Linear Arrays

Introduction 28/02/2000
All the individual radiators of an array are usually similar, with the most common array element being the
half-wave dipole !he radiation pattern of an array in free-space depends on four factors"
!he relative positions of the individual radiators with respect to each other
!he relative phases of the currents of fields in them
!he relative magnitudes of the individual radiator currents or fields
!he patterns of the individual radiators
!he basic theory of arrays is developed in terms of the first three factors !he fourth factor assumes that the
individual radiators are fictional isotropic point sources An isotropic point source is one that radiates with
uniform intensity in all directions and has no physical si#e and also no $electrical% si#e, hence, it does not
bloc& or otherwise affect the radiation of the other elements of the array An array radiation pattern can be
calculated on the basis of these assumptions and then a correction to it can be made to ta&e into account
that in reality the individual radiators do affect each other and do not radiate isotropically
'efinition of Antenna (attern
If an antenna is imagined to be located at the centre of a spherical coordinate system, its radiation pattern is
determined by measuring the electric field intensity over the surface of a sphere at some fi)ed distance r
*ince the field + is then a function of the two variables and , it is written +,, -
A measurement of the electric field intensity +,, - of an electromagnetic filed in free space is e.uivalent
to a measurement of the magnetic field intensity /,, -, since the magnitudes of the two .uantities are
directly related by the e)pression + 0 122/ 3ectorially + and / are at right angles to each other and their
phase angles are e.ual !herefore, the pattern could e.ually well be given in terms of + or / It is
customary, however, to discuss patterns in terms of the electric field intensity +
!he power density of the field (,, -, can also be computed when +,, - is &nown !he relation being ( 0
+
2
/122 !herefore a plot of the antenna pattern in terms of (,, - conveys the same information as a plot of
the magnitude of +,, -
Assuming free-space propagation, a pattern that represents field strength as a function of angular direction
at a fi)ed distance from the antenna is identical to a plot of distance for constant field strength !herefore, a
field-strength pattern can be interpreted in either of these ways, by simply changing the labelling of the
radial coordinate scale A similar statement, however, is not true for power-density patterns
4elative pattern
5ften the pattern is plotted in relative terms, that is, the field strength or power density is represented in
terms of its ratio to some reference value !he reference usually chosen is the field level in the ma)imum-
field-strength direction !he field strength or power density is given the value unity in this direction and
fractional values in other directions !his relative pattern provides as much information about the antenna
as does an absolute pattern and therefore relative patterns are usually plotted when it is desired to describe
only the properties of the antenna, without reference to an associated transmitter or receiver
!wo-Isotropic-element Array
!he simplest array comprises two isotropic point-source radiators It is discussed here because it covers
most of the principles of multi radiator arrays
!he meaning of the term phase as applied to the radiating elements of an array is illustrated by considering
the following6 At some distant point the fields of the two radiating elements are e)amined and are found to
7
be in phase with each other If then the phase of the current in 4adiator 2 is changed by an amount
radians and the phase of the current in 4adiator 7 is left unchanged, it will be observed that the two fields at
the distant point are now out-of-phase by the amount radians !he phase of a radiating element in an
array is always discussed in relation to the phases of the other elements If the phases of the two radiators in
a two-element array are changed by the same amount, the array pattern is unaffected, but changing the
phases between the two elements affects the way that the individual fields of the tow radiators add up at a
specified distant point
8igure 7 shows the significant geometry of a two-element isotropic-point-source array
Figure 1 Array of two isotropic sources
!he coordinates of ( are $r% and are two coordinates of the spherical coordinate system !he third
spherical coordinate , is not shown because only a plane in which is constant is being considered here,
namely, the )y-plane in which
2

radians !he point ( is the field point, that is, any point in space
where the field is to be calculated /ence, an e)pression of the field strength a ( applies to all points and
thus defines the radiation pattern of the array 9alculation of the field due to the array at an arbitrary point
( is the basic problem of array theory :ecause the distances r7 and r2 are much larger than the distance d,
the amplitudes of the separate fields of the two radiators at point ( will be very nearly the same due to the
large attenuation that the radiation from each radiator e)periences
5n the other hand, the relative phases of the two fields at ( arising from the two radiators will be dependent
on , where
2 7
r r , as shown in 8igure 2
Figure 2 Two-element array geometry
2
!he resulting phase difference of the fields due to is e.ual to
2

radians, where 4adiator 7 or 47 is

ta&en to be the reference phase !o this difference must be added the initial phase difference , between the
two radiators themselves !he total phase difference of the two fields at ( is then given by
2

,7-
!he resultant field at ( is the superposition of two fields of e.ual amplitude, +o, and phase difference If
two fields of the same amplitude +o, are considered as two vectors separated by an angle , then the
resultant field +, can be found by using the cosine rule, as
2 cos
2
o
E E

