Unit –II

Antenna Array
The study of a single small antenna indicates that the radiation fields are uniformly
distributed and antenna provides wide beam width, but low directivity and gain. For
example, the maximum radiation of dipole antenna takes place in the direction normal to its
axis and decreases slowly as one moves toward the axis of the antenna. The antennas of
such radiation characteristic may be preferred in broadcast services where wide coverage is
required but not in point to point communication. Thus to meet the demands of point to point
communication, it is necessary to design the narrow beam and high directive antennas, so
that the radiation can be released in the preferred direction. The simplest way to achieve
this requirement is to increase the size of the antenna, because a larger-size antenna leads
to more directive characteristics. But from the practical aspect the method is inconvenient as
antenna becomes bulky and it is difficult to change the size later. Another way to improve
the performance of the antenna without increasing the size of the antenna is to arrange the
antenna in a specific configuration, so spaced and phased that their individual contributions
are maximum in desired direction and negligible in other directions. This way particularly, we
get greater directive gain. This new arrangement of multi-element is referred to as an array
of the antenna. The antenna involved in an array is known as element. The individual
element of array may be of any form (wire. dipole. slot, aperture. etc.). Having identical
element in an array is often simpler, convenient and practical, but it is not compulsory. The
antenna array makes use of wave interference phenomenon that occurs between the
radiations from the different elements of the array. Thus, the antenna array is one of the
methods of combining the radiation from a group of radiators in such a way that the
interference is constructive in the preferred direction and destructive in the remaining
directions. The main function of an array is to produce highly directional radiation. The field
is a vector quantity with both magnitude and phase. The total field (not power) of the array
system at any point away from its centre is the vector sum of the field produced by the
individual antennas. The relative phases of individual field components depend on the
relative distance of the individual clement and in turn depend on the direction.

ARRAY CONFIGURATIONS
Broadly, array antennas can be classified into four categories:
(a) Broadside array
(b) End-fire array
(c) Collinear array

Page 1
(d) Parasitic array
Broadside Array- This is a type of array in which the number of identical elements
is placedon a supporting line drawn perpendicular to their respective axes.
Elements are equally spaced and fed with a current of equal magnitude and all in same
phase. The advantage of this feed technique is that array fires in broad side direction (i.e.
perpendicular to the line of array axis, where there are maximum radiation and small
radiation in other direction). Hence the radiation pattern of broadside array is bidirectional
and the array radiates equally well in either direction of maximum radiation. In Fig. 1 the
elements are arranged in horizontal plane with spacing between elements and radiation is
perpendicular to the plane of array (i.e. normal to plane of paper.) They may also be
arranged in vertical and in this case radiation will be horizontal. Thus, it can be said that
broadside array is a geometrical arrangement of elements in which the direction of
maximum radiation is perpendicular to the array axis and to the plane containing the array
clement. Radiation pattern of a broad side array is shown in Fig. 2. The bidirectional pattern
of broadside array can be converted into unidirectional by placing an identical array behind
0
this array at distance of λ/4 fed by current leading in phase by 90 .

Fig. 1 Geometry of broadside array Fig. 2 Radiation pattern of broadside array

End Fire Array-The end fire array is very much similar to the broadside array from thepoint
of view of arrangement. But the main difference is in the direction of maximum radiation. In
broadside array, the direction of the maximum radiation is perpendicular to the

axis of array; while in the end fire array, the direction of the maximum radiation is
along the axis of array.

Fig. 3 End fire array

Thus in the end fire array number of identical antennas are spaced equally along a

Page 2
line. All the antennas are fed individually with currents of equal magnitudes but their
phases vary progressively along the line to get entire arrangement unidirectional
finally. i.e. maximum radiation along the axis of array.
Thus end fire array can be defined as an array with direction of maximum radiation
coincides with the direction of the axis of array to get unidirectional radiation.

Collinear Array-In collinear array the elements are arranged co-axially, i.e., antennas
areeither mounted end to end in a single line or stacked over one another. The collinear
array is also a broadside array and elements are fed equally in phase currents. But the
radiation pattern of a collinear array has circular symmetry with its main lobe
everywhere normal to the principal axis. This is reason why this array is called
broadcast or Omni-directional arrays. Simple collinear array consists of two elements:
however, this array can also have more than two elements (Fig. 4). The performance
characteristic of array does not depend directly on the number of elements in the array.
For example, the power gain for collinear array of 2, 3, and 4 elements are respectively
2 dB, 3.2 dB and 4.4 dB respectively. The power gain of 4.4 dB obtained by this array is
comparatively lower than the gain obtained by other arrays or devices. The collinear
array provides maximum gain when spacing between elements is of the order of 0.3λ to
0.5λ; but this much spacing results in constructional and feeding difficulties. The
elements are operated with their ends are much close to each other and joined simply
by insulator.

Fig. 4 (a) Vertical collinear antenna array (b) Horizontal collinear antenna array
Increase in the length of collinear arrays increases the directivity: however, if the
number of elements in an array is more (3 or 4), in order to keep current in phase in all
the elements, it is essential to connect phasing stubs between adjacent elements. A

Page 3
collinear array is usually mounted vertically in order to increase overall gain and
directivity in the horizontal direction. Stacking of dipole antennas in the fashion of
doubling their number with proper phasing produces a 3 dB increase in directive gain.
Parasitic Arrays-In some way it is similar to broad side array, but only one element is
feddirectly from source, other element arc electromagnetically coupled because of its
proximity to the feed element. Feed element is called driven element while other elements
are called parasitic elements. A parasitic element lengthened by 5% to driven element act
as reflector and another element shorted by 5% acts as director. Reflector makes the
radiation maximum in perpendicular direction toward driven element and direction helps in
making maximum radiation perpendicular to next parasitic element. The simplest parasitic
array has three elements: reflector, driven element and director, and is used, for example in
Yagi-Uda array antenna. The phase and amplitude of the current induced in a parasitic
element depends upon its tuning and the spacing between elements and driven element to
which it is coupled. Variation in spacing between driven element and parasitic elements
changes the relative phases and this proves to be very convenient. It helps in making the
radiation pattern unidirectional. A distance of λ/4 and phase difference of π/2 radian
provides a unidirectional pattern. A properly designed parasitic array with spacing 0.1λ to
0.15λ provides a frequency bandwidth of the order of 2%, gain of the order of 8 dB and FBR
of about 20 dB. It is of great practical importance, especially at higher frequencies between
150 and 100 MHz, for Yagi array used for TV reception.

The simplest array configuration is array of two point sources of same polarization and
separated by a finite distance. The concept of this array can also be extended to more
number of elements and finally an array of isotropic point sources can be formed.
Based on amplitude and phase conditions of isotropic point sources, there are three
types of arrays:
(a) Array with equal amplitude and phases
(b) Array with equal amplitude and opposite phases
(c) Array with unequal amplitude and opposite phases
Two Point Sources with Currents Equal in Magnitude and Phase

...(1)

Page 4
 Fig. 5 Two element array
Consider two point sources A1 and A2, separated by distance d as shown in the Fig. 5. Consider
that both the point sources are supplied with currents equal in magnitude and phase. Consider
point P far away from the array. Let the distance between point P and point sources A 1 and A2

be r1 and r2 respectively. As these radial distances are extremely large as compared with the
distance of separation between two point sources i.e. d, we can assume,

r1= r2 = r
The radiation from the point source A2 will reach earlier at point P than that from point
source A1 because of the path difference. The extra distance is travelled by the radiated
wave from point source A1 than that by the wave radiated from point source A2.
Hence path difference is given
by, Path difference = d cos ʋ
The path difference can be expressed in terms of
wavelength as, Path difference = (d cos ʋ) / λ...(2)
Hence the phase angle ʋ is given by,
Phase angle ʋ = 2π (Path difference)

But phase shift β = 2π/λ, thus equation (3) becomes,

Let E1 be the far field at a distant point P due to point source A l. Similarly let E2 be
the far field at point P due to point source A2. Then the total field at point P be the
addition of the two field components due to the point sources A 1 and A2. If the
phase angle between the two fields is ʋ = βdcosʋ then the far field component at
point P due to point source A1 is given by,

Similarly the far field component at point P due to the point source A 2 is given by,

Note that the amplitude of both the field components is E0 as currents are same and
the point sources are identical.
The total field at point P is given by,

Page 5
Rearranging the terms on R.H.S., we get,

By trigonometric identity,

Hence equation (7) can be written as,

Substituting value of Ψ from equation (4), we get,.

