Nikolova 2012

1
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 2012
2
To enhance the directivity of an end-fire array, Hansen and Woodyard
proposed that the phase shift of an ordinary EFA
kd | = (14.1)
be increased as

2.94
kd
N
|
| |
= +
|
\ .
for a maximum at 0 u = , (14.2)

2.94
kd
N
|
| |
= + +
|
\ .
for a maximum at 180 u = . (14.3)
Conditions (14.2)(14.3) are known as the HansenWoodyard conditions for
end-fire radiation. They follow from a procedure for maximizing the directivity,
which we outline below.
The normalized pattern AF
n
of a uniform linear array is

( )
( )
sin cos
2
cos
2
n
N
kd
AF
N
kd
u |
u |
(
+
(

~
+
(14.4)
if the argument cos kd u | = + is sufficiently small (see previous lecture). We
are looking for an optimal | , which results in maximum directivity. Let
pd | = , (14.5)
where d is the array spacing and p is the optimization parameter. Then,

( )
( )
sin cos
2
cos
2
n
Nd
k p
AF
Nd
k p
u
u
(

(

=

.
For brevity, use the notation / 2 Nd q = . Then,

| |
sin ( cos )
( cos )
n
q k p
AF
q k p
u
u

=

, (14.6)
or
sin
n
Z
AF
Z
= , where ( cos ) Z q k p u = .
Nikolova 2012
3
The radiation intensity becomes

2
2
2
sin
( )
n
Z
U AF
Z
u = = , (14.7)

| |
2
sin ( )
( 0)
( )
q k p
U
q k p
u

= =
`

)
, (14.8)

2
( ) sin
( )
( 0) sin
n
U z Z
U
U z Z
u
u
u
| |
= =
|
=
\ .
, (14.9)
where
( ) z q k p = ,
( cos ) Z q k p u = , and
( )
n
U u is normalized power pattern with respect to 0 u = .

The directivity at 0 u = is

0
4 ( 0)
rad
U
D
P
t u =
= (14.10)
where ( )
rad n
P U d u
O
= O
}}
. To maximize the directivity,
0
/ 4
rad
U P t = is
minimized.

2
2
0
0 0
1 sin
sin
4 sin
z Z
U d d
z Z
t t
u u u
t
| |
=
|
\ .
} }
, (14.11)

| |
2
2
0
0
sin ( cos )
1
sin
2 sin ( cos )
q k p
z
U d
z q k p
t
u
u u
u

| |
=
`
|

\ .
)
}
, (14.12)

2
0
1 cos 2 1 1
Si(2 ) ( )
2 sin 2 2 2
z z
U z g z
kq z z kq
t
| | (
= + + =
|
(
\ .
. (14.13)
Here,
( )
0
Si (sin / )
z
z t t dt =
}
. The minimum of ( ) g z occurs when
( ) 1.47 z q k p = ~ , (14.14)
Nikolova 2012
4
( ) 1.47
2
Nd
k p ~ .
1.47, where
2 2
Ndk Ndp
dp | ~ =
( ) 1.47
2
N
dk | + ~

2.94 2.94
kd kd
N N
|
| |
~ = +
|
\ .
. (14.15)
Equation (14.15) gives the Hansen-Woodyard condition for improved
directivity along 0 u = . Similarly, for 180 u = ,

2.94
kd
N
|
| |
= + +
|
\ .
. (14.16)
Usually, conditions (14.15) and (14.16) are approximated by
kd
N
t
|
| |
~ +
|
\ .
, (14.17)
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 minimum beamwidth
(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 0 u = [reminder: cos kd u | = + ]

0
0
180
2.94
2.94
2.94
2
N
kd
N
kd
N
u
u
u

=
=
=
=
| |
= +
|
\ .
=

(14.18)
Since we want to have a minimum of the pattern in the 180 u =

direction, we
must ensure that
Nikolova 2012
5

180
| |
u
t
=
~ . (14.19)
It is easier to remember the Hansen-Woodyard conditions for maximum
directivity in the 0 u = direction as

180
0
2.94
| | , | |
N N
u
u
t
t
=
=
= ~ ~ . (14.20)

(b) Maximum at 180 u =

180
180
0
2.94
2.94
2.94
2
N
kd
N
kd
N
u
u
u

=
=
=
=
= +
= +

(14.21)
In order to have a minimum of the pattern in the 0 u =

direction, we must
ensure that

0
| |
u
t
=
~ . (14.22)
We can now summarize the Hansen-Woodyard conditions for maximum
directivity in the 180 u = direction as

180 0
2.94
| | , | |
N N
u u
t
t
= =
= ~ ~ . (14.23)
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 that
the complementary requirement | | t = at the opposite direction can be met if
the following relation is observed:

1
4
N
d
N

| |
~
|
\ .
. (14.24)
If N is large, / 4 d ~ . Thus, for a large uniform array, Hansen-Woodyard
condition can yield improved directivity only if the spacing between the array
elements is approximately / 4 .

