FREQUENCY INDEPENDENT ANTENNAS

INTRODUCTION
In antenna scale modelling, characteristics such as impedance, pattern, polarization
and so forth are invariant to a change of the physical size if a similar change is also
made in the operating frequency or wavelength. For example, if all the physical
dimensions are reduced by a factor of two, the performance of the antenna will remain
unchanged if the operating frequency is increased by a factor of two. In other words,
the performance remains unchanged if the electrical dimensions remain unchanged.
The scaling characteristics of antenna model measurements also indicate that if the
shape of the antenna were completely specified by the angles, its performance would
have to be independent of frequency. The infinite bi-conical dipole is one such
structure. To make infinite structures more practical, the designs usually require that
the current on the structure decrease with distance away from the input terminals.
After a certain point the current is negligible, and the structure beyond that point to
infinity can be truncated and removed. Practically then the truncated antenna has a
lower cut-off frequency above which its radiation characteristics are the same as those
of infinite structure. The lower cut-off frequency is that for which the current at the
point of truncation becomes negligible. The upper cut-off is limited to frequencies for
which the dimensions of the feed transmission line cease to look like a “point” (usually
about 2 8 where 2 is the wavelength at the highest desirable frequency). Practical
bandwidths are often of about 40:1. Even higher ratios i.e, 1000:1) can be achieved in
antenna design but they are not necessary, since they would far exceed the
bandwidths of receivers and transmitters.
Evan though the shape of the bi-conical antenna can be specified by angles, the
current on its structure does not diminish with distance away from the input
terminals, and its pattern does not have a limiting form with frequency. By examining
the current distribution of a bi-conical antenna it is evident that that there are phase
but no amplitude variations with the radial distance r. Thus the bi-conical structure
cannot be truncated to form a frequency independent antenna. In practice, however,
antenna shape exist which satisfy the general shape equation to have frequency
independent characteristics in pattern, impedance, polarization and so forth and with
current distribution which diminishes rapidly.

EQUIANGULAR SPIRAL ANTENNA

The equiangular spiral is one geometrical configuration whose surface can be
described by angles. It thus fulfils all the requirements for shape that can be used to
design frequency independent antennas.

PLANAR SPIRAL ANTENNA

The system is composed of the two conducting arms, P and Q, constitutes a balanced
system, and is shown in figure. The antenna excited at the origin by a voltage applied
between the two arms in such a manner that the arms are balanced to ground. The
current flow outward along the spiral arms with small attenuation until a region of
certain size in wavelengths is reached. In this region, known as the active or radiating
region, essentially all of the power guided along the spiral arms is radiated. Beyond
this region the presence or absence of the arms is of little consequence. Because the
radiating region has fixed dimensions in wavelengths it moves inward or outward as
the frequency is raised or lowered. Consequently the size of the effective radiating
aperture automatically adjusts or scales with frequency to produce an antenna that
has the same pattern and impedance at all frequencies. Because of the spiralling of
the arms the radiation pattern actually rotates about the axis of the antenna as
frequency changed.

There are two different versions of spiral antenna. One antenna would consist of two
metallic arms suspended in free-space, as shown in figure above, and the other of a
spiralling slot on a large conducting plane, as shown in figure below. The second is
also usually tapered to provide better matching termination. The slot antenna is most
practical, because it can be conveniently fed by a balanced coaxial arrangement to
maintain its overall balancing. Spiral slot antennas, with good radiation
characteristics, can be built with one-half to three turns. The most optimum design
seems to be that with 1.25 to 1.5 turns with an overall length equal to or greater than
one wavelength. The rate of expansion should not exceed about 10 per turn.
 The planar equiangular spiral antenna is bidirectional, radiating a broad lobed beam
on each side of the plane. As long as the expansion rate of the spiral is not too rapid,
the field is circularly polarized and the pattern is approximately cos    where  is the
angle measured from the axis.
For a fixed cut the beam width vary with frequency since pattern rotates. Typical
variations are on the order of 10o. In general, however slot antennas with more broad
arms and / or more tightly wound spirals exhibit smother and more uniform patterns
with smaller variations in beam width with frequency. For symmetrical structure the
pattern is also symmetrical with no tilt to the lobe structure.
To maintain the symmetrical characteristics, the antenna must be fed by an
electrically and geometrically balanced line. One method that achieves geometrical
balancing requires that the coax is embedded into one of the arms in the spiral. To
maintain symmetry, a dummy cable is usually placed into the other arm. If the feed
line is electrically unbalanced, a balun must be used. This limits the bandwidth of the
system.
The polarization of the radiated wave is controlled by the length of the arms. For very
low frequencies, such that the total arm length is small compared to the wave length,
pattern is the radiated field is linearly polarized. As the frequency increases, the wave become
unaltered as elliptically polarized and eventually achieves circular polarization.
its arms Since the pattern is essential unaltered through this frequency range, the polarization
change in
size to adjust change with frequency can be used as a convenient criterion to select the lower cut-off
changes in frequency of the useable band width. In many practical cases, this is chosen to be the
frequency to point where the axial ratio is equal to or less than 2 to 1, and it occurs typically when
keep it
the overall arm length is about one wavelength. The off-axis radiated field has nearly
independent
of frequency circular polarization over a smaller part of the band width.
In addition to the limitation imposed on the bandwidth by the over all length of the
arms, another critical factor that can extend or reduce the band width is the
construction precision of the feed.
The input impedance of a balanced equiangular slot antenna converges rapidly as the
frequency is increased, and it remains reasonably constant for frequencies for which
the arm length is greater than about one wavelength.
For slot antennas radiating in free-space, without dielectric material or cavity backing,
typical measured efficiencies are about 98% for arm length equal to or greater than
one wavelength. Rapid decreases are observed for shorter arms.