Above equation represents total field in intensity at point P. due to two point sources
having currents of same amplitude and phase. The total amplitude of the field at
point P is 2E0 while the phase shift is βdcosʋ/2
The array factor is the ratio of the magnitude of the resultant field to the magnitude
of the maximum field.

But maximum field is Ernax =2E0

The array factor represents the relative value of the field as a function of ʋ defines
the radiation pattern in a plane containing the line of the array.
Maxima direction

From equation (9), the total field is maximum when is maximum. As we know,

the variation of cosine of a angle is ± 1. Hence the condition for maxima is given by,

Page 6
Let spacing between the two point sources be λ/2. Then we can write,

If n = 0, then

Minima direction

Again from equation (9), total field strength is minimum when is minimum i.e.

0 as cosine of angle has minimum value 0. Hence the condition for minima is given by,

Again assuming d = λ/2 and β=2π/λ, we can write

Half power point direction:

When the power is half, the voltage or current is 1/√2 times the maximum value.
Hence the condition for half power point is given by,

Page 7
Let d=λ/2 and β=2π/λ, then we can write,

The field pattern drawn with ET against ʋ for d=λ/2, then the pattern is bidirectional
as shown in Fig 6. The field pattern obtained is bidirectional and it is a figure of
eight.
0
If this pattern is rotated by 360 about axis, it will represent three dimensional doughnut
shaped space pattern. This is the simplest type of broadside array of two point sources
and it is called Broadside couplet as two radiations of point sources are in phase.

Fig. 6 Field pattern for two point source with spacing d=λ/2 and fed with currents equal in
magnitude andphase.

Page 8
Two Point Sources with Currents Equal in Magnitudes but Opposite in Phase
Consider two point sources separated by distance d and supplied with currents
equal in magnitude but opposite in phase. Consider Fig. 5 all the conditions are
exactly same except the phase of the currents is opposite i.e. 180°. With this
condition, the total field at far point P is given by,

Assuming equal magnitudes of currents, the fields at point P due to the point
sources A1 and A2 can be written as,

Substituting values of E1 and E2 in equation (1), we get

Rearranging the terms in above equation, we get,

By trigonometry identity,

Equation (4) can be written as,

Now as the condition for two point sources with currents in phase and out of phase
is exactly same, the phase angle can be written as previous case.
Phase angle = βdcosʋ ...(6) Substituting value of phase angle in
equation (5), we get,

Page 9
Maxima direction

From equation (7), the total field is maximum when is maximum i.e. ±1 as

the maximum value of sine of angle is ±1. Hence condition for maxima is given by,

Let the spacing between two isotropic point sources be equal to d=λ/2
Substituting d=λ/2 and β=2π/λ, in equation (8), we get,

If n = 0. then

Minima direction

Again from equation (7), total field strength is minimum when is

minimum i.e. 0.
Hence the condition for minima is given by,

Assuming d=λ/2 and β=2π/λ in equation (10), we get,

If n = 0, then

Page
10
Half Power Point Direction (HPPD)
When the power is half of maximum value, the voltage or current equals to 1/√2
times the respective maximum value. Hence the condition for the half power point
can be obtained from equation (7) as,

Let d=λ/2 and β=2π/λ, we can write,

Thus from the conditions of maxima, minima and half power points, the field pattern
can be drawn as shown in the Fig. 7.

Fig. 7 Field pattern for two point sources with spacing d = d=λ/2 and fed
with currents equal in magnitude but out of phase by
0
180 .

As compared with the field pattern for two point sources with inphase currents, the
maxima have shifted by 90° along X-axis in case of out-phase currents in two point
source array. Thus the maxima are along the axis of the array or along the line joining
two point sources. In first case, we have obtained vertical figure of eight. Now in above
case, we have obtained horizontal figure of eight. As the maximum field is along the line

Page
11
joining the two point sources, this is the simple type of the end fire array.

Two point sources with currents unequal in magnitude and with any phase
Let us consider Fig. 5. Assume that the two point sources are separated by distance d and
supplied with currents which are different in magnitudes and with any phase difference say
α. Consider that source 1 is assumed to be reference for phase and amplitude of the fields

E1 and E2, which are due to source 1 and source 2 respectively at the distant point P. Let us

assume that E1 is greater than E2 in magnitude as shown in the vector diagram in Fig. 8.

Fig. 8 Vector diagram of fields El and E2

Now the total phase difference between the radiations by the two point sources at
any far point P is given by,

where α is the phase angle with which current I2 leads current Il. Now if α = 0, then
the condition is similar to the two point sources with currents equal in magnitude and
phase. Similarly if α = 180", then the condition is similar to the two point source with
currents equal in magnitude but opposite in phase. Assume value of phase
0
difference as 0 <α< 180 . Then the resultant field at point P is given by,

Note that E1> E2, the value of k is less than unity. Moreover the value of k is given by, 0 ≤ k

≤ 1

Page
12
The magnitude of the resultant field at point P is given by,

The phase angle between two fields at the far point P is given by,

n Element Uniform Linear Arrays

At higher frequencies, for point to point communications it is necessary to have a
pattern with single beam radiation. Such highly directive single beam pattern can be
obtained by increasing the point sources in the arrow from 2 to n say. An array of n
elements is said to be linear array if all the individual elements are spaced equally along
a line. An array is said to be uniform array if the elements in the array are fed with
currents with equal magnitudes and with uniform progressive phase shift along the line.
Consider a general n element linear and uniform array with all the individual elements
spaced equally at distance d from each other and all elements are fed with currents
equal in magnitude and uniform progressive phase shift along line as shown in the Fig.
9.

Fig. 9 Uniform, linear array of n elements

The total resultant field at the distant point P is obtained by adding the fields due to n

Page
13
individual sources vectorically. Hence we can write,

Note that ʋ= (βdcosʋ + α) indicates the total phase difference of the fields from adjacent
sources calculated at point P. Similarly α is the progressive phase shift between two adjacent
0 0 0
point sources. The value of α may lie between 0 and 180 . If α = 0 we get n element uniform
0
linear broadside array. If α = 180 we get n element uniform linear endfire array.
jʋ
Multiplying equation (1) by e , we get,

Subtracting equation (2) from (1), we get,

.
Simply mathematically, we get

According to trigonometric identity,

The resultant field is given by,

This equation (4) indicates the resultant field due to n element array at
distant point P. The magnitude of the resultant field is given by,

Page
14
The phase angle θ of the resultant field at point P is given by,

Array of n elements with Equal Spacing and Currents Equal in Magnitude and Phase •
Broadside Array
Consider 'n' number of identical radiators carries currents which are equal in magnitude and in

phase. The identical radiators are equispaced. Hence the maximum radiation occurs in the

directions normal to the line of array. Hence such an array is known as Uniform broadside array.

Consider a broadside array with n identical radiators as shown in the Fig. 10.

Fig 10 Array of n elements with Equal Spacing

The electric field produced at point P due to an element A0 is given by,

As the distance of separation d between any two array elements is very small as
compared to the radial distances of point P from A0, A1, ...An-1, we can assume r0,
r1, ...rn-1 are approximately same.

Now the electric field produced at point P due to an element A 1 will differ in phase
as r0 and r1 are not actually same. Hence the electric field due to A 1 is given by,

Page
15
Exactly on the similar lines we can write the electric field produced at point P due to
an element A2 as,

Page
16
But the term inside the bracket represent E1

From equation (2), substituting the value of E1, we get,

Similarly, the electric field produced at point P due to element An-1 is given by,

The total electric field at point P is given by,

Let βdcosʋ = ʋ, then rewriting above equation,

Consider a series given

2
by s = 1 + r + r +..... +
n-1 jʋ
r where r = e
... (i)
Multiplying both the sides of the equation (i)
2 n
by r, s . r = r + r +..... + r
Subtracting equation (ii) from (i), we
n
... (ii)
get. s(1-r) = 1-r

Using equation (iii), equation (5) can be modified as,

Page
17
From the trigonometric identities,

Equation (6) can be written as,

The exponential term in equation (7) represents the phase shift. Now considering
magnitudes of the electric fields, we can write,

Properties of Broadside Array

1. Major lobe
In case of broadside array, the field is maximum in the direction normal to the
axis of the array. Thus the condition for the maximum field at point P is given by,

0 0
Thus ʋ = 90 and 270 are called directions of principle maxima.
2. Magnitude of major lobe

Page
18
 The maximum radiation occurs when ʋ=0. Hence we can write,

where, n is the number of elements in the array.