_

,
,2-
!his is the first step toward finding the radiation pattern of the array, which is an e)pression of + as a
function of , the angle of the direction of ( from the centre of the array !o obtain the radiation pattern, it
is necessary to e)press in terms of
8rom figure 2,
sin d ,1-
5n substituting e.uation 1 into e.uation 7 and the resultant bac& into e.uation 2, gives the e.uation for the
field +,- at (
( )
sin
2 cos
2
o
d
E E

1

1
]
,;-
+.uation ; is the e)pression for the field at ( as a function of the angle that the direction of ( ma&es with
the line perpendicular to the line of the array +.uation ; gives the shape of the pattern in the )y-plane or
2

plane !he absolute-value brac&ets are used to indicate that the field intensity being calculated is
proportional to the amplitude or to the rms value and is therefore a positive number, although the
e)pression inside the brac&ets may be signed
In order to obtain the relative pattern, for which the field strength in the ma)imum-intensity direction has
the value unity, e.uation ; must be divided by the ma)imum value of +, which is usually, but not always
2+0
!hree-dimensional pattern
+.uation ; describes the array pattern in the )y-plane, in which the angle of a three-dimensional
coordinate system is constant ,
2

- :ecause the angle is constant it does not appear in e.uation ;

!he three-dimensional pattern is obtained by revolving the )y-pattern about the y-a)is, which is the line of
the array !his is possible as the )y-pattern is identical in shape and si#e at any value of rotation in the y#-
plane !he pattern in the y#-plane is e)pressed as a function of the angle instead of the angle In other
planes, both angles are involved !he e)pression for the complete three-dimensional pattern is given by,
( )
sin sin
, 2 cos
2
o
d
E E

1

1
]
,;a-
(arallel dipole two-element array
!he pattern defined by e.uation ; can be plotted as a function of the angle for different values of the
parameters d and ,<ote that a parameter is defined in mathematics as a .uantity that can vary but is held
constant in a particular problem so that the effect of some other variable may be studied, uncontaminated
by any variation of the parameter- Although the various resulting patterns obtained are for an array of
isotropic elements, they also represent the patterns of an array of two dipoles in the plane perpendicular to
their a)is (roviding that the dipoles are parallel to each other and perpendicular to the line =oining their
1
centres ,line of the array-, as shown in 8igure 1 !his diagram may be thought of as the hori#ontal-plane
patterns of a pair of vertical dipoles or monopoles separated by a distance d and with currents having a
phase difference
Figure 3 Geometrical positioning of pattern for two dipoles
It should be noted that when the two elements are in phase , 0 0-, the radiation is always ma)imum in the
direction perpendicular to the line =oining the elements, that is in the )-direction in 8igure 1 above !hat is,
because the distances from the elements to the field point ( are e.ual in that direction , 0 0-, the phase
difference due to path difference is #ero also6 hence the total phase difference of the superimposed fields is
0 0 9onse.uently, the fields add directly and the ma)imum possible resultant field is obtained >hen
0 0 and the pattern ma)imum is in the direction perpendicular to the array line, as shown in 8igure 1, the
antenna is called a broadside array
8or certain conditions the resultant field in some directions is #ero, that is when the sum of the radiator
phase difference , and the phase difference due to path difference
2

is an odd integral multiple of

radians !he fields of the individual radiators are in this case, of e.ual amplitude and opposite phase, so
they cancel !his occurs, for e)ample, when 0 0 and d 0 /2, in the 0 ?0 and 220 directions, since in
these directions the field phase-difference is 780 degrees, whereas the phase difference due to radiator
phase difference is #ero !he same result occurs when the 780 degrees net phase difference is due to a
combination of path difference and radiator phase difference !hese directions of #ero intensity in a pattern
are called nulls
9ertain combinations of d and result in ma)imum radiation in the direction of the line =oining the array
elements !he array is then said to be operating as an endfire array !he radiation of an endfire array may
be either bidirectional or unidirectional A bidirectional array is where the radiation lobes are in both
directions along the line of the array >hereas, a unidirectional array has a lobe in one direction and a null
in the opposite direction
@ulti-element Aniform Linear Arrays
>hen more than two elements are used in an array, the principle of calculating the pattern ,from which in
turn the beamwidth and directive gain can be computed- is the same as for a two-element array +)cept
that the fields of all the elements must be superposed at the field point !he simplest type of multi-element
array is one in which all the radiators are in a line, with e.ual spacing between ad=acent pairs !his is
shown in 8igure ; below !he method of analysis is suggested by showing a field-point ( =oined by ray
lines to each element *uch an array is called a linear array >hen all the elements are radiating with e.ual
intensity and the phase difference between ad=acent elements is constant, the array is called uniform !he
applet provided demonstrates a uniform linear array of up to nine elements
;
Figure 4 Linear array of four radiating elements
4adiation (attern
If the array contains n isotropic point-source elements, with e.ual spacing d and phase difference
between ad=acent elements, the pattern in a plane containing the line of the array can be shown to be
sin
sin
2
sin
sin
2
rel
d
n
E
d
n

