Lecture-6

6.3.6 N-Element Linear Array: Uniform Spacing, Nonuniform Amplitude

Uniform amplitude arrays produce small half-power beamwidth and possess the largest directivity. But in
certain instances the sidelobe level of the radiation pattern has to be maintained at a desired level. The
sidelobe level can be reduced by varying the amplitude excitations of the array elements. Nonuniform
amplitude excitations of a linear antenna array produce a pattern with smaller sidelobe level and a slightly
increased half power beamwidth in comparison to the uniform linear antenna array.

In this section, we will discuss arrays with uniform spacing but nonuniform amplitude distribution. Often,
the broadside arrays are classified according to the type of their excitation amplitude. The categories are:

(a) Uniform amplitude array: Relatively high directivity, but the side-lobe levels are high;

(b) Dolph–Chebyscheff array: For a given number of elements, its maximum directivity is next to that of
the uniform array. Side-lobe levels are the lowest in comparison with the other two types of arrays for
a given directivity;

(c) Binomial array: Does not have good directivity but has very low side-lobe levels (when the element
spacing is equal or less than λ/2, there are no side lobes at all).

Array Factor Analysis for Non-uniform Array

A simplification to the array factor can be made by symmetrically placing the array elements. Consider an
array with an even number of isotropic elements (2M, where M is an integer) and suppose these are
located symmetrically about the origin of the coordinate system (defined as the z-axis in this case, which
is also the physical center of the array). The separation between the elements is d.

aM a2 a1 a1 a2 aM

Z
d d/2 d/2 d

Fig.6.16 Nonuniform amplitude array of even number of elements.

Assuming that the amplitude excitation is symmetrical about the origin, the array factor for a nonuniform
amplitude, broadside array can be written as:

kd cos  3 kd cos  ( 2 M 1) kd cos  kd cos 

 j( )  j( ) ( )  j( )
( AF ) 2 M  a1e 2
 a2 e 2
   a M e 2
 a1e 2

3 kd cos  ( 2 M 1) kd cos 
 j( )  j( )
a2 e 2
   a M e 2

M
(2n  1)
or ( AF ) 2 M  2 a n cos[ kd cos  ]
n 1 2
In the normalized form, it reduces to
M
(2n  1)
( AF ) 2 M   a n cos[ kd cos  ] (6.46)
n 1 2
where an' s are the excitation coefficients of the array element.

Similarly, for odd number of elements (2M+1), the array factor can be written as
M 1
( AF ) 2 M 1  a
n 1
n cos[( n  1) kd cos  ] (6.47)

The amplitude excitation of the center element is 2a1, in this case.

6.3.7 Binomial Array:

The binomial array was investigated and proposed by J. S. Stone to synthesize patterns without side lobes.
Let us first consider a 2–element array with equal current amplitudes and spacing, the array factor is given
by

AF 1  e j (6.48)

For a broadside array ( = 0) with element spacing d less than one-half wavelength, the array factor has
no sidelobes. This can be proved in the following way:

 1  cos 2  sin 2   21  cos   4 cos 2  

2
AF
 2

where ψ = kd cos. The first null of this array factor can be obtained as:

1 2    
  d cos  n     n   cos 1  
2  2  2d 

As long as the d < λ/2, the first null does not exist. If d = λ/2, then null will be at  = 00 and 1800. Thus, in
the “visible” range of , all secondary lobes are eliminated.

An array formed by taking the product of two arrays of this type gives:

  
AF  1  e j 1  e j  1  2e j  e j 2 (6.49)

This array factor, being the square of an array factor with no sidelobes, will also has no sidelobes.
Mathematically, the array factor above represents a 3-element equally-spaced array driven by current
amplitudes with ratios of 1:2:1. In a similar fashion, equivalent arrays with more elements may be
formed.