Thus from equation (10) and (11) it is clear that, all the field components add
up together to give total field which is ‘n’ times the individual field when ʋ =
0 0
90 and 270 .
3. Nulls
The ratio of total electric field to an individual electric field is given by,

Equating ratio of magnitudes of the fields to zero,

The condition of minima is given by,

Hence we can write,

Page
19
 where, n= number of elements in the array d=
spacing between elements in meter

λ= wavelength in meter
m= constant= 1, 2 , 3....
Thus equation (13) gives direction of nulls
4. Side Lobes Maxima
The directions of the subsidary maxima or side lobes maxima can be obtained
if in equation (8),

Hence sin(nʋ/2), is not considered. Because if nʋ/2=π/2 then sin nʋ/2 =1

which is the direction of principle maxima.
Hence we can skip sin nʋ/2 = ±π/2 value Thus, we get

Now equation for ʋ can be written as,

The equation (15) represents directions of subsidary maxima or side lobes maxima.

Page
20
5. Beamwidth of Major Lobe
Beamwidth is defined as the angle between first nulls. Alternatively
beamwidth is the angle equal to twice the angle between first null and the
major lobe maximum direction. Hence beamwidth between first nulls is given
by,

Also

Hence
Taking cosine of angle on both sides, we get

If γ is very small, then sin γ ≈ γ. Substituting n above equation we get,

For first null i.e. m=1,

But nd≈ (n-1)d if n is very large. This L= (nd) indicates total length of the
array.

BWFN in degree is written as,

Page
21
 Now HPBW is given by,

HPBW in degree is written as,

6. Directivity
The directivity in case of broadside array is defined as,

where, U0 is average radiation intensity which is given by,

From the expression of ratio of magnitudes we can write,

or
For the normalized condition let us assume E0 = 1, then

Thus field from array is maximum in any direction θ when ʋ = 0. Hence

normalized field pattern is given by,

Hence the field is given by,

where ʋ = βdcosʋ

Page
22
Equation (23) indicated array factor, hence we can write electric field due to n
array as

Assuming d is very small as compared to length of an array,

Then,

Substituting value of E in equation (24) we get

Let

Rewritting above equation we get,

Page
23
For large array, n is large hence nβd is also very large (assuming tending to
infinity). Hence rewriting above equation.

Interchanging limits of integration, we get

By integration formula,

Using above property in above equation we can write,

From equation (23), the directivity is given by,

But Umax = 1 at ʋ = 90° and substituting value of U0 from equation (28), we get,

But β= 2π/λ
Hence

Page
24
 The total length of the array is given by, L = (n - 1) d ≈ nd, if n is very large. Hence

the directivity can be expressed in terms of the total length of the array as,

Array of n Elements with Equal Spacing and Currents Equal in Magnitude but with
Progressive Phase Shift - End Fire Array
Consider n number of identical radiators supplied with equal current which are not in
phase as shown in the Fig. 11. Assume that there is progressive phase lag of βd
radians in each radiator.

Fig.11 End fire array

Consider that the current supplied to first element A0 be I0. Then the current
supplied to A1 is given by,

Similarly the current supplied to A2 is given by,

Thus the current supplied to last element is

The electric field produced at point P, due to A0 is given by,

Page
25
The electric field produced at point P, due to A1 is given by,

But r1 = r0 – dcosʋ

Let ʋ = βd(cosʋ -1)

The electric field produced at point P, due to A2 is given by,

Similarly electric field produced at point P, due to An-1 is given by,

The resultant field at point p is given by,

Considering only magnitude we get,

Page
26
Properties of End Fire Array
1. Major lobe
For the end fire array where currents supplied to the antennas are equal in amplitude
but the phase changes progressively through array, the phase angle is given by,
ʋ = βd(cosʋ -1) ...(9)

In case of the end fire array, the condition of principle maxima is given by,
ʋ = = 0 i.e.

i.e. cosʋ
= 1
0
i.e. ʋ = 0 ...(11)
0
Thus ʋ = 0 indicates the direction of principle maxima.

2. Magnitude of the major lobe

The maximum radiation occurs when ʋ= 0. Thus we can write,

where, n is the number of elements in the array.

3. Nulls
The ratio of total electric field to an individual electric field is given by,

Equating ratio of magnitudes of the fields to zero,

Page
27
The condition of minima is given by,

Henc

e we can write,

Substituting value of ʋ from equation (9), we get,

But β= 2π/λ

Note that value of (cosʋ-1) is always less than 1. Hence it is always negative.
Hence only considering -ve values, R.H.S., we get

where, n= number of elements in the array d=

spacing between elements in meter

λ= wavelength in meter
m= constant= 1, 2 , 3....
Thus equation (15) gives direction of nulls
Consider equation(14),

Expressing term on L.H.S. in terms of halfangles, we get,

Page
28
4. Side Lobes Maxima
The directions of the subsidary maxima or side lobes maxima can be obtained
if in equation (8),

Hence sin(nʋ/2), is not considered. Because if nʋ/2=±π/2 then sin nʋ/2

=1 which is the direction of principle maxima.
Hence we can skip sin nʋ/2 = ±π/2 value Thus, we get

Putting value of ʋ from equation (9) we get

Now equation for ʋ can be

written as,But β = 2π/λ

Note that value of (cosʋ-1) is always less than 1. Hence it is always negative. Hence
only considering -ve values, R.H.S., we get

Page
29
5. Beamwidth of Major Lobe
Beamwidth is defined as the angle between first nulls. Alternatively
beamwidth is the angle equal to twice the angle between first null and the
major lobe maximum direction.
From equation (16) we get,

ʋminis very low

Hence sin ʋmin/2 ≈ ʋmin/2

But nd≈ (n-1)d if n is very large. This L= (nd) indicates total length of
the array. So equation (20) becomes,

BWFN is given by,

BWFN in degree is expressed as

For m=1,

Page
30
6. Directivity
The directivity in case of endfire array is defined as,

where, U0 is average radiation intensity which is

given by, For endfire array, Umax = 1and

The total length of the array is given by, L = (n - 1) d ≈ nd, if n is very large. Hence

the directivity can be expressed in terms of the total length of the array as,

Multiplication of patterns
In the previous sections we have discussed the arrays of two isotropic point sources
radiating field of constant magnitude. In this section the concept of array is extended
to non-isotropic sources. The sources identical to point source and having field
patterns of definite shape and orientation. However, it is not necessary that
amplitude of individual sources is equal. The simplest case of non-isotropic sources
is when two short dipoles are superimposed over the two isotopic point sources
separated by a finite distance. If the field pattern of each source is given by

Page
31
Then the total far-field pattern at point P becomes

...(1)

where

Equation (1) shows that the field pattern of two non-isotropic point sources (short dipoles) is
equal to product of patterns of individual sources and of array of point sources. The pattern of
array of two isotropic point sources, i.e., cos ʋ/2 is widely referred as an array factor. That is

ET= E (Due to reference source) x Array factor

This leads to the principle of pattern multiplication for the array of identical elements.
In general, the principle of pattern multiplication can he stated as follows:

The resultant field of an array of non-isotropic hut similar sources is the product of the
fields of individual source and the field of an array of isotropic point sources, each
located at the phase centre of individual source and hating the relative amplitude and
phase. The total phase is addition of the phases of the individual source and that of
isotropic point sources. The same is true for their respective patterns also.

The normalized total field (i.e., ETn), given in Eq. (1), can re-written as

where E1(θ) = sin θ = Primary pattern of array

= Secondary pattern of array.

Thus the principle of pattern multiplication is a speedy method of sketching the field pattern
of complicated array. It also plays an important role in designing an array. There is no
restriction on the number of elements in an array; the method is valid to any number of
identical elements which need not have identical magnitudes, phase and spacing between
then). However, the array factor varies with the number of elements and their
arrangement, relative magnitudes, relative phases and element spacing. The array
of elements having identical amplitudes, phases and spacing provides a simple array

Page
32
factor. The array factor does not depend on the directional characteristic of the array
elements; hence it can be formulated by using pattern multiplication techniques. The
proper selection of the individual radiating element and their excitation are also
important for the performance of array. Once the array factor is derived using the
point-source array, the total field of the actual array can be obtained using Eq. (2).