1 _

1
, ]

1
]
,B-
8or n 0 2, this e)pression reduces to e.uation ;
As in the two-element case, the three dimensional pattern is obtained simply by replacing sin by the
product sinsin !he factor n in the denominator of e.uation B is a normali#ing factor that is used to
ma&e +rel a true relative pattern !hat is, it permits +rel 0 7 for the particular values of in the direction of
ma)imum field intensity, in accordance with the definition of relative pattern given above
It is e.uation B that is used in the applet to determine the polar pattern of an n element array
:roadside Linear Arrays
>hen 0 0, all the elements are in phase and pattern ma)ima occur at 0 0 and
0 780, that is, in the directions perpendicular to the line of the array !his configuration is called a
broadside array !he pattern will be ma)imum in these directions regardless of the element spacing, d
!hese will be the only primary ma)ima if d C If d 0 , additional ma)ima occur at 0 ?0 and 0 220
As d is increased still further, additional ma)ima occur as cones of radiation about the a)is of the array
!hey are &nown as grating lobes, analogous to the lobes observed in the optical study of a diffraction or
reflection grating 5rdinarily, therefore, the spacing of elements in a broadside array is &ept less than a
wavelength /owever, there is an advantage in spacing the elements of an array by more than half a
wavelength 8or a two-element array the optimum spacing is about 02, for a four-element array it is about
08 and for a large number of elements the optimum is about 0?B !he directivity increases gradually as
the spacing is increased until the optimum is reached, then drops rather sharply with further increase !he
advantage of the wider spacing is in the directivity obtainable with a given number of elements In terms of
the ratio of the directivity to the total length of the array, there is no advantage in the wider spacing !hese
observations can be seen in the applet Although the above discussion is based on a uniform linear
broadside array of point-source isotropic radiators, they apply also to a similar array of dipoles !hese
B
dipoles have their centres on the array line and their a)es perpendicular to the array line and parallel to each
other !he pattern is in the plane perpendicular to the dipole a)es, that is, the )y-plane as shown in 8igure
1 !he polari#ation of the array follows the direction of the dipoles If the array line is hori#ontal and the
dipole a)es are vertical, the radiation is vertically polari#ed If the array line is hori#ontal and the dipole
a)es are hori#ontal, as shown in 8igure B, the radiation is hori#ontally polari#ed
8igure B shows a broadside array and its polar pattern for n 0 D, d 0 02 and 0 0
Figure 5 Linear roadside array of si! radiating dipoles all elements in p"ase s"owing t"e position
of t"e polar pattern
!he in-phase currents in the individual dipoles, re.uired for a broadside pattern, may be obtained by
properly connecting a branched transmission line to the feed point of each dipole !hat is, if the total line
length from the transmitter to each dipole is the same the dipoles will be fed in phase It is important to
ensure that the same side of the line is connected to the same side of each dipole 4eversing this connection
reverses the phase
:roadside arrays may also b e formed from other types of elements, such as horns, slots, heli)es and
polyrods If the elements are unidirectional radiators, such as sectoral or pyramidal horns, waveguide slots,
a)ial mode heli)es and polyrods, a unidirectional broadside array results

:eamwidth and Eain of :roadside Linear Arrays
Isotropic elements spaced a "alf-wa#elengt"
8or a uniform broadside linear array of n isotropic elements spaced a half-wavelength apart, the beamwidth
is given by
8or n 0 2 D0
8or n 0 1 1D
8or n F1 702/n
!he directivity of such an array is e.ual to n
Isotropic elements spaced other than a half-wavelength
D
>hen the element spacing is not half-wavelength but is not greatly different from this value, the beamwidth
formula becomes
B7
BW
nd

,D-
+ndfire Linear Arrays
If in e.uation B the phase-difference between ad=acent elements is e.ual to
2 d

radians, the
condition for a ma)imum of radiation is satisfied when 0 ?0 and for this value only, provided that d C
/2 !he ma)imum field intensity is radiated in a direction along the line of the array, Goff the endH rather
than off the side /ence the name endfire array. !he ma)imum is toward only one of the ends of the array,
rather than in both the endfire directions
In terms of 8igure ;, if the progressive phase change is a retardation going in the direction of the positive y-
a)is by the amount per element, the beam will be in the 0 ?0 direction If the sign of is changed, or
its amount is increased by 780 degrees without changing the element spacing, the beam will be in the
direction 0 220 Although the condition stated above for the value of results in an endfire array, it does
not result in an endfire pattern with the ma)imum possible directivity and narrowest possible beam /ansen
and >oodyard showed that an endfire beam with a greater gain results if the phase change per element
satisfies
Hansen-Woodyard condition
2 d
n