2-element AF 1  e j

3-element AF  1  e j 
2
 1  2e j  e j 2
 
4-element AF  1  e j 3
 1  3e j  3e j 2  e j 3

Similarly for N-element the array factor can be expressed as

N-element AF  1  e j  
N 1
(6.50)

If d ≤ λ/2, the above AF does not have side lobes regardless of the number of elements N. The excitation
amplitude distribution can be obtained easily by the expansion of the binome in (6.50). Making use of
Pascal’s triangle, this can be given by:

1 1

1 2 1

1 3 3 1

1 4 6 4 1

1 5 10 10 5 1

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

The relative excitation amplitudes at each element of an (N+1) element array can be determined from this
traiangle. An array with a binomial distribution of the excitation amplitudes is called a binomial array.


AF  1  e j 
N 1
 1  N  1e j 
N  1N  2 e j 
N  1N  2N  3 e j 3  (6.51)
2! 3!

The excitation distribution as given by the binomial expansion gives the relative values of the amplitudes.
It is immediately seen that there is too wide variation of the amplitude, which is a disadvantage of the
binomial arrays. The overall efficiency of such an antenna would be low. Besides, the binomial array has
a relatively wide beam. Its HPBW is the largest as compared to the uniform or the Dolph–Chebyshev
array.

An approximate closed-form expression for the HPBW of a binomial array with d = λ/2 is

1.06 1.06 1.75

HPBW    (6.52)
N 1 2L /  L/

where L = (N-1)d is the array length.

The directivity of a broadside binomial array with spacing d = λ/2 can be calculated as:

2
D0   2 N 1
  
cos cos  
0
 2 
d
D0 
2 N  2  2 N  4    2
2 N  3  2 N  5    1

D0  1.77 N  1.77 1  2 L /  (6.52)

The array factor of a 10 element broadside binomial array (N = 10) is shown below.

(a) d = λ/2 (b) d = 3λ/4

Fig.6.17 Radiation pattern for 10- element broadside binomial array

 Lecture-7
6.3.8 Dolph–Chebyshev array:
Dolph proposed (in 1946) a method to design arrays with any desired sidelobe levels and any HPBWs.
This method is based on the approximation of the pattern of the array by a Chebyshev polynomial of
order m, high enough to meet the requirement for the side-lobe levels.

Chebyshev Polynomials: The Chebyshev polynomial of order m is defined by

 
 1m cosh m  cos 1  z  , z  1


 
Tm z   cos m  cos 1 z  ,1  z  1, (6.53)

 
cosh m  cos z  , z  1
1

A Chebyshev polynomial Tm(z) of any order m can be derived from a recursion formula, provided Tm-1(z)
and Tm-2(z) are known, in the following way:

Tm z   2 zTm1 z   Tm2 z  (6.54)

From the above equation we can express the Chebyshev polynomials as:

m  0, T0 z   1
m  1, T1 z   z
m  2, T2 z   2 z 2  1
(6.55)
m  3, T3 z   4 z 3  3 z
m  4, T4 z   8 z 4  8 z 2  1
m  5, T5 z   16 z 5  20 z 3  5 z

The array factor expressed in (6.46) and (6.47) is the summation of cosine terms and the largest
harmonics of the cosine terms is one less than the total number of elements in the array. Each cosine term
having an argument which is the interger multiple of the fundamental frequency can be expressed as a
series of cosine functions with the fundamental frequency as the argument. i.e.

m=0 cos(mu)=1

m=1 cos(mu)=cos u (6.56)

m=2 cos(mu)= cos 2u= 2cos2u-1

m=3 cos(mu)=cos 3u= 4cos3u – 3cos u

If the -1 ≤ z ≤ 1, the Chebyshev polynomials are related to the cosine functions. So by comparing
equation (6.55) and (6.56) we see that the Chebyshev argument z is related to cosine argument u by

z  cos u  u  cos 1 z

e.g. cos(2u) can be expressed as

    
cos 2 cos 1 z  2 cos cos 1 z 
2
 1  2 z 2  1  T2 z 

Similarly:

m  0, cosmu   T0 z   1
m  1, cosmu   cosu   T1 z   z
m  2, cosmu   cos2u   T2 z   2 z 2  1
(6.57)
m  3, cosmu   cos3u   T3 z   4 z 3  3 z
m  4, cosmu   cos4u   T4 z   8 z 4  8 z 2  1
m  5, cosmu   cos5u   T5 z   16 z 5  20 z 3  5 z

Properties of the Chebyshev polynomials:

1) All polynomials of any order m pass through the point (1, 1).
2) Within the range -1 ≤ z ≤ 1, the polynomials have values within [–1, 1].
3) All nulls occur within -1 ≤ z ≤ 1.
4) The maxima and minima in the z   1, 1] range have values +1 and –1, respectively.
5) The higher the order of the polynomial, the steeper the slope for z  1.