Binomial Array
In order to increase the directivity of an array its total length need to be increased. In this
approach, number of minor lobes appears which are undesired for narrow beam applications. In
has been found that number of minor lobes in the resultant pattern increases whenever spacing

between elements is greater than λ/2. As per the demand of modern communication where
narrow beam (no minor lobes) is preferred, it is the greatest need to design an array of only main
lobes. The ratio of power density of main lobe to power density of the longest minor lobe is
termed side lobe ratio. A particular technique used to reduce side lobe level is called tapering.
Since currents/amplitude in the sources of a linear array is non-uniform, it is found that minor
lobes can be eliminated if the centre element radiates more strongly than the other sources.
Therefore tapering need to be done from centre to end radiators of same specifications. The
principle of tapering are primarily intended to broadside array but it is also applicable to end-fire
array. Binomial array is a common example of tapering scheme and it is an array of n-isotropic
sources of non-equal amplitudes. Using principle of pattern multiplication, John Stone first
proposed the binomial array in 1929, where amplitude of the radiating sources arc arranged
according to the binomial expansion. That is. if minor lobes

appearing in the array need to be eliminated, the radiating sources must have
current amplitudes proportional to the coefficient of binomial series, i.e. proportional
to the coefficient of binomial series, i.e.

...(1)
where n is the number of radiating sources in the array.

For an array of total length nλ/2, the relative current in the nth element from the one
end is given by

where r = 0, 1, 2, 3, and the above relation is equivalent to what is known as Pascal's triangle.

Page
33
For example, the relative amplitudes for the array of 1 to 10 radiating sources are as follows:

Since in binomial array the elements spacing is less than or equal to the half-wave
length, the HPBW of the array is given by

and directivity

Using principle of multiplication, the resultant radiation pattern of an n-source

binomial array is given by
In particular, if identical array of two point sources is superimposed one above other,
then three effective sources with amplitude ratio 1:2:1 results. Similarly, in case
three such elements are superimposed in same fashion, then an array of four
sources is obtained whose current amplitudes are in the ratio of 1:3:3:1.
The far-field pattern can be found by substituting n = 3 and 4 in the above
expression and they take shape as shown in Fig. 14(a) and (b).

Fig. 14(a) Radiation pattern of 2-element array with amplitude ratio 1:2:1.

Page
34
 Fig 14(b) Radiation pattern of 3-element array with amplitude ratio 1:3:3:1.

It has also been noticed that binomial array offers single beam radiation at the cost
of directivity, the directivity of binomial array is greater than that of uniform array for
the same length of the array. In other words, in uniform array secondary lobes
appear, but principle lobes are narrower than that of the binomial array.
Disadvantages of Binomial Array
(a) The side lobes are eliminated but the directivity of array reduced.
(b) As the length of array increases, larger current amplitude ratios are required.

Page
35
LECTURE 14: LINEAR ARRAY THEORY - PART II
(Linear arrays: Hansen-Woodyard end-fire array, directivity of a linear array,
linear array pattern characteristics – recapitulation; 3-D characteristics of an
N-element linear array.)

1. Hansen-Woodyard End-fire Array (HWEFA)

The end-fire arrays (EFA) have relatively large HPBW as compared to
broadside arrays.

[Fig. 6-11, p. 270, Balanis]

Nikolova 2022 1
 To enhance the directivity of an end-fire array, Hansen and Woodyard
proposed that the phase shift of an ordinary EFA
β = ± kd (14.1)
be increased as
 2.94 
β = −  kd +  for a maximum at θ = 0° , (14.2)
 N 
 2.94 
β = +  kd +  for a maximum at θ = 180° . (14.3)
 N 
Conditions (14.2)–(14.3) are known as the Hansen–Woodyard conditions for
end-fire radiation. They follow from a procedure for maximizing the directivity,
which we outline below.
The normalized pattern AFn of a uniform linear array is
N 
sin  ( kd cosθ + β ) 
AFn ≈ 2  (14.4)
N
( kd cosθ + β )
2
if ψ = kd cosθ + β is sufficiently small (see previous lecture). We are looking
for an optimal β , which results in maximum directivity. Let
β = − pd , (14.5)
where d is the array spacing and p is the optimization parameter. Then,
 Nd
sin  ( k cosθ − p )
AFn =  2 .
Nd
( k cosθ − p )
2
For brevity, use the notation Nd / 2 = q . Then,
sin [ q (k cosθ − p )]
AFn = , (14.6)
q (k cosθ − p)
sin Z
or AFn = , where Z = q(k cosθ − p) .
Z
Nikolova 2022 2
The radiation intensity becomes
2 sin 2 Z
U (θ ) = AFn = , (14.7)
Z2
2
 sin [ q (k − p ) ] 
U (θ = 0) =   , (14.8)
 q ( k − p ) 
2
U (θ )  z sin Z 
U n (θ ) = = ⋅  , (14.9)
U (θ = 0)  sin z Z 
where
z = q (k − p ) ,
Z = q(k cosθ − p) , and
U n (θ ) is normalized power pattern with respect to θ = 0° .

The directivity at θ = 0° is
4π U (θ = 0)
D0 = (14.10)
Prad
where Prad =  ΩU n (θ )d Ω . To maximize the directivity, U 0 = Prad / 4π is
minimized.
2π π 2
1  z sin Z 
U0 =
4π    sin z Z  sin θ dθ dθ , (14.11)
0 0

2 π 2
1 z   sin [ q (k cosθ − p) ] 
U0 =  
2  sin z    q(k cosθ − p)  sin θ dθ , (14.12)
0
2
1  z   π cos 2 z − 1  1
U0 =   + + Si(2 z )  2kq g ( z ) .
= (14.13)
2kq  sin z   2 2z
z
Here, Si ( z ) =  (sin t / t )dt . The minimum of g ( z ) occurs when
0

z = q(k − p ) ≈ −1.47 , (14.14)

Nikolova 2022 3
 Nd
 (k − p) ≈ −1.47 .
2
Ndk Ndp
 − ≈ −1.47, where dp = − β
2 2
N
 ( dk + β ) ≈ −1.47
2
2.94  2.94 
β ≈− − kd = −  kd + . (14.15)
N  N 
Equation (14.15) gives the Hansen-Woodyard condition for improved directivity
along θ = 0° . Similarly, for θ = 180° ,
 2.94 
β = +  kd + . (14.16)
 N 
Usually, conditions (14.15) and (14.16) are approximated by
 π 
β ≈ ±  kd +  , (14.17)
 N
which is easier to remember and gives almost identical results since the curve
g ( z ) at its minimum is fairly flat.
Conditions (14.15)-(14.16), or (14.17), ensure maximum directivity in the
end-fire direction. There is, however, a trade-off in the side-lobe level, which is
higher than that of the ordinary EFA. Besides, conditions (14.15)-(14.16) have
to be complemented by additional requirements, which would ensure low level
of the radiation in the direction opposite to the main lobe.
(a) Maximum at θ = 0° [reminder: ψ = kd cosθ + β ]
2.94
ψ θ = 0° = −
 2.94  N
β = −  kd +   (14.18)
 N  θ = 0° 2.94
ψ θ =180° = −2kd − .
N
Since we want to have a minimum of the pattern in the θ = 180° direction, we
must ensure that

Nikolova 2022 4
 |ψ |θ =180° ≈ π . (14.19)
It is easier to remember the Hansen-Woodyard conditions for maximum
directivity in the θ = 0° direction as
2.94 π
| ψ |θ =0° = ≈ , | ψ |θ =180° ≈ π . (14.20)
N N

(b) Maximum at θ = 180°

2.94
ψ θ =180° =
2.94 N
β = kd +  (14.21)
N θ =180° 2.94
ψ θ = 0° = 2kd + .
N
In order to have a minimum of the pattern in the θ = 0° direction, we must ensure
that
|ψ |θ =0° ≈ π . (14.22)
We can now summarize the Hansen-Woodyard conditions for maximum
directivity in the θ = 180° direction as
2.94 π
|ψ |θ =180° = ≈ , |ψ |θ =0° ≈ π . (14.23)
N N
If (14.19) and (14.22) are not observed, the radiation in the opposite of the
desired direction might even exceed the main beam level. It is easy to show (use
the relation 2kd + π / N ≈ π ) that the complementary requirement |ψ |= π at the
opposite direction can be met if the following relation is observed:
 N −1 λ . (14.24)
d ≈ 
 N 4
If N is large, d ≈ λ / 4 . Thus, for a large uniform array, Hansen-Woodyard
condition can yield improved directivity only if the spacing between the array
elements is approximately λ / 4 .