CONICAL SPIRAL ANTENNA

A conical spiral metal strip antenna is shown below. It is a balanced structure which
may be fed (at the apex) by means of a balanced transmission line carried up inside
along the axis of the cone. Alternatively it may be fed by a coaxial cable carried along
and soldered in contact with one of the arms. This type of feed automatically provides
a frequency – independent balun, permitting the balanced antenna to be fed by means
of an unbalanced coaxial line. To maintain physical symmetry a dummy cable is
usually soldered to the other arm. Conical equiangular or log-spiral antennas have
been constructed to operate over band width with a frequency range of over more than
40 to 1. The upper usable frequency is determined by the truncated region at the apex
which must remain small in terms of wavelengths. The lower usable frequency is set
by the cone base diameter which must be at least 3λ/8 at the lowest frequency of
operation for fairly tightly wrapped spirals.

A distinct difference between the planar and conical spirals is that the latter provides
unidirectional radiation (single lobe) toward the apex of the cone with the maximum
along the axis. Circular polarization and relatively constant impedances are preserved
over large bandwidths. Smoother patterns have been observed for unidirectional
designs. Conical spirals can be used in conjunction with a ground plane, with a
reduction in bandwidth when the flash mounted on the plane.

LOG-PERIODIC ANTENNA

PLANAR AND WIRE SURFACES

A planar log-periodic structure is shown in figure below. It consists of a metal strip
whose edges are specified by the angle  2 . However, in order to specify the length
from the origin to any point on the structure, a distance characteristic must be
included. Therefore it is not truly frequency independent.
A typical log-periodic antenna configuration is shown in figure below. The pattern is
unidirectional toward the apex of the cone formed by the two arms, and it is linearly
polarized. Although the pattern of this and other log-periodic structures are not
completely frequency independent, the amplitude variations of certain designs are very
slight. Thus practically they are frequency independent.

The fields on the conductor of the log-periodic surface structure of the figure (a)
attenuated very sharply with distance. This is due to strong current concentration at
or near the edge of the conductors. Thus removing part of the inner surface to form a
wire antenna as shown in figure (b) should not seriously degrade the performance of
the antenna.

The geometries of the above figure use uniform periodic teeth, we define the geometric
ratio of the log-periodic structure by
  Rn Rn1 (23)
and the width of the antenna slot by
  rn Rn1 (24)
The geometric ratio  defines the period of operation. For example, if two frequencies
f1 and f2 are one period apart, they are related to the geometric ratio  by
  f1 f2 where f2 >f1 (25)

If the wires or the edges of the plates are linear (instead of curved), the geometries of
the above figure (a) and (b) reduce, respectively, to the trapezoidal tooth log-periodic
structures of the figure above. These simplifications result in more convenient
fabrication geometries with no loss in operational performance.