Fig. 6.18 Plot of Chebyshev polynomials

 Lecture-8
6.3.9 Chebyshev Array Design
In this case, the main goal is to approximate the desired AF with a Chebyshev polynomial such that
 The side-lobe level meets the requirements, and
 The main beam width is as small as possible.

An array of N elements has an AF approximated with a Chebyshev polynomial of order m, which is

always m = N-1.
where N = 2M, if N is even; and N = 2M+1, if N is odd.

Statement: Design a broadside Dolph-Tschebyscheff array of N elements with spacing d between the
elements. The sidelobes are R0 dB below the maximum of the major lobe. Find the excitation co-efficients
and form the array factor.

Procedure:

1. Select the appropriate AF for the total number of elements N.

(2n  1)
M
( AF ) 2 M  a
n 1
n cos[
2
kd cos  ] N = 2M

M 1
( AF ) 2 M 1  a
n 1
n cos[(n  1)kd cos  ] N = 2M+1

2. Replace each cos(mu) term in the array factor by its appropriate expansion in terms of powers of
cos(u).

3. Determine the point z = z0 such that Tm= R0 (Voltage ratio). Find z0 such that


R0  TN 1 ( z 0 )  cosh N  1 cosh 1 z 0 
 cosh 1 R0 
z 0  cosh  
 N  1 

Substitute cos u  
z
4. in the array factor of step 2. This substitution normalizes the array factor
z0
sidelobes to a peak value of unity.

5. Equate the array factor of step 4 to TN-1 (z) and determine the array coefficients.

Example-6: Design a Dolph–Chebyshev array (broadside) of N = 10 elements

with a major-to-minor lobe ration of R0=26dB. Find the excitation coefficients.
Solution:
The order of the Chebyshev polynomial is m = N-1 = 9. The AF for an even-number array is:
5
AF2 M   a cos  2n  1 u 
n 1
n

d
u cos , M  5.

Step 1: The array factor for 10 element array :
AF10  a1 cos u  a2 cos3u  a3 cos5u  a4 cos 7u  a5 cos9u

Expand the cos(mu) terms of powers of cos u :

cos3u  4cos3 u  3cos u ,

cos5u  16cos5 u  20cos3 u  5cos u ,
cos 7u  64cos7 u  112cos5 u  56cos3 u  7 cos u ,
cos9u  256cos9 u  576cos7 u  432cos5 u  120cos3 u  9cos u

Step 2: Determine z0 :
R0  26 dB
26
 R0  10 20 20
 T9 ( z0 )  20

cosh 9cosh 1  zo    20,

 
z0  cosh 0.41  z0  1.08515

Step 3: Express the AF from Step 1 in terms of cos u  z / z0

z
AF10   a1  3a2  5a3  7a4  9a5 
z0
z3
  4a2  20a3  56a4  120a5 
z03
z5
 16a3  112a4  432a5 
z05
z7
  64a4  576a5 
z07
z9
  256a5  
z09
 9
z  120 z 3  432 z 5  576 z 7  256
 z9
T9 ( z )

Step 4: Finding the coefficients by matching the power terms:

256a5  256 z09  a5  2.0860
64a4  576a5  576 z07  a4  2.8308
16a3  112a4  432a5  432 z05  a3  4.1184
4a2  20a3  56a4  120a5  120 z07  a2  5.2073
a1  3a2  5a3  7 a4  9a5  9 z09  a1  5.8377

Normalize coefficients with respect to edge element ( N  5) :

a5  1; a4  1.357; a3  1.974; a2  2.496; a1  2.789

 Leccture--9
6.4 Plan
nar array
A planar array providees a large apeerture and maay be used forr directional bbeam controllled by varyinng the
hase of each element, sym
relative ph mmetrical patteerns with low
w side lobes, mmuch higher directivity (nnarrow
main beam m) than that of
o their individ
dual.