Nikolova 2022 5
 ARRAY FACTORS OF A 10-ELEMENT UNIFORM-AMPLITUDE HW EFA

Solid line: d = λ / 4
Dash line: d = λ / 2
N = 10
 π 
β = −  kd + 
 N
Fig. 6.12, p. 273, Balanis

Nikolova 2022 6
2. Directivity of a Linear Array
2.1. Directivity of a BSA
Using the approximate expression for the AF, the normalized radiation intensity
is obtained as
2
 N 
 sin  kd cos θ  2
2  2   sin Z 
U (θ ) = AFn =   =  (14.25)
 N   Z 
kd cosθ
 2 
U U
D0 = 4π 0 = 0 , (14.26)
Prad U av
where U av = Prad / (4π ) . The radiation intensity in the direction of maximum
radiation θ = π / 2 in terms of AFn is unity:
U 0 = U max = U (θ = π / 2) = 1 ,
 D0 = U av
−1 . (14.27)
The radiation intensity averaged over all directions is calculated as
2
N 
2π π 2 π sin  kd cosθ 
1 sin Z 1 2  sin θ dθ .
4π 0 0 Z 2 2 0
U av = sin θ dθ d φ =
N
kd cosθ
2
Change variable:
N N
Z= kd cosθ  dZ = − kd sin θ dθ . (14.28)
2 2
Then,
Nkd
−
2 2
1 2  sin Z 
U av =−
2 N kd 
Nkd

 Z 
 dZ , (14.29)
2

Nikolova 2022 7
 Nkd
2 2
1  sin Z 
U av =
Nkd Nkd  Z  dZ . (14.30)
−
2

The function ( Z −1 sin Z )2 is a relatively fast decaying function as Z increases.

That is why, for large arrays, where Nkd / 2 is big enough ( ≥ 20 ) , the integral
(14.30) can be approximated by
∞ 2
1  sin Z  π
U av ≈
Nkd   Z  dZ =
Nkd
, (14.31)
−∞

1 Nkd d
D0 = ≈ = 2N   . (14.32)
U av π λ
Substituting the length of the array L = ( N − 1) d in (14.32) yields
 L  d 
D0 ≈ 2  1 +   . (14.33)
 d  λ 
N

For a large array ( L ≫ d ) ,

D0 ≈ 2 L / λ . (14.34)

2.2. Directivity of ordinary EFA

Consider an EFA with maximum radiation at θ = 0 °, i.e., β = −kd .
2
 N 
sin
  2 kd ( cos θ − 1)    sin Z 2
U (θ ) = AFn
2
=  
 =  , (14.35)
 
kd ( cosθ − 1)    
 N Z
  2  
N
where Z = kd (cosθ − 1) . The averaged radiation intensity is
2
2π π 2 π 2
P 1  sin Z  1  sin Z 
U av = rad =
4π 4π   Z 
  sin θ d θ d φ =
2   Z  sin θ dθ .
0 0 0

Nikolova 2022 8
Since
N N
Z= kd (cos θ − 1) and dZ = − kd sin θ dθ , (14.36)
2 2
it follows that
− Nkd /2 2
1 2  sin Z 
U av =−
2 Nkd  
 Z
 dZ ,

0
Nkd /2 2
1  sin Z 
U av =
Nkd  
 Z 
 dZ . (14.37)
0

If ( Nkd ) is sufficiently large, the above integral can be approximated as

∞ 2
1  sin Z  1 π
U av ≈  
Nkd 0  Z 
 dZ = ⋅ .
Nkd 2
(14.38)

The directivity then becomes

1 2 Nkd d 
 D0 ≈ = = 4N   . (14.39)
U av π λ 
The comparison of (14.39) and (14.32) shows that the directivity of an EFA is
approximately twice as large as the directivity of the BSA.
Another (equivalent) expression can be derived for D0 of the EFA in terms of
the array length L = (N−1)d:
 L  d 
D0 = 4 1 +   . (14.40)
 d  λ 
For large arrays, the following approximation holds:
D0 = 4 L / λ if L ≫ d . (14.41)

2.3. Directivity of HW EFA

If the radiation has its maximum at θ = 0° , then the minimum of U av is
obtained as in (14.13):

Nikolova 2022 9
 2
min
1 2  Z min   π cos(2 Z min ) − 1 
U av =    + + Si(2 Z min )  , (14.42)
2k Nd  sin Z min   2 2 Z min 
where Z min = −1.47 ≈ −π / 2 .
2
1  π  π 2  0.878
 U av
min =    + − 1.8515 = . (14.43)
Nkd  2   2 π  Nkd
The directivity is then
1 Nkd   d 
D0 = = = 1.789  4 N    . (14.44)
min
U av 0.878   λ 
From (14.44), we can see that using the HW conditions leads to improvement of
the directivity of the EFA with a factor of 1.789. Equation (14.44) can be
expressed via the length L of the array as
  L  d     L 
D0 = 1.789  4  1 +    = 1.789  4    . (14.45)
  d  λ     λ 

Example: Given a linear uniform array of N isotropic elements (N = 10), find the
directivity D0 if:
a) β = 0 (BSA)
b) β = −kd (ordinary EFA)
c) β = −kd − π / N (Hansen-Woodyard EFA)
In all cases, d = λ / 4 .

a) BSA
d 
D0 ≈ 2 N   = 5 ( 6.999 dB )
λ 
b) Ordinary EFA
d 
D0 ≈ 4 N   = 10 (10 dB )
λ 

Nikolova 2022 10
 θ g = arccos ( ±0.8 )  θ g1,2 ≈ 37°,143° .

5. Ordinary End-fire Array

An end-fire array is an array, which has its maximum radiation along the
axis of the array (θ = 0°,180°) . It may be required that the array radiates only in
one direction – either θ = 0° or θ = 180° . For an AF maximum at θ = 0° ,
ψ = kd cosθ + β θ =0° = kd + β = 0 , (13.42)

 β = −kd , for θmax = 0° . (13.43)

For an AF maximum at θ = 180° ,
ψ = kd cosθ + β θ =180° = −kd + β = 0 ,
 β = kd , for θ max = 180° . (13.44)
If the element separation is multiple of a wavelength, d = nλ , then in addition
to the end-fire maxima there also exist maxima in the broadside direction
(θ = 90 ). As with the broadside array, in order to avoid grating lobes, the
maximum spacing between the element should be less than λ :
d <λ.
(Show that an end-fire array with d = λ / 2 has 2 maxima for β = −kd : at
θ = 0 and at θ = 180 .)

Nikolova 2022 18
 AF pattern of an EFA: N = 10, d = λ / 4

Fig. 6-11, p. 270, Balanis

Nikolova 2022 19
6. Phased (Scanning) Arrays
It was already shown that the 0th order maximum (m=0) of AFn occurs when
ψ = kd cosθ 0 + β = 0 . (13.45)
This gives the relation between the direction of the main beam θ 0 and the phase
difference β . Therefore, the direction of the main beam can be controlled by
β . This is the basic principle of electronic scanning for phased arrays.
When the scanning is required to be continuous, the feeding system must be
capable of continuously varying the progressive phase β between the elements.
This is accomplished by ferrite or diode (varactor) phase shifters.