DIPOLE ARRAY
A basic log-periodic structure is shown in figure (a) below. Although this antenna has
similar directivities as the Yagi-Uda array (7-12 dB), they are achievable and
maintained over much wider bandwidths. There are, however, major differences
between them.
While the geometrical dimensions of the Yagi-Uda array elements do not follow any set
pattern, the lengths  ln  , spacing Rn  , diameters  dn  and even gap spacing at dipole
centres  sn  of the log-periodic array increase logarithmically as defined by the inverse
of the geometric ratio  . That is,
1 ln 1 Rn 1 dn 1 sn 1
    (26a)
 ln Rn dn sn
Another parameter that is usually associated with a dipole array is the spacing factor
 defined by
Rn 1  Rn
 (26b)
2ln 1
straight lines through the dipole ends meet to form an angle 2 which is a
characteristic of frequency independent structure.
Because it is usually very difficult to obtain wires or tubing of many different
diameters and to maintain tolerance of very small gap spacing, constant dimensions in
these can be used. These relatively minor factors will not sufficiently degrade the
overall performance.
While only one element of the Yagi-Uda array is directly energized by the feed line,
while the others operate in a parasitic mode, all the elements of the log-periodic array
are connected. There are two basic methods, as shown in figure (b) and (c) below,
which could be used to connect and feed elements of a log-periodic dipole array. In
both cases the antenna is fed at the small end of the structure.

The current in the elements of figure (b) have the same phase relationship as the
terminal phases. If in addition the elements are closely spaced, the phase progression
of the currents is to the right. This produces an end-fire beam in the direction of
longer elements and interference effects to the pattern result.
It was recognized that by mechanically crisscrossing or transposing the feed between
adjacent elements, as shown in figure (c), a 180o phase is added to the terminals of
each element. Since the phase between the adjacent closely spaced short elements is
almost in opposition, very little energy is radiated by them and their interference
effects are negligible. However, at the same time, the longer and larger spaced
elements radiate. The mechanical phase reversal between these elements produces a
phase progression so that the energy is beamed end-fire in the direction of the shorter
elements. The most active elements for this feed arrangement are those that are near
resonant with a combined radiation patter toward the vertex of the array. The feed
arrangement of figure (c) is convenient provided the input feed line is a balanced line
like two-conductor transmission line.
If the geometrical pattern of the log-periodic array is to be maintained to achieve a
truly log-periodic configuration, an infinite structure would result. However, to be
useful as a practical broadband radiator, the structure is truncated at both ends. This
limits the frequency of operation to a given bandwidth. The cut-off frequencies of the
truncated structure can be determined by the electrical lengths of the longest and
shortest elements of the structure. The lower cut-off frequency occurs approximately
when the longest element is  2 ; however, the high cut-off frequency occurs when the
shortest element is nearly  2 when the active region is very narrow. Usually it
extends beyond that element. The active region of log-periodic dipole array is near the
elements whose lengths are nearly or slightly smaller than  2 . The role of the active
elements is passed from the longer to the shorter elements as the frequency increases.
Also the energy from the shorter active elements travelling toward the longer inactive
elements decreases very rapidly so that a negligible amount is reflected from the
truncated end. The movement of the active region of the antenna, and its associated
phase centre, is an undesirable characteristic in the design of feeds for reflector and
lens antennas.
The region of constant voltage along the structure is referred to as the transmission
region, because it resembles that of a matched transmission line. Along the structure,
there is about 150o phase change for every  4 free-space length of transmission line.
This indicates that that the phase velocity of the wave travelling along the structure is
vp  0.6c , where c is the free-space velocity. The smaller velocity results from the shunt
capacitive loading of the line by the smaller elements. The loading is almost constant
per unit length because there are larger spacing between the longer elements.
The radiated wave of a single log-periodic dipole array is linearly polarized, and it has
horizontal polarization when the plane of the antenna is parallel to the ground.
Bidirectional patterns and circular polarization can be obtained by phasing multiple
log-periodic dipole arrays. For these, the overall effective phase centre can be
maintained at the feed.
If the input impedance of a log-periodic antenna is plotted as a function of frequency,
it will be repetitive. However, if it is plotted as a function of logarithm of the frequency,
it will be periodic (not necessarily sinusoidal) with each cycle being exactly identical to
the preceding one. Hence the name log-periodic, because the variations are periodic
with respect to the logarithm of the frequency. A typical variation of the impedance as
a function of frequency is shown in the figure above. Other parameters that undergo
similar variations are the pattern, directivity, beam width and side lobe level.