_
+

,
radians ,2-
!his condition does not necessarily result in a unidirectional pattern as does the basic endfire condition
!he basic endfire analysis assumes isotropic radiators, but it applies also to other radiators that have
radiation along the line of the array !hus, parallel dipoles can be used buy not collinear dipoles 8or
isotropic elements the beam has three-dimensional a)ial symmetry where the hori#ontal and vertical
beamwidths are the same
!he case where d 0 /8 with 0 71B and d 0 /; with 0 ?0 provide a complete null or #ero field
intensity in the bac& direction for a 2 element array !hese patterns are advantageous in applications where
a high front-to-back ratio is re.uired !he condition for a null condition for any uniform linear endfire
array with an even number of elements is given by
Condition for null in one endfire direction
2
7
d

_
t

,
radians ,8-
A null is never absolute because of unavoidable imperfections in the spacing and phasing of the elements
and many-element arrays are more susceptible to such imperfections than those of few elements 8igure D
shows an endfire arrangement
Figure $ Linear endfire array of si! radiating dipoles all elements in p"ase s"owing t"e position of
t"e polar pattern%
2
(arasitically +)cited +ndfire Arrays
It is not necessary to feed each element of an endfire array by direct connection to a transmission line If
only one dipole as such an array is directly fed, or driven the field that it sets up will cause currents to flow
in ad=oining elements !his process is called parasitic e)citation and the elements thus e)cited are parasitic
elements +ndfire arrays employing this principle are &nown as Iagi-Ada antennas, which are often =ust
called Iagis ,(rofessor Ada first described these antennas in Japanese and / Iagi was the first to describe
the antenna in +nglish, giving full credit to Ada%s wor&- !he Iagi-Ada antenna can also be regarded as a
surface-wave antenna
(arasitic e)citation cannot be employed in broadside arrays !o produce the in-phase currents re.uired for a
broadside pattern, a full-wavelength element spacing would be re.uired !here would then be two endfire
lobes of radiation as well as the two broadside lobes !his type of pattern is not a true broadside pattern
!herefore, broadside arrays are always driven arrays rather than parasitic arrays
!he phases of the currents in parasitic dipole elements are determined by their spacing from the ad=acent
element and also by their lengths A parasitic dipole cut e)actly a half wavelength or slightly longer will be
inductive and the phase of its current will lag the induced emf A dipole cut shorter than a half wavelength
will be capacitive and the current in it will lead the induced emf 9omparatively close spacing of elements
is used in parasitic arrays to obtain good e)citation and the induction fields of the elements play a ma=or
role !he e)act analysis is very complicated (roperly spaced dipole elements slightly shorter than a half
wavelength act as directors, reinforcing the field of the driven element in the direction away from the
driven element !hus, a line of directors may be used with each one e)citing the ne)t one 5n the other
hand, an element one-half-wavelength long or slightly longer will act as a reflector, if correctly spaced
!his reflector reinforces the field of the driven element in a direction toward the driven element from the
reflector !herefore, if a reflector element is placed ad=acent to a driven element, another element placed
beyond the reflector will not be appreciably e)cited Although parasitic elements are usually longer or
shorter than a half wavelength when used as reflectors or directors, respectively, they may also be one half-
wavelength long and made to act as reflectors or directors by proper spacing
8or these reasons a Iagi-Ada endfire array usually consist of one driven element, one reflector on one side
of it, and a number of directors on the other side of it Antennas of this type offer the advantages of a
unidirectional beam of moderate directivity with light weight, simplicity of feed design and low cost !he
design becomes critical if high directivity is attempted trough the use of many elements Ap to fie or si)
may be used without difficulty and arrays of thirty or forty elements are possible !he input impedance of a
yagi-Aday array tends ot be low and the bandwidth is limited to around 2K, typically 'irective gain of
around 70 d: is readily achieved with a moderate number of elements, usually five or si) /igher gains
may be achieved by ma&ing a broadside array of which the elements are Iagi-Ada arrays
4eferences
7 !he above theory was ta&en from the e)cellent boo& by, L3 :la&e, GAntennas,H Artech /ouse, Inc
,7?8;- ,I*:< 0-8?00D-7B;-8- ,5riginally published by John >iley L *ons, 7?DD-
2 >> /ansen and J4 >oodyard, GA <ew (rinciple in 'irectional Antenna 'esign,H (roceedings
I4+, 2D, 111-11B ,@arch 7?18-
8