Example: Derive the values of the progressive phase shift β as dependent on

the direction of the main beam θ 0 for a uniform linear array with d = λ / 4 .
From equation (13.45):
2π λ π
β = −kd cosθ 0 = − cosθ 0 = − cosθ 0
λ 4 2

θ0 β
0˚ -90˚
60˚ -45˚
120˚ +45˚
180˚ +90˚

The approximate HPBW of a scanning array is obtained using (13.31) with

β = −kd cosθ 0 :
 λ  2.782  
θ h 1,2 = arccos   − β ±  . (13.46)
 2π d  N  
The total beamwidth is

Nikolova 2022 20
 HPBW = θ h1 − θ h 2 , (13.47)
 λ  2.782    λ  2.782  
HPBW = arccos   kd cos θ −  − arccos  2π d  kd cos θ + 
N   N  
0 0
 2π d   
(13.48)
Since k = 2π / λ ,
 2.782   2.782 
HPBW = arccos cosθ0 − − arccos cos θ + . (13.49)
Nkd   Nkd 
0

We can use the substitution N = ( L + d ) / d to obtain
  λ 
HPBW = arccos  cosθ 0 − 0.443   −
  L + d  
(13.50)
  λ 
arccos  cos θ 0 + 0.443   .
  L + d 
Here, L is the length of the array.
Be aware that the equations in (13.49) and (13.50) can be used to calculate
the HPBW of an array as long as it is not an end-fire array. End-fire arrays have
circularly symmetric beams around the end-fire direction, in which case
 2.782 
HPBW = 2arccos  1 − . (13.51)
 Nkd 

Nikolova 2022 21
LECTURE 15: LINEAR ARRAYS – PART III
(N-element linear arrays with uniform spacing and non-uniform amplitude:
Binomial array; Dolph–Tschebyscheff array. Directivity and design.)

1. Advantages of Linear Arrays with Nonuniform Amplitude Distribution

The most often met BSAs, classified according to the type of their excitation
amplitudes, are:
a) the uniform BSA – relatively high directivity, but the side-lobe levels are
high;
b) Dolph–Tschebyscheff (or Chebyshev)1 BSA – for a given directivity with a
fixed number of array elements, achieves the lowest side-lobe level;
c) binomial BSA – does not have good directivity (for a given number of
elements) but has low side-lobes (if d = λ / 2 , no side lobes at all).

2. Array Factor of Linear Arrays with Nonuniform Amplitude

Distribution
Let us consider a linear array with an even number (2M) of elements,
located symmetrically along the z-axis, with excitation, which is also
symmetrical with respect to z = 0 . For a broadside array ( β = 0) ,
1 3 2 M −1
j kd cosθ j kd cosθ j kd cosθ
AF e = a1e 2 + a2 e 2 + ⋯ + aM e 2 +
(15.1)
1 3 2 M −1
− j kd cosθ − j kd cosθ −j kd cosθ
+ a1e 2 + a2 e 2 + ⋯ + aM e 2 ,
M
 2n − 1  
 AF e = 2 an cos   kd cos θ . (15.2)
n =1   2  
If the linear array consists of an odd number (2M+1) of elements, located
symmetrically along the z-axis, the array factor is
AF o = 2a1 + a2 e jkd cosθ + a3e j 2 kd cosθ + ... + aM +1e jMkd cosθ +
(15.3)
+ a2e − jkd cosθ + a3e− j 2 kd cosθ + ... + aM +1e− jMkd cosθ ,
M +1
 AF o = 2  an cos ( n − 1) kd cos θ  . (15.4)
n =1

1
Russian spelling is Чебышёв.

Nikolova 2022 1
 EVEN- AND ODD-NUMBER ARRAYS

Fig. 6.17, p. 291, Balanis

Nikolova 2022 2
The normalized AF derived from (15.2) and (15.4) can be written in the form
M
AF e =  an cos [ (2n − 1)u ], for N = 2 M , (15.5)
n =1
M +1
AF o =  an cos [ 2(n − 1)u ], for N = 2M + 1, (15.6)
n =1

1 πd
where u = kd cosθ = cosθ .
2 λ

Examples of AFs of arrays of nonuniform amplitude distribution

a) uniform amplitude distribution (N = 5, d = λ / 2 , max. at θ 0 = 90 °)

pp. 148-149, Stutzman

Nikolova 2022 3
 b) triangular (1:2:3:2:1) amplitude distribution (N = 5, d = λ / 2 , max. at
θ 0 = 90 °)

pp. 148-149, Stutzman

Nikolova 2022 4
 c) binomial (1:4:6:4:1) amplitude distribution (N = 5, d = λ / 2 , max. at
θ 0 = 90 °)

pp. 148-149, Stutzman

Nikolova 2022 5
 d) Dolph-Tschebyschev (1:1.61:1.94:1.61:1) amplitude distribution (N = 5,
d = λ / 2 , max. at θ 0 = 90 °)

pp. 148-149, Stutzman

Nikolova 2022 6
 e) Dolph-Tschebyschev (1:2.41:3.14:2.41:1) amplitude distribution (N = 5,
d = λ / 2 , max. at θ 0 = 90 °)

pp. 148-149, Stutzman

Notice that as the current amplitude is tapered more gradually toward the edges
of the array, the side lobes tend to decrease, and the beamwidth tends to
increase.

Nikolova 2022 7
3. Binomial Broadside Array
The binomial BSA was investigated and proposed by J. S. Stone2 to
synthesize patterns without side lobes. First, consider a 2–element array (along
the z-axis).
z

x
The elements of the array are identical and their excitations are the same. The
array factor is of the form
AF = 1 + Z , where Z = e jψ = e j ( kd cosθ + β ) . (15.7)
If the spacing is d ≤ λ / 2 and β = 0 (broad-side maximum), the array pattern
|AF| has no side lobes at all. This is proven as follows.
| AF |2 = (1 + cosψ )2 + sin 2 ψ = 2(1 + cosψ ) = 4cos 2 (ψ / 2) (15.8)
where ψ = kd cosθ . The first null of the array factor is obtained from (15.8) as
1 2π π  λ 
⋅ ⋅ d cos θ n1,2 = ±  θ n1,2 = ± arccos  . (15.9)
2 λ 2  2d 
As long as d < λ / 2 , the first null does not exist. If d = λ / 2 , then θ n1,2 = 0,
180 °. Thus, in the “visible” range of θ, all secondary lobes are eliminated.
Second, consider a 2–element array whose elements are identical and the
same as the array given above. The distance between the two arrays is again d.
z

d
y

2
US Patents #1,643,323, #1,715,433.

Nikolova 2022 8
This new array has an AF of the form
AF = (1 + Z )(1 + Z ) = 1 + 2 Z + Z 2 . (15.10)
Since (1 + Z ) has no side lobes, (1 + Z )2 does not have side lobes either.
Continuing the process for an N-element array produces
AF = (1 + Z ) N −1 . (15.11)
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 (15.11). Making use of Pascal’s triangle,
1
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. An array with a binomial distribution of the excitation
amplitudes is called a binomial array. The excitation distribution as given by
the binomial expansion gives the relative values of the amplitudes. It is
immediately seen that there is a fairly wide variation of the amplitude, which is
a disadvantage of the BAs. The overall efficiency of such an array would be
low. Besides, the BA has relatively wide beam. Its HPBW is the largest as
compared to the uniform BSA or the DCA for a give number of elements.
An approximate closed-form expression for the HPBW of a BA with
d = λ / 2 is
1.06 1.06 1.75
HPBW ≈ = = , (15.12)
N −1 2L λ Lλ
where L = ( N − 1)d is the array’s length. The AFs of 10-element broadside
binomial arrays (N = 10) are given below.

Nikolova 2022 9
 d =λ
d = 3λ / 4
d =λ/2 d = 3λ / 4

d =λ/4

d =λ

Fig. 6.18, p.293, Balanis

Nikolova 2022 10
 The directivity of a broadside BA with spacing d = λ / 2 can be calculated
as
4π 2
D0 = = 2( N −1)
, (15.13)
Prad π   π 
   2
cos  cos θ 

dθ
0

(2 N − 2) ⋅ (2 N − 4) ⋅ ... ⋅ 2
D0 = , (15.14)
(2 N − 3) ⋅ (2 N − 5) ⋅ ... ⋅1
D0 ≈ 1.77 N = 1.77 1 + 2 L λ . (15.15)

4. Dolph–Chebyshev Array (DCA)

Dolph proposed (in 1946) a method for designing arrays with any desired
side-lobe level for a given HPBW. 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. A DCA with no side lobes
(side-lobe level of −∞ dB) reduces to the binomial design.
4.1. Chebyshev polynomials
The Chebyshev polynomial of order m is defined by
(−1) m cosh(m ⋅ arccosh | z |), z ≤ −1,

Tm ( z ) = cos ( m ⋅ arccos( z ) ) , −1 ≤ z ≤ 1, (15.16)

cosh ( m ⋅ arccosh(z ) ) , z ≥ 1.
A Chebyshev polynomial Tm(z) of any order m can be derived via a recursion
formula, provided Tm−1(z) and Tm−2(z) are known:
Tm ( z ) = 2 zTm −1 ( z ) − Tm − 2 ( z ) . (15.17)
Explicitly, from (15.16) we see that
m = 0, T0 ( z ) = 1
m = 1, T1 ( z ) = z .
Then, (15.17) produces:

Nikolova 2022 11
 m = 2, T2 ( z ) = 2 z 2 − 1
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 , etc.
If | z |≤ 1, then the Chebyshev polynomials are related to the cosine functions
through z = cos x so that Tm ( x) = cos(mx) ; see (15.16). Thus, the Chebyshev
polynomials are handy in expanding the function cos(mx) as a polynomial of
cos(x) of order m. For example, for m = 2 ,
cos 2 x = 2cos 2 x − 1 . (15.18)
Similar expansion holds for the hyperbolic cosine function cosh. In general, the
Chebyshev argument z can be related to the cosine argument x by
z = cos x ↔ x = arccos z ,| z |≤ 1
(15.19)
z = cosh x ↔ x = arccosh z ,| z |> 1 .
For example, (15.18) can be written as:
2
cos(2 arccos z ) = 2 [ cos( arccos z ) ] −1 cos(2 arccos z ) = 2 z 2 −1=T2 ( z ) . (15.20)
Properties of the Chebyshev polynomials of z
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.

Nikolova 2022 12
 Fig. 6.19, pp. 296, Balanis

Nikolova 2022 13
4.2. Chebyshev array design
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
m = N −1. (15.21)
In general, for a given side-lobe level, the higher the order m of the polynomial,
the narrower the beamwidth. However, for m > 10, the difference is not
substantial – see the slopes of Tm ( z ) in the previous figure. The AFs of an N-
element array in (15.5) or in (15.6) are shaped by a Chebyshev polynomial by
requiring that
 e M
 AF =  an cos [ (2n − 1)u ], M = N / 2, even
 n =1
TN −1 ( z ) =  (15.22)
M +1
 AF o =
  an cos [ 2(n − 1)u ], M = ( N − 1) / 2, odd
n =1

where, as before, u = (π d / λ ) cosθ . Let the side-lobe level be

Emax 1
R0 = = (aka voltage ratio). (15.23)
Esl AFsl
Then, we require that the maximum of TN −1 is fixed at an argument z0 , where
TNmax
−1 ( z0 ) = R0 , | z0 |> 1 . (15.24)
Equation (15.24) the maximum AF value AF (u ) = AF max (u0 ) , and z0 must
satisfy the condition | z0 |> 1 so that TN −1 > 1. The maxima of | TN −1 ( z ) | for
| z |≤ 1 are equal to unity and they correspond to the side lobes of the AF. Thus,
AF (u ) has side-lobe levels equal to 1 / R0 .
The AFs in (15.22) are sums of cosine functions of the form cos(mu ) ,
where m = 2n − 1 for an even-number array and m = 2(n − 1) for an odd-number
array. Therefore, they can be expanded into polynomials of cos(u) of order m

Nikolova 2022 14
using the Chebyshev recursion formula. On the other hand, TN −1 ( z ) is a
polynomial of z where z is limited to the range
−1 ≤ z ≤ z0 > 1. (15.25)
Since −1 ≤ cos u ≤ 1 , the relation between z and cosu must be normalized as
cos u = z / z0 , where u = (π d / λ ) cosθ . (15.26)

Design of a DCA of N elements – general procedure:

1) Expand the AF, as given by the right side of (15.22), by replacing each
cos(mu ) term with the power series of cosu .
2) Determine z0 such that TN −1 ( z0 ) = R0 (voltage ratio).
3) Substitute cos u = z / z0 in the AF found in step 1.
4) Equate the AF found in Step 3 to TN −1 ( z ) and determine the coefficients for
each power of z.

Example: Design a DCA (broadside) of N=10 elements with a major-to-minor

lobe ratio of R0 = 26 dB. Find the excitation coefficients and the AF.

Solution:
The order of the Chebyshev polynomial is m = N − 1 = 9 . The AF for an even-
number array is:
5
πd
AF2 M =  an cos [ (2n − 1)u ], u = cosθ , M = 5 .
n =1 λ
Step 1: Write AF10 (see sum above) explicitly:
AF10 = a1 cos u + a2 cos3u + a3 cos5u + a4 cos 7u + a5 cos9u .
Expand the cos(mu ) terms as powers of cosu :
cos3u = 4 cos3 u − 3cos u , ( a2 terms)
cos5u = 16 cos5 u − 20 cos3 u + 5cos u , ( a3 terms)
cos7u = 64 cos7 u − 112 cos5 u + 56 cos3 u − 7cos u , ( a4 terms)
cos 9u = 256 cos9 u − 576 cos7 u + 432 cos5 u − 120 cos3 u + 9cos u . ( a5 terms)
Nikolova 2022 15
Note that the above expansions can be readily obtained from the recursive
Chebyshev relation (15.17) and the substitution z = cos u . For example,
m = 3, T3 ( z ) = 4 z 3 − 3z
translates into: cos(3u ) = 4cos3 u − 3cos u .
Step 2: Determine z0 :
26
R0 = 26 dB  R0 = 10 20 ≈ 20  T9 ( z0 ) = 20 ,
cosh [9arccosh( z0 )] = 20 ,
9arccosh( z0 ) = arccosh20 = 3.69 ,
arccosh( z0 ) = 0.41,
z0 = cosh 0.41  z0 = 1.08515 .

Step 3: Express the AF from Step 1 in terms of cos u = z / z0 and make equal to
the Chebyshev polynomial:
z
AF10 = ( a1 − 3a2 + 5a3 − 7a4 + 9a5 )
z0
z3
+ 3 ( 4a2 − 20a3 + 56a4 − 120a5 )
z0
z5
+ 5 (16a3 − 112a4 + 432a5 )
z0
z7
+ 7 ( 64a4 − 576a5 )
z0
z9
+ 9
( 256a5 ) =
z0
= 9 z − 120 z 3 + 432 z 5 − 576 z 7 + 256 z 9
T9 ( z )

Step 4: Find 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

Nikolova 2022 16
 4a2 − 20a3 + 56a4 − 120a5 = −120 z07  a2 = 5.2073
a1 − 3a2 + 5a3 − 7a4 + 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
 AF10 = 2.789cos(u ) + 2.496cos(3u ) +1.974cos(5u ) +1.357 cos(7u ) + cos(9u )
πd
where u = cosθ . Remember that the Chebyshev variable z relates to u as
λ
z = z0 cos u .

Fig. 6.20b, p. 298, Balanis

Nikolova 2022 17
 Fig. 6.21, p. 300, Balanis

Nikolova 2022 18
4.3. Maximum affordable d for Dolph-Chebyshev arrays
This restriction arises from the requirement for a single major lobe – see
also equation (15.25), −1 ≤ z ≤ z0 :
πd
z ≥ −1, cos u = z / z0 , u = cosθ ,
λ
πd 
 z = z0 cos  cos θ  ≥ −1 . (15.27)
 λ 
For a broadside array, when θ varies from 0° to 180°, the argument z assumes
values
πd 
from z(θ =0 ) = z0 cos   (15.28)
 λ 
to z(θ =90 ) = z0 (15.29)
 πd 
back to z(θ =180 ) = z0 cos  −  = z(θ = 0 ) . (15.30)
 λ 
The extreme value of z to the left on the abscissa corresponds to the end-fire
directions of the AF (θ = 0,180° ). This value must not go beyond z = −1 .
Otherwise, end-fire lobes of levels higher than 1 (higher than R0 ) will appear.
Therefore, the inequality (15.27) must hold for θ = 0° or 180°:
πd  πd  1
z0 cos   ≥ − 1  cos   ≥ − . (15.31)
 λ   λ  z0
Let the angle γ be such that cos γ = 1 / z0 (see figure below). Then,
γ = arccos ( z0−1 ) and
cos (π d / λ ) ≥ − cos γ . (15.32)
1

π d max
λ
−1 cos γ γ 1
1 allowed cos(π d / λ )
− cos(π d / λ )
z0
ILLUSTRATION OF EQUATION (15.31) AND THE REQUIREMENT IN (15.32) AND (15.33)

Nikolova 2022 19
Remember that z0 > 1; thus γ is a real-positive angle. Then, from (15.32),
π d / λ < π − γ = π − arccos ( z0−1 ) (15.33)
or
π d max d 1 1
= π − arccos ( z0−1 )  max = 1 − arccos   (15.34)
λ λ π  z0 
γ

For the case of the previous example,

d 1  1  0.39879
< 1− arccos   = 1 − = 0.873 ,
λ π  1.08515  π
d max = 0.873λ .

5. Directivity of Non-uniform Arrays

It is difficult to derive closed form expressions for the directivity of non-
uniform arrays. Here, we derive expressions in the form of series in the most
general case of a linear array when the excitation coefficients are known.
The non-normalized array factor is
N −1
AF =  ane jβ e jkz cosθ ,
n n (15.35)
n=0

where
an is the amplitude of the excitation of the n-th element;
β n is the phase angle of the excitation of the n-th element;
zn is the z-coordinate of the n-th element.
The maximum AF is
N −1
AFmax =  an . (15.36)
n =0

The normalized AF is

Nikolova 2022 20
 N −1

AF
 ane jβ e jkz cosθ
n n

n =0
AFn = = N −1
. (15.37)
AFmax
 an
n =0

The beam solid angle of a linear array along z is

π
2
Ω A = 2π  AFn (θ ) sin θ dθ ,
0
N −1 N −1 π
2π
ΩA = 2   am a p e j( βm − β p )
 e jk ( z m − z p ) cos θ sin θ dθ , (15.38)
 N −1  m=0 p =0
 
0
an
 n =0 
where
π
2sin  k ( zm − z p ) 
 e jk ( zm − z p ) cosθ sin θ dθ = k ( zm − z p )
.
0
4π N −1 N −1 sin  k ( zm − z p ) 
 ΩA = 2   am a p e j ( β m − β p ) ⋅ k ( zm − z p )
. (15.39)
 N −1  m =0 p =0
 n a
 n =0 
From
D0 = 4π / Ω A ,
we obtain
2
 N −1 
  an 
 D0 = N −1 N −1  n =0  . (15.40)
sin  m
k ( z − z p 
)
  am a p e j( βm − β p ) ⋅ k( zm − z p ) 
m =0 p =0

For equispaced linear ( zn = nd ) arrays, (15.40) reduces to

Nikolova 2022 21
 2
 N −1 
  an 
D0 = N −1 N −1  n =0  . (15.41)
sin [ ( m − p ) kd ]
  am a p e j ( βm − β p ) ⋅ (m − p)kd
m =0 p =0

For equispaced broadside arrays, where β m = β p for any (m,p), (15.41)

reduces to
2
 N −1 
  an 
D0 = N −1 N −1  n = 0  . (15.42)
sin [ (m − p )kd ]
  am a p ⋅ (m − p)kd
m =0 p = 0

For equispaced broadside uniform arrays,

N2
D0 = N −1 N −1 . (15.43)
sin [ (m − p )kd ]
  (m − p)kd
m = 0 p =0

When the spacing d is a multiple of λ / 2 , equation (15.42) reduces to

2
 N −1 
  an 
λ
D0 = Nn−=10  , d = , λ ,... . (15.44)
2
 ( an ) 2
n =0

Example: Calculate the directivity of the Dolph–Chebyshev array designed in

the previous example if d = λ / 2 .

The 10-element DCA has the following amplitude distribution:

a5 = 1; a4 = 1.357; a3 = 1.974; a2 = 2.496; a1 = 2.798 .
We make use of (15.44):

Nikolova 2022 22
 2
 5 
4   an  2
D0 = 5 n =1  = 2⋅
( 9.625 )
= 8.9090 (9.5 dB) .
20.797
2 (an ) 2
n =1

Output from ARRAYS.m: D0 = 8.9276 .

6. Half-power Beamwidth of a Broadside DCA

For large DCAs with side lobes in the range from –20 dB to –60 dB, the
HPBW HPBWDCA can be found from the HPBW of a uniform array HPBWUA
by introducing a beam-broadening factor f given by
2
2 
f = 1 + 0.636  cosh  (arccoshR0 )2 − π 2   , (15.45)
 R0  
so that
HPBWDCA = f × HPBWUA . (15.46)
In (15.45), R0 denotes the side-lobe level (the voltage ratio).

Nikolova 2022 23
Electronic scanning arrays
This is considered to be the best feature of a phased array antenna. To be able to scan in a
given space, without having to rotate the antenna is called inertia less scanning and serves
many advantages, especially in field operations.

To illustrate the principle, let us take a simple three element linear array arranged with
the following phase relationship due to its beam tilted towards an angle θ to receive
signal, as shown below:

Figure 1: A simple three element linear array

The path difference between adjacent elements is= d sin (θ) and it progressively increases
from right to left. So BB’= d sin (θ) whereas CC’= 2d sin (θ). If we assume that an
incoming field with planar wavefront having equal amplitude and equiphase is
across CC’, one can visualize the extra path that has to be traveled to reach
element A and B, in comparison to the element C. Translating this into phase shift, we
have a phase delay= φB = (2π/λ) d sin (θ) at element B, and φA= (2π/λ)(2d)sin(θ) at
element A, all reckoned with element C as the point of phase reference. So in the final
tally, the field captured along the three-element array comes out to be EResultant = E (1+e-jφ+
e-j2φ), where φ = (2π/λ) d sin(θ) and E is the field value of the incoming equiphase front.
The magnitude of the resultant field can be simplified to E(1+2 cos (φ)). It is clear that
the field captured by all elements adds up to 3E only when cos (φ) equals zero, which is
the case when the antenna is looking at its perpendicular direction, called the broadside.
But this is the initial position without scanning. The expression further shows that array
gain falls as the angle of scan increases, theoretically falling to E at φ=± π/2. (A simple
vector diagram will reveal that at this state, the field received by element A will oppose
that of element C). This position also coincides with the limits of scan in the visible
region corresponding to a scan limit of θ= ± 90°, for (d/λ) =1/2.
The above example forms the basis for exploring the method of electronic scanning,
further.

If we need to receive a signal from an angle inclined from the bore-sight (θ=0°), then
there is an inevitable phase shift (delay) to signal reaching the different elements. Hence
the uniform phase front along the linear antenna array is no longer along its axis but lifted
(refer line CC’) at an angle equal to the angle of scan, (θ in the above example). If we
wish to bring back this tilted equiphase plane along the axis of the array, it is obvious
each element should now have a phase shift of its own to counter the phase delay they
suffered. The required phase shift would be equal in magnitude but in opposite direction.
Let the array be now modified with each element provided with a phase shifter
individually. A phase shift + ψ is introduced in a progressive manner from left to right to
compensate the phase delays produced earlier by the incoming signal.

Figure 2: Array elements with phase shifters

The modified array equation gives the resultant field = E (1+e-j(φ-ψ) + e-j2(φ-ψ))
Making (φ-ψ) = δ, the above reduces to Ee-j δ(e+jδ + 1 + e-jδ) = E e-j δ [1+2 cos (δ)], which
attains maximum at δ =0, making φ=ψ. Remember that φ=(2π/λ) d sin (θ).
Suppose we want the beam to position at θ=30°, then ψ = (2π/λ) d sin (30°). If we need
to scan to a direction θ=-30°, then ψ =- (2π/λ) d sin(30°), will be a lagging phase shift.
What essentially resulted was that the equiphase plane inclined at θ was brought back to
the array surface by phase weighting (ψ) of the element as arranged in Figure 2. The
phase weight is dependent on the angle chosen for the scan, once the designer has already
selected the spacing among elements (d/λ). This phase weight will then have to
continually track its value in sync with the position of the scan angle ‘θ ‘.
A general equation can now be attempted to describe the pattern of a linear array antenna
with N elements:
A simulation run with an array antenna is shown in Figure 1. A linear array of 100
elements was chosen (N=100), with the inter-distance between elements d=λ/2. The
desired angle of scan was u1=sin(θ0) This was set for three values: u=0 (broadside), and
at u1=± 0.5 (angle of ±30°).

Figure 3: Array scan pattern

It is true that the gain of the scanned patterns becomes a function of the scanned angle
[Gθ = G cos(θ)]; this fact is not visible in this plot as each pattern is normalized with
respect to its own peak.