The periodicity of the structure does not ensure broadband operation. However, if the
variations of the impedance, pattern, directivity and so forth within one cycle are made
sufficiently small and acceptable for the corresponding bandwidth of the cycle,
broadband characteristics are ensured within acceptable limits of variation. The total
bandwidth is determined by the number of repetitive cycles for the given truncated
structure. The relative frequency span  of each cycle is determined by the geometric
ratio, which gives
  ln  f2   ln  f1   ln 1/   (27)
Typical design of log-periodic dipole arrays have apex half angle of 10o    45o and
0.95    0.7 . There is a relation between  and  . As  increase, the corresponding 
value decreases, and vice versa. Larger values of  or smaller values of  require a
larger number of elements that are close together. For this type of a design, there are
more elements in the active region which are nearly  2 . Therefore the variations of
the impedance and other characteristics as a function of frequency are smaller,
because of the smoother transition between the elements, and the gains are larger.
There are other configurations of log-periodic dipole array designs, including those
with v instead of linear elements. This array provides moderate bandwidths with good
directivities at the higher frequencies, and it is widely used as a single TV antenna
covering the entire frequency spectrum from the lowest VHF channel to the highest
UHF.
The general configuration of a log-periodic array is described in terms of the design
parameters ,  and  related by
1   
  tan1   (28)
 4 
Once two of them are specified, the other can be found. Directivity (in dB) contour
curves as a function of  for various values of  are shown in figure below.
While the bandwidth of the system determines the lengths of the shortest and longest
elements of the structure, the width of the active region depends on the specific
design. The bandwidth of the active region Bar is given by
Bar  1.1  7.7 1    cot    (29)
2

In practice a slightly larger bandwidth Bs  is usually designed than that which is
required (B). The two are related by
Bs =BBar  B 1.1  7.7 1    cot    
2
(30)
 
where Bs : Designed bandwidth
B: desired bandwidth
Bar : Active region bandwidth
The total length of the structure L, from the shortest lmin  to the longest  lmax  element,
is given by
max  1
L  1   cot    (31a)
4  Bs 
where
max  2lmax  v / fmin (31b)
from the geometry of the system, the number of elements are determined by
ln Bs 
N  1 (32)
ln 1/ 
The centre to centre spacing s of the feeder line conductors can be determined by
specifying the required input impedance (assumed to be real), and the diameter of the
dipole elements and the feeder line conductors. To accomplish this, we first define an
average characteristic impedance of the elements given by
  l  
Za  120 ln  n   2.25 (33)
  dn  
where ln / dn is the length-to-diameter ratio of the nth element of the array. For an ideal
log-periodic design, this ratio should be the same for all the elements of the array.
Practically, however, the elements are usually divided into one, two, three or more
groups with all the elements in each group having the same diameter but not the same
length. The number of groups is determined by the total number of the array. Usually
three groups (for the small, middle and the large elements) should be sufficient.

The effective loading of the dipole elements on the input line is characterized by the
graphs shown in the figure above, where
    =Relative mean spacing
Za =Average characteristic impedance of the elements
Rin =Input impedance (real)
Z0 =characteristic impedance of the feeder line
The centre to centre spacing s between the two rods of the feeder line, each of identical
diameter d, is determined by
s  dcosh  Z0 /120  (34)
A design procedure is outlined here, based on the equations introduced above and
assuming that the directivity (in dB), input impedance Rin (real), diameter of the
elements of the feeder line (d), and the lower and the upper frequencies B  fmax / fmin  of
the bandwidth are specified. It then proceeds as follows:
1. Given D0 , determine  and  from the figure.
2. Determine  using (28)
3. Determine Bar using (29) and Bs using (30)
4. Find l using (31) and N using (32).
5. Determine Za using (33) and    
6. Determine Z0 /Rin using figure above.
7. find s using (34).
ARRAY THEORY OF LOG-PERIODIC AND FREQUENCY INDEPENDENT ANTENNAS
The formulation of an array theory for log-periodic or frequency independent
structures could be expected to be most difficult because all of the usual parameters,
such as element length in wavelengths, element spacing in wavelengths, magnitude
and phasing of element currents vary along the array, and of course are also a
function of frequency. Fortunately a method exists for arriving at a first approximation
of array behaviour. This method consists of viewing the log-periodic structure as a
locally periodic structure whose period varies slowly, increasing linearly with distance
to the apex. From this view point, analysis of a uniform periodic structure as a
function of frequency corresponds approximately to an examination of the log-periodic
structure as a function of distance from the apex. Using this approach, the standard
uniform array theory can be used directly to give a first approximation to the
performance of log-periodic and frequency independent structures. Let us consider the
figure below. It is helpful to consider separately three main regions.