Array Factor
F

Fig.6.19 Planar arraay

If M elem
ments are initiaally placed alo
ong x-axis, th
he array factoor can be exprressed as

M
AFx1  I
m 1
m1e
j  m 1 kd x sin  cos    x 
((6.58)

It is assum med that all elements aree equispaced with an inteerval of d x annd a progresssive shift  x . I m1
denotes th he excitation coefficient of o the elemen nt with coordiinates x   m  1 d x , y  0. , w
nt at the point where
sin  cos   cos  x is the directional
d coosine with resspect to the xx-axis (  x is tthe angle betw
ween r and the x-
th st
axis). In the
t figure abo he element off the m row and the 1 coolumn of the array matrixx. If N
ove, this is th
such array ys are placed d at even interrvals along thhe y-directionn, a rectangullar array is foormed. We asssume
again thatt they are equ a there is a progressive pphase shift  y along eachh row.
uispaced at a distance d y and
We also assume that the normalized current distribution along each of the x-directed arrays is the same
but the absolute values correspond to a factor of I1n  n  1,..., N  . Then, the AF of the entire M×N array is
M  j  n 1 kd sin  sin   
 y y
N
AF   
I1n 

I m1e   x
j m 1 kd sin  cos   x 
e
 (6.59)
n 1  m 1 
or
AF  SxM  S yN

M
S xM  AFx1  I
m 1
m1e
j  m 1 kd x sin  cos   x 

(6.60)
 
N
j  n 1 kd y sin  cos   y
S yN  AF1 y  I
n 1
1n e

In the array factors above

sin  cos   xˆ  rˆ = cos x
sin  cos   yˆ  rˆ = cos y (6.61)

Thus, the pattern of a rectangular array is the product of the array factors of the linear arrays in the x and
y-directions.

In the case of a uniform planar rectangular array, I m1  I1n  I 0 for all m and n, i.e., all elements have the
same excitation amplitudes. Thus,

 
M N

e 
j n 1 kd y sin  sin    y
AF  I 0 
m 1
e   x
j m 1 kd sin  cos    x 

n 1
(6.62)

The normalized array factor is obtained as

   x    sin  N  y  
 1 sin  M    
 2    1  2 
AFn  ,     (6.63)
M    N  y  
 sin  x    sin   
  2     2  

where
 x  kd x sin  cos    x
 y  kd y sin  cos    y

The major lobe (principal maximum) and grating lobes of the terms
  
sin  M x 
S xM 
1  2 
M   (6.64)
sin  x 
 2 
  y 
sin  N 
S yN 
1  2 
N  y  (6.65)
sin  
 2 
are located at angles such that
kd x sin  m cos m   x  2m , m  0,1,...,
kd y sin  n sin n   y  2n , n  0,1,..., (6.66)

The principal maximum corresponds to m = 0, n = 0.

In general,  x and  y are independent of each other. But, if it is required that the main beams of
S xM and S yN intersect (which is usually the case), then the common main beam is in the direction:
   0 and   0 , m  n  0 .

If the principal maximum is specified by 0 ,0  , then the progressive phases  x and  y must satisfy
 x  kd x sin 0 cos 0 (6.67)

 y   kd y sin  0 sin 0 . (6.68)

When  x and  y are specified, the direction of the main beam can be found by simultaneously solving
(6.67) and (6.68)
 yd x
tan 0  (6.69)
 xd y

2
  y 
2
 
sin 0    x     (6.70)
 kd x   kd y 


The grating lobes can be located by substituting (6.67) and (6.68) in (6.66)

sin 0 sin 0  n
dy
tan mn 
sin 0 sin 0  m (6.71)
dx

sin 0 cos 0  m sin 0 sin 0  n

dx dy
sin  mn   (6.72)
cos mn sin mn

To avoid grating lobes, the spacing between the elements must be less than   d y   and d y   . In order a
true grating lobe to occur, both equations (6.71) and (6.72) must have a real solution mn ,mn  .