Nikolova 2012
6
ARRAY FACTORS OF A 10-ELEMENT UNIFORM-AMPLITUDE HW EFA

Solid line: / 4 d =
Dash line: / 2 d =
N = 10
kd
N
t
|
| |
= +
|
\ .

Fig. 6.12, p. 273, Balanis
Nikolova 2012
7
2. Directivity of a linear array
2.1. Directivity of a BSA

2
2
2
sin cos
sin 2
( )
cos
2
n
N
kd
Z
U AF
N
Z
kd
u
u
u
(
| |
|
(
| |
\ .
( = = =
|
\ . (
(

(14.25)

0 0
0
4
rad av
U U
D
P U
t = = , (14.26)
where / (4 )
av rad
U P t = . The radiation intensity in the direction of maximum
radiation / 2 u t = in terms of
n
AF is unity:

0 max
( / 2) 1 U U U u t = = = = ,

1
0 av
D U

= . (14.27)
The radiation intensity averaged over all directions is calculated as

2
2
2
2
0 0 0
sin cos
1 sin 1
2
sin sin
4 2
cos
2
av
N
kd
Z
U d d d
N
Z
kd
t t t
u
u u | u u
t
u
| |
|
\ .
= =
} } }
.
Change variable:
cos sin
2 2
N N
Z kd dZ kd d u u u = = . (14.28)
Then,

2
2
2
1 2 sin
2
Nkd
av
Nkd
Z
U dZ
N kd Z

| |
=
|
\ .
}
, (14.29)

2
2
2
1 sin
Nkd
av
Nkd
Z
U dZ
Nkd Z

| |
=
|
\ .
}
. (14.30)
Nikolova 2012
8
The function
1 2
( sin ) Z Z

is a relatively fast decaying function as Z increases.

That is why, for large arrays, where / 2 Nkd is big enough
( )
20 > , the integral
(14.30) can be approximated by

2
1 sin
av
Z
U dZ
Nkd Z Nkd
t

| |
~ =
|
\ .
}
, (14.31)

0
1
2
av
Nkd d
D N
U t
| |
= ~
|
\ .
= . (14.32)
Substituting the length of the array
( )
1 L N d = in (14.32) yields

0
2 1
N
L d
D
d
| || |
= +
| |
\ .\ .
_
. (14.33)
For a large array
( )
L d > ,

0
2 / D L ~ . (14.34)

2.2. Directivity of ordinary EFA
Consider an EFA with maximum radiation at 0 u = , i.e., kd | = .

( )
( )
2
2
2
sin cos 1
sin
2
( )
cos 1
2
n
N
kd
Z
U AF
N
Z
kd
u
u
u

(

(
| |

= = =
`
|
(
\ .

(


)
, (14.35)
where (cos 1)
2
N
Z kd u = . The averaged radiation intensity is

2
2 2
0 0 0
1 sin 1 sin
sin sin
4 4 2
rad
av
P Z Z
U d d d
Z Z
t t t
u u | u u
t t
| | | |
= = =
| |
\ . \ .
} } }
.
Since
(cos 1)
2
N
Z kd u = and sin
2
N
dZ kd d u u = , (14.36)
Nikolova 2012
9
it follows that
2 / 2
0
1 2 sin
2
Nkd
av
Z
U dZ
Nkd Z

| |
=
|
\ .
}
,

2 / 2
0
1 sin
Nkd
av
Z
U dZ
Nkd Z
| |
=
|
\ .
}
. (14.37)
If ( Nkd ) is sufficiently large, the above integral can be approximated as

2
0
1 sin 1
2
av
Z
U dZ
Nkd Z Nkd
t

| |
~ =
|
\ .
}
. (14.38)
The directivity then becomes

0
1 2
4
av
Nkd d
D N
U t
| |
~ = =
|
\ .
. (14.39)
Comparing (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 D
0
in terms of the array
length L = (N1)d:

0
4 1
L d
D
d
| || |
= +
| |
\ .\ .
. (14.40)
For large arrays, the following approximation holds:

0
4 / if D L L d = > . (14.41)

2.3. Directivity of HW EFA
If the radiation has its maximum at 0 u = , then the minimum of
av
U is
obtained as in (14.13):

2
min min
min
min
min min
1 2 cos(2 ) 1
Si(2 )
2 sin 2 2
av
Z Z
U Z
k Nd Z Z
t ( (
= + +
( (

, (14.42)
where
min
1.47 / 2 Z t = ~ .
Nikolova 2012
10

2
min
1 2 0.878
1.8515
2 2
av
U
Nkd Nkd
t t
t
| | (
= + =
|
(
\ .
. (14.43)
The directivity is then

0
min
1
1.789 4
0.878
av
Nkd d
D N
U
(
| |
= = =
|
(
\ .

. (14.44)
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

0
1.789 4 1 1.789 4
L d L
D
d
( (
| || | | |
= + =
| | |
( (
\ .\ . \ .

. (14.45)

Example: Given a linear uniform array of N isotropic elements (N = 10), find
the directivity
0
D if:
a) 0 | = (BSA)
b) kd | = (Ordinary EFA)
c) / kd N | t = (Hansen-Woodyard EFA)
In all cases, / 4 d = .

a) BSA

( )
0
2 5 6.999 dB
d
D N

| |
~ =
|
\ .

b) Ordinary EFA

( )
0
4 10 10 dB
d
D N

| |
~ =
|
\ .

c) HW EFA

( )
0
1.789 4 17.89 12.53 dB
d
D N

(
| |
~ =
|
(
\ .



Nikolova 2012
11

3. Pattern characteristics of linear uniform arrays - recapitulation
A. Broad-side array
NULLS ( 0
n
AF = ):
arccos
n
n
N d

u
| |
=
|
\ .
, where 1, 2,3, 4,... n = and , 2 ,3 ,... n N N N =
MAXIMA ( 1
n
AF = ):
arccos
n
m
d

u
| |
=
|
\ .
, where 0,1, 2,3,... m =
HALF-POWER POINTS:

1.391
arccos
h
Nd

u
t
| |

|
\ .
= , where
d
1
t

<
HALF-POWER BEAMWIDTH:

1.391
2 arccos , 1
2
h
d
Nd
t t
u
t
(
| |
A =
| (
\ .

<
MINOR LOBE MAXIMA:

2 1
arccos
2
s
s
d N

u
+ (
| |

|
(
\ .

= , where 1, 2,3,... s = and
d
1
t

<
FIRST-NULL BEAMWIDTH (FNBW):
2 arccos
2
n
Nd
t
u
(
| |
A =
| (
\ .


FIRST SIDE LOBE BEAMWIDH (FSLBW):

3
2 arccos , 1
2 2
s
d
Nd
t t
u

(
| |
A =
| (
\ .

<



Nikolova 2012
12

B. Ordinary end-fire array
NULLS ( 0
n
AF = ):
arccos 1
n
n
N d

u
| |
=
|
\ .
, where 1, 2,3,... n = and , 2 ,3 ,... n N N N =
MAXIMA ( 1
n
AF = ):
arccos 1
n
m
d

u
| |
=
|
\ .
, where 0,1, 2,3,... m =
HALF-POWER POINTS:

1.391
arccos 1
h
Nd

u
t
| |
=
|
\ .
, where
d
1
t

<
HALF-POWER BEAMWIDTH:

1.391
2arccos 1 , 1
h
d
Nd
t
u
t
| |
A =
|
\ .
<
MINOR LOBE MAXIMA:

( )
2 1
arccos 1
2
s
s
Nd

u
+ (
=
(

, where 1, 2,3,... s = and 1
d t

<
FIRST-NULL BEAMWIDTH:
2arccos 1
n
Nd

u
| |
A =
|
\ .

FIRST SIDE LOBE BEAMWIDH:

3
2arccos 1 , 1
2
s
d
Nd
t
u

| |
A =
|
\ .
<




Nikolova 2012
13

C. Hansen-Woodyard end-fire array
NULLS:
( )
arccos 1 1 2
2
n
n
Nd

u
(
= +
(

, where 1, 2,... n = and , 2 ,... n N N =
MINOR LOBE MAXIMA:
arccos 1
s
s
Nd

u
| |
=
|
\ .
, where 1, 2,3,... s = and 1
d t

<
SECONDARY MAXIMA:
arccos 1 [1 (2 1)]
2
m
m
Nd

u

= + +
`
)
, where 1, 2,... m = and 1
d t

<
HALF-POWER POINTS:
arccos 1 0.1398
h
Nd

u
| |
=
|
\ .
, where 1, -large
d
N
t

<
HALF-POWER BEAMWIDTH:
2arccos 1 0.1398
h
Nd

u
| |
A =
|
\ .
, where 1, -Large
d
N
t

<
FIRST-NULL BEAMWIDTH:
2arccos 1
2
n
Nd

u
| |
A =
|
\ .








Nikolova 2012
14

4. 3-D characteristics of a linear array
In the previous considerations, it was always assumed that the linear-array
elements are located along the z-axis, thus, creating a problem, which is
symmetrical around the z-axis. If the array axis has an arbitrary orientation, the
array factor can be expressed as

( )( ) ( ) 1 cos 1
1 1
N N
j n kd j n
n n
n n
AF a e a e
| +
= =
= =

, (14.46)
where
n
a is the excitation amplitude and cos kd | = + .
The angle is subtended between the array axis and the position vector to
the observation point. Thus, if the array axis is along the unit vector

a,

sin cos sin sin cos
a a a a a
u | u | u = + + a x y z (14.47)
and the position vector to the observation point is

sin cos sin sin cos u | u | u = + + r x y z (14.48)
the angle can be found as
cos sin cos sin cos sin sin sin sin cos cos ,
a a a a a
u | u | u | u | u u = = + + a r x y z
cos sin sin cos( ) cos cos
a a a
u u | | u u = + . (14.49)
If
( 0 )
a
u = = a z , then cos cos , u u = = .