Transmission line region:

In this region the antenna elements are short compared with the resonant lengths
l  / 2 so that elements present relatively high capacitive impedance. The element
current is small and leads the base voltage supplied by the transmission line by
approximately 90o. The element spacing is small in wavelengths and the phase
reversal introduced by the transposition of the transmission line means that the
adjacent elements are nearly 180o out of phase. More precisely, each element current
leads the preceding element currents approximately by     d , where d is the
element separation and   2 /    / v is the phase shift constant along the line. In
general, ,  and v will differ from their free space values owing to the loading effect of
the elements on the transmission line. Because of the phasing and close spacing of the
elements, radiation from this region will be very small and in the backfire direction.

Active region:
In this region the element lengths approach the resonant length (l slightly less than
 / 2 ) , so the element impedance has an appreciable resistive component. The element
current is large and more nearly in phase with the base voltage; the current is slightly
leading just below resonance and slightly lagging just above resonance. The element
spacing is now sufficiently large that the phase of current in a given element leads that
in the preceding element by an angle     d which may approximate  / 2 radians.
This combination of condition will produce a strong radiation in the backfire direction.

Reflection region:
In this region the elements length are greater than the resonant length l   / 2 , so the
element impedance becomes inductive and the element currents lag the base voltage.
The base voltage provided by the transmission line is now quite small because in a
properly designed array, nearly all of the energy transmitted down the line has been
abstracted and radiated by the active region. The element spacing may now be greater
than  / 4 , but the phase shift per unit length along the line in this region is small so
that the resulting phasing between the elements (including the phase reversal
introduced by the transposition) is such that any small amount of radiation is still in
the backfire direction. In addition, the transmission line characteristic impedance
becomes reactive in this region. This means that any small amount of incident energy
transmitted through the active region is not accepted in not reflection region but is
reflected back towards the source.

THE ARRAY AS A LOADED TRANSMISSION LINE:

Some of the remarkable properties of log-periodic and frequency-independent
antennas are attributable to the propagation characteristics of the equivalent loaded
transmission line which conveys energy from the source to the radiating portion of the
antenna. These effects are particularly easy to see in the case of the log-periodic dipole
array, and are indicated in the figure above. On the feed line to the antenna, region
(0), the series inductance and shunt capacitance per unit length are shown as l and c,
respectively. In the transmission region of the antenna, region (1), the transmission
line is loaded by a capacitance per unit length, Ca , which represents the loading effect
of the short dipoles which have a capacitive reactance. It is noted that to the first
approximation Ca is nearly constant throughout its region because at the beginning of
the region the capacitance per element is small, but the element are closely spaced,
whereas near the end of the region the capacitance per element is larger, but so is the
spacing. The effect of augmented shunt capacitance of the line  C  Ca  is to increase
the phase delay per unit length, and since   2 /    / v , a decrease of wavelength 
and a decrease of phase velocity v along the below the free space values result. This is
said to be “slow-wave” region of the transmission line. Note, however, that because of
the transposition of the feed line between elements, successive elements are fed with a
leading phase shift of      per section. This rapid phase shift in the reverse direction
corresponds to a slow wave in the backward direction along the antenna elements.
In the active region (2), the element lengths approach the resonant length and the
transmission line loading becomes resistive, designated by the shunt resistance Ra in
series with the antenna capacitance Ca and antenna inductance La . The phase shift
per unit length, the wavelength and phase velocity all approach their free space
values. Because of the transposition between elements, and accounting for the fact
that the element current leads the base voltage by lesser amounts in successive
elements as the resonant length is approached, it turns out that phasing of currents
in the elements corresponds to a backward travelling wave having a velocity v
somewhat less than c, the velocity of light.
In the inactive or reflection region (3), the element lengths become longer than
resonant wavelength, the antenna inductive reactance predominates, and the loading
effect on the line is represented by the shunt inductance La . When (and if) the parallel
combination of La and C is inductive, we have the equivalent of the attenuation region
of a filter. The phase shift per unit length is then zero (for the lossless case) and the
phase velocity is infinite (that is there is no wave motion). The incident energy
propagating down the line is no longer accepted but is reflected back toward the
source.

DIRECTIVITY AND IMPEDANCE:

The radiation characteristics of the log-periodic dipole array are functions of the array
geometry as expressed by the factor  and the angle  ; the input impedance depends
mainly on the characteristic impedance , Z0 , of the transmission line feeding the
elements. The gain of a well designed array ranges from 7.5 to 12 dB (over an isotropic
radiator) with the larger gain values associated with the smaller angles  , that is ,
longer arrays. The input impedance can be made to have any value from about 50 to
200 Ohms.