The uniqueness theorem follows from Poynting’s theorem in its integral form:
(EH* )ds+ j ( |H|2 − |E|2 )dv+ |E|2 dv =−(EJ i* +H*M i )dv . (17.1)
S VS VS VS
1
A more general statement of the theorem asserts that any one of the following boundary conditions at S ensure the solution’s
uniqueness: (1) E tan S , or (2) H tan S , or (3) Etan1 S and H tan1 S , or (4) E tan2 S and H tan2 S . Here, Etan = E − E nˆ is the tangential
component of E at the surface S while Etan1 and E tan2 are its components. The same notations hold for H .
[N.K. Nikolova, “Electromagnetic boundary conditions and uniqueness revisited,” IEEE Antennas & Propagation Magazine,
vol. 46, no. 5, pp. 141–149, Oct. 2004.]
Nikolova 2020 1
We start with the supposition that a given EM problem has two solutions (due to
the same sources and the same boundary conditions): (Ea , H a ) and (Eb , H b ) .
The difference field is then formed:
E = E a − Eb ,
(17.2)
H = H a − Hb .
The difference field has no sources; thus, it satisfies the source-free form of
(17.1):
Since both fields satisfy the same boundary conditions, then Etan = 0 or
H tan = 0 over S, which makes the surface integral in (17.3) zero. This results in
j ( | H |2 − | E |2 )dv + | E |2 dv = 0 , (17.4)
VS VS
imaginary real
3. Equivalence Principles
The equivalence principle follows from the uniqueness theorem. It allows for
the simplification of certain EM problems. As long as a problem is re-formulated
Nikolova 2020 2
so that it preserves the boundary conditions for the original field (Eo , H o ) at S,
it is going to produce the only one possible solution for the region VS bounded
by S. Such a re-formulated problem is referred to as an equivalent problem.
(E o , H o ) (E o , H o ) (E o , H o )
VS VS VS
J se Js
(E o , H o ) (E e , H e ) no fields
S sources S no sources M se S no sources Ms
n̂ n̂ n̂
(a) Original problem (b) General equivalent (c) Equivalent problem
problem with zero fields
The equivalent problem in (b) assumes that the field inside the volume enclosed
by S is given by (Ee , H e ) , which is different from the original field (Eo , H o ) .
This results in a field discontinuity at the surface S, which demands the existence
of surface current densities (as per Maxwell’s equations):
J se = nˆ (H o − H e ),
(17.6)
M se = (Eo − Ee ) nˆ .
For the equivalent problem in (c), where (Ee , H e ) is set to zero, these surface
current densities are
J s = nˆ H o ,
(17.7)
M s = Eo nˆ .
The zero-field formulation is often referred to as Love’s equivalence principle.
We can apply Love’s equivalence principle in three different ways.
(a) We can assume that the boundary S is a perfect conductor. As per image
theory, in an equivalent open problem, this eliminates the surface electric
currents, i.e., J s = 0 , and leaves just surface magnetic currents of double
strength 2M s . Such an equivalent problem is illustrated below.
Nikolova 2020 3
(E o , H o ) (E o , H o ) no fields (E o , H o ) no fields (E o , H o )
VS S S S
n̂ n̂ n̂
sources no sources no sources
Ms 2M s
Js = 0 Js = 0
(b) We can assume that the boundary S is a perfect magnetic conductor. As per
image theory, in an equivalent open problem, this eliminates the surface
magnetic currents, i.e., M s = 0 , and leaves just surface electric currents of
double strength 2J s . This approach is illustrated below.
(E o , H o ) (E o , H o ) (E o , H o ) (E o , H o )
no fields no fields
VS
S S S
n̂ n̂ n̂
sources no sources no sources
Js 2J s
Ms = 0
Ms = 0
(c) Make no assumptions about the materials inside S, and define both J s and
M s currents, which radiate in free space (no fictitious conductors behind
them). It can be shown that these equivalent currents must lead to zero fields
inside VS . [Ewald-Oseen extinction theorem: A. Ishimaru, Electro-
magnetic Wave Propagation, Radiation, and Scattering, Prentice Hall,
1991, p. 173]
Nikolova 2020 4
The first two approaches are not very accurate in the general case of a curved
boundary surface S because the image theory can be applied to curved surfaces
only if the curvature radius is large compared to the wavelength. However, in the
case of flat infinite planes (walls), the image theory holds exactly and all three
approaches should produce the same external field according to the uniqueness
theorem.
The above approaches are used to compute fields in half-space as excited by
apertures. The field behind S is assumed known and is used to define the
equivalent surface currents. The open-region far-zone solutions for the vector
potentials A (resulting from J s ) and F (resulting from M s ) are
e− j r
A( P) =
4 r J s (r)e j rˆrds , (17.8)
S
e− j r
F( P) =
4 r M s (r)e j rˆrds . (17.9)
S
Here, r̂ denotes the unit vector pointing from the origin of the coordinate system
to the point of observation P(r). The integration point Q(r) is specified through
the position vector r . In the far zone, it is assumed that the field propagates
radially away from the antenna. It is convenient to introduce the propagation
vector or wave vector,
β = rˆ , (17.10)
which characterizes both the phase constant (wave number) and the direction of
propagation of the wave. The vector potentials can then be written as
e− j r
A( P) =
4 r J s (r)e jβ(r)rds , (17.11)
S
e− j r
F( P) =
4 r M s (r)e jβ(r)rds . (17.12)
S
The relations between the far-zone field vectors and the vector potentials are
E far = − j ( A θˆ + A φˆ ) , (due to A only)
A (17.13)
H Ffar = − j ( F θˆ + F φˆ ) , (due to F only). (17.14)
Nikolova 2020 5
Since
E Ffar = H Ffar rˆ , (17.15)
the total far-zone electric field (due to both A and F) is found as
E far = E Afar + E Ffar = − j ( A + F )θˆ + ( A − F )φˆ . (17.16)
Equation (17.16) involves both vector potentials as arising from both types of
surface currents. Computations are reduced in half if image theory is used in
conjunction with an electric or magnetic wall assumption.
Nikolova 2020 6
e− j r
F( P) = −
4 r nˆ Ea (r) e jβ(r)rds . (17.19)
SA
We can work with the general vector expression for the far field E [see (17.16)]
written as
E far = − j A − j F rˆ , (17.20)
where the longitudinal Ar component is to be neglected. Substituting (17.18)
and (17.19) into (17.20) yields
e− j r
E far (r ) = − j rˆ nˆ Ea (r) − rˆ ( nˆ H a (r ) ) e jβ (r)r ds . (17.21)
4 r SA
This is the full vector form of the radiated field resulting from the aperture field,
and it is referred to as the vector diffraction integral (or vector Kirchhoff
integral).
We now consider a practical case of a flat aperture lying in the xy plane with
nˆ zˆ . Now, (17.18) and (17.19) are written as
e− j r
A( P) = zˆ Η a (r) e jβ (r)r ds , (17.22)
4 r SA
e− j r
F( P) = − zˆ Ea (r) e jβ (r)r ds . (17.23)
4 r SA
For brevity, the radiation integrals in (17.22) and (17.23) are denoted as
I H = I xH xˆ + I yH yˆ = H a e jβr ds , (17.24)
SA
Then,
e− j r
A= (− I yH xˆ + I xH yˆ ) , (17.26)
4 r
e− j r
F = − (− I yE xˆ + I xE yˆ ) . (17.27)
4 r
Nikolova 2020 7
The integrals in the above expressions can be explicitly written for the case nˆ zˆ
in spherical coordinates, bearing in mind that the source-point position is
r = xxˆ + yyˆ :
I xE ( , ) = Eax ( x , y )e j ( x sin cos + y sin sin ) dx dy , (17.28)
SA
Note that the above integrals can be viewed as 2-D Fourier transforms of the
aperture field components where x transforms into x = − sin cos and y
transforms into y = − sin sin .
The transverse components of the magnetic vector potential A in spherical
terms are obtained from (17.26) as
e− j r
A = ( − I yH cos cos + I xH cos sin ) , (17.32)
4 r
e− j r H
A = ( I y sin + I xH cos ) , (17.33)
4 r
which can also be written in the vector form:
e− j r ˆ
A⊥ = θ cos ( I xH sin − I yH cos ) + φˆ ( I xH cos + I yH sin ) . (17.34)
4 r
Analogously,
e− j r ˆ
F⊥ = − θ cos ( I xE sin − I yE cos ) + φˆ ( I xE cos + I yE sin ) . (17.35)
4 r
By substituting the above expressions in (17.16), we obtain the far-zone E field:
e− j r E
E = j I x cos + I yE sin + cos ( I yH cos − I xH sin ) , (17.36)
4 r
Nikolova 2020 8
e− j r
E = j - ( I xH cos + I yH sin ) + cos ( I yE cos − I xE sin ) . (17.37)
4 r
For apertures illuminated from open space (e.g., reflector antennas), the dual
current formulation is used. Then, the usual assumption is that the aperture field
resembles that of a locally-plane wave, i.e.,
H a = zˆ Ea / . (17.41)
This implies that
1 I yE I xE
IH = zˆ I E or I xH =− , I yH = . (17.42)
This assumption is valid for apertures that are at least a couple of wavelengths in
extent where the reflector is in the far zone of the primary illuminating antenna.
Then, (17.36)-(17.37) reduce to
e− j r (1 + cos ) E
E ( , ) = j ( I x cos + I yE sin ) , (17.43)
4 r 2
e− j r (1 + cos ) E
E ( , ) = j ( I y cos − I xE sin ) . (17.44)
4 r 2
Nikolova 2020 9
Compare (17.43)-(17.44) to (17.39)-(17.40). The terms in the brackets are
identical. If the aperture has high gain, the factors containing cos are not going
to affect the pattern significantly and the two sets of formulas are going to be
nearly equivalent.
Ea
Ly
x
Lx
If the field is uniform in amplitude and phase across the aperture, it is referred to
as a uniform rectangular aperture. Let us assume that the aperture field is y-
polarized:
Lx Ly
Ea = E0 yˆ , for | x | and | y | ,
2 2 (17.45)
Ea = 0, elsewhere .
Using the equivalence principle, let us assume an electric wall at z = 0 , where
the equivalent magnetic current density is given by M s ,e = E0 nˆ . Applying
image theory, we double the equivalent source radiating in open space:
M s = 2M s ,e = 2 E0 yˆ zˆ = 2 E0xˆ . (17.46)
The only non-zero radiation integral is [see (17.29)]
Nikolova 2020 10
Lx /2 Ly /2
I yE ( , ) = 2 E0 e j x sin cos dx e j y sin sin dy , (17.47)
− Lx /2 − Ly /2
Nikolova 2020 11
sin ( 0.5 Lx sin )
F ( ) = E ( ) = cos , E = 0 (17.53)
( 0.5 Lx sin )
60 60
120 120
150 150
180
For electrically large apertures, the main beam is narrow and the
(1 − sin 2 cos 2 )1/2 in (17.51) is negligible, i.e., it is roughly equal to 1 for all
observation angles within the main beam. That is why, in the theory of large
apertures and arrays, it is assumed that the amplitude pattern is
Nikolova 2020 12
sin u sin v
f (u, v) , (17.54)
u v
where u = 0.5 Lx sin cos and v = 0.5 Ly sin sin ; see (17.49).
Below is a view of the | (sin u ) / u | function for Lx = 20 and = 0 (H-plane
pattern):
|sin[20*pi*sin(theta)]/[20*pi*sin(theta)]|
1
0.8
0.6
0.4
0.2
0
-1 -0.5 0 0.5 1
sin(theta)
Nikolova 2020 13
Here is a view of the | sin v / v | function for Ly = 10 and = 90 (E-plane
pattern):
0.8
0.6
0.4
0.2
0
-1 -0.5 0 0.5 1
sin(theta)
Notice that the side-lobe level in both patterns is the same regardless of the fact
that the size of the aperture is different in the x and y directions ( Lx = 20 ,
Ly = 10 ). This is due to the first minor maximum of the function
| sin x / x | 0.2172 reached at x 4.494 . The value of this maximum does not
depend on the size of the aperture as long as this size exceed a wavelength.2
2
The first minor maximum of the sinc function’s absolute value is reached when its argument solves the transcendental equation
tan x = x .
Nikolova 2020 14
Point for Discussion: The field of a narrow slot (a slit) ( Ly ).
y L
x
z
Ea = E0 yˆ
The radiation integral for the case of a slit is a particular case of (17.48):
sin ( 0.5 Lx sin cos ) sin ( 0.5 Ly sin sin )
( I yE )slit =2 E0 Lx Ly lim , (17.55)
( 0.5 Lx sin cos ) Ly →0
( 0.5 Ly sin sin )
1
which leads to
sin ( 0.5 Lx sin cos )
( I yE )slit = 2E0 Lx Ly . (17.56)
(0.5 Lx sin cos )
The total field pattern of the slit is then
sin ( 0.5 Lx sin cos )
F ( , ) = 1 − sin 2 cos 2 (17.57)
0.5 Lx sin cos
The principal plane patterns are:
E-plane pattern ( = / 2) - omnidirectional (like a dipole)!
F ( ) = E = 1 (17.58)
H-plane pattern ( = 0)
sin ( 0.5 Lx sin )
F ( ) = E =| cos | (17.59)
( 0.5 Lx sin )
Nikolova 2020 15
Beamwidths
(a) First-null beamwidth (FNBW)
We need the locations of the first nulls in the pattern in order to calculate the
FNBW. The nulls of the E-plane pattern are determined from (17.52) as
Ly
sin / =n = n , n = 1,2, , (17.60)
2
n
n = arcsin , rad. (17.61)
y
L
The first null occurs at n = 1 .
FNBWE = 2 n = 2arcsin , rad. (17.62)
Ly
In a similar fashion, FNBWH is determined to be
FNBWH = 2arcsin , rad. (17.63)
Lx
It is apparent that larger aperture widths lead to narrower beams.
Nikolova 2020 16
A first-order approximation is possible for very small arguments in (17.67), i.e.,
when Ly 0.443 (large aperture):
HPBWE 0.886 . (17.68)
Ly
The half-power beamwidth in the H-plane is analogous:
0.443
HPBWH 2arcsin . (17.69)
Lx
Side-lobe level
It is obvious from the properties of the | sin x / x | function that the first side lobe
has the largest maximum of all side lobes, and it is
sin 4.494
| E ( = s ) | 0.217 −13.26 , dB. (17.70)
4.494
When evaluating side-lobe levels and beamwidths in the H-plane, one has to
include the cos factor. The larger the aperture, the less important this factor is.
Directivity
The antenna solid angle A is needed to calculate the directivity from
D0 = 4 / A . (17.71)
The radiation intensity in any direction can be expressed through the normalized
field pattern as
U ( , ) = U max | F ( , ) |2 . (17.72)
The far-field pattern | F ( , ) | is available from (17.51), namely,
sin u sin v
| E |= F ( , ) = 1 − sin 2 cos 2 . (17.73)
u v
The antenna solid angle is then calculated as
2 /2
A = | F ( , ) |2 sin d d , (17.74)
0 0
Nikolova 2020 17
However, in the case of an aperture illuminated by a TEM wave, we can use
a simpler approach. Generally, for slot and reflector (dish) antennas, the
assumption of a TEM wave at the aperture is quite accurate. Then, if E = yˆ E0 ,
H a = −xˆ E0 / , (17.75)
where is the intrinsic impedance of the medium. Analogous expression is used
for an open-end waveguide antennas where is replaced by the waveguide’s
wave impedance Z w . The far-field components in this case were already derived
in (17.43) and (17.44). They lead to the following radiation intensity:
2
U ( , ) = (1 + cos )2 (| I xE ( , ) |2 + | I yE ( , ) |2 ) . (17.76)
32
2
The maximum value of the function in (17.76) is derived after substituting the
radiation integrals from (17.28) and (17.29), which leads to
2 2
U max = 2
8 S A
Ea ds . (17.77)
The integration of the radiation intensity (17.76) over a closed sphere is not
easy. It can be avoided by observing that the total power reaching the far zone
must have passed through the aperture in the first place. In an aperture, where the
field obeys (17.75), this power is determined as
1
= Pav ds =
2 S A
| Ea |2 ds . (17.78)
S
Nikolova 2020 19
sin ( 0.5 Lx sin cos ) sin ( 0.5 Ly sin sin )
I yE ( , ) = 2 E0 Lx Ly . (17.85)
(0.5 Lx sin cos ) ( 0.5 Ly sin sin )
The far-field components are obtained by substituting (17.85) into (17.43) and
(17.44):
E = C sin
(1 + cos ) sin u sin v ,
2 u v
(17.86)
E = C cos
(1 + cos ) sin u sin v ,
2 u v
where
e− j r
C = j Lx Ly E0 ,
2 r
u = 0.5 Lx sin cos ,
v = 0.5 Ly sin sin .
The far-field expressions in (17.86) are very similar to those of the aperture
mounted on a ground plane, see (17.50). For small values of , the patterns of
both apertures are practically identical.
An exact analytical evaluation of the directivity is difficult. However,
according to the approximations made, the directivity formula derived in (17.79)
should provide accurate enough value. According to (17.79), the directivity is the
same as in the case of the aperture mounted on a ground plane.
Nikolova 2020 20
equivalent sources distribution from a maximum at the aperture’s center to zero
values at its edges.
One practical aperture of tapered source distribution is the open rectangular
waveguide. The dominant TE10 mode has the following distribution:
− Lx / 2 x Lx / 2
Ea = yˆ E0 cos x , (17.87)
Lx − Ly / 2 y Ly / 2
Ey x
The general procedure for the far-field analysis is the same as before (Sections 5
and 6). The only difference is in the field distribution. Again, only the integral
I yE ( , ) is evaluated:
Ly /2
Lx /2
I yE ( , ) = 2 E0 cos x e j x sin cos dx e j y sin sin dy . (17.88)
− Lx /2 Lx − L y /2
cosine distribution along x constant distribution along y
Nikolova 2020 21
j x sin cos
Lx /2
I x ( , ) = Lx x e
cos dx =
− Lx /2
Lx /2
= Lx x cos ( x sin cos ) + j sin ( x sin cos ) dx =
cos
− Lx /2
1 x
L /2
2 − Lx /2 Lx
=
cos − sin cos
x + cos + sin cos x dx +
x
L
j x
L /2
2 − Lx /2
+ sin sin cos −
x + cos sin cos + x dx
Lx Lx
Lx
cos sin cos
Lx 2
I x ( , ) = (17.90)
2 2 Lx
− sin cos
2 2
Nikolova 2020 22
where
e− j r
C = j Lx Ly E0 ,
2 r
u = 0.5 Lx sin cos ,
v = 0.5 Ly sin sin .
E-plane ( = 90 ):
Ly
sin sin
F ( ) = E = 2 (17.93)
Ly
2 sin
H-plane ( = 0 ):
L
cos x sin
F ( ) = E ( ) = cos 2 . (17.94)
Lx
2 2
sin −
2 2
Nikolova 2020 23
H-PLANE PATTERN – UNIFORM VS. TAPERED ILLUMINATION ( Lx = 3 ):
0
uniform
30 30 tapered
60 60
120 120
150 150
180
The lower SLL of the tapered-source pattern is obvious. It is better seen in the
rectangular plot given below. The price to pay for the lower SLL is the decrease
in directivity (the beamwidth of the major lobe increases).
Nikolova 2020 24
H-PLANE PATTERN RECTANGULAR PLOT – UNIFORM VS. TAPERED ILLUMINATION
WHEN Lx = 3
tapered
1 uniform
0.9
0.8
H-plane amplitude pattern
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Nikolova 2020 25
0 H-plane
30 30 E-plane
60 60
120 240
150 210
180
In the plot above, the polar patterns are shown for an X-band waveguide of cross-
section defined by Lx = 2.286 cm, Ly = 1.016 cm. The frequency f 0 = 9.84 GHz
is considered when the free-space wavelength is 0 = 3.048 cm.
The case of a dominant-mode open-end waveguide radiating in free space can
be analyzed following the approaches outlined in this Section and in Section 6.
The calculation of the beamwidths and the directivity is analogous to the previous
cases. Only the final results will be given here for the case of the x-tapered
(cosine taper) aperture on a ground plane.
4 8
Directivity: D0 = Lx Ly (17.95)
2 2
8
Effective area: Aeff = Lx Ly 0.81 Ap (17.96)
2
Note the decrease in the directivity and the effective area compared to the
uniform-aperture case.
Nikolova 2020 26
Half-power beamwidths:
50.6
HPBWE = , deg. (= HPBWE of the uniform aperture) (17.97)
Ly /
68.8
HPBWH = , deg. (> HPBWH of the uniform aperture) (17.98)
Lx /
The above results are approximate. Better results are obtained if the following
factors are taken into account:
• the phase constant of the waveguide g and its wave impedance Z w are
not equal to the free-space phase constant 0 = 0 0 and intrinsic
impedance Z 0 = 0 / 0 ; they are dispersive;
• the abrupt termination at the waveguide open end introduces reflection,
which affects the field at the aperture;
• there are strong fringe currents at the waveguide walls, which contribute to
the overall radiation.
Nikolova 2020 27