gain ripple in small offset gregorian antennas
TRANSCRIPT
Gain Ripple in Small Offset Gregorian Antennas
Dirk I.L. de Villiers
Department of Electrical and Electronic Engineering
Stellenbosch University
Stellenbosch, 7600, South Africa
Email: [email protected]
Abstract—This paper investigates the effect of sub-reflectordiffraction on gain ripple performance in small clear apertureoffset Gregorian reflector antennas. A spherical mode expansionof an analytically specified feed pattern is used to illuminatea given sub-reflector, and method of moment simulations areperformed to find the resulting radiation pattern. It is foundthat the diffracted rays from the sub-reflector edge can interferewith the main beam to produce a sinusoidal form gain ripple forthe dual reflector system. The relative amplitude of the ripple isfrequency dependent, and can be predicted from the sub-reflectordiffraction results. The ripple frequency can be found from theantenna geometry. The effect on the ripple of changing some ofthe important parameters is investigated, and it is shown thataccurate near field simulations of the feed pattern illuminatingthe sub-reflector is required to produce accurate results.
I. INTRODUCTION
In wide band radio telescope systems, such as the South
African Square Kilometer Array (SKA) [1] pathfinder project
MeerKAT [2], special care must be taken in the design of the
reflector and feed systems to limit gain ripple over frequency.
This effect is known as chromatic aberration which, in optics,
equates to differential diffraction of light of different wave-
lengths. Chromatic aberration becomes a limiting factor when
doing radio frequency spectroscopy of radio sources with radio
telescopes exhibiting gain ripple over frequency. Such gain
ripple is usually caused by feed and receiver mismatches, as
well as effects relating to the feed and reflector configuration.
A thorough treatment of some of these effects is given in
[3], where predictions of gain ripple due to blocking and feed
scattering in prime focus as well as symmetrical secondary
focus fed radio telescopes are presented. Typical techniques for
the reduction of gain ripple in symmetrical reflector systems
include apex ’splash’ plates or cones and the use of support
struts with sharp edges toward the reflector to reduce the
backscattered energy into the feed.
When a clear aperture offset reflector configuration, such
as an offset Gregorian, is used the gain ripple caused by the
multipath backscattered energy from the reflector and struts
can be theoretically completely eliminated. However, since
the sub-reflector in such offset designs is normally relatively
small, the diffraction from the sub-reflector edge can still cause
a significant gain ripple in the reflector system. This paper
investigates this diffraction gain ripple in clear aperture offset
Gregorian reflector systems through full wave simulations of
the sub-reflector as well as the full dual reflector system. It
is shown that increasing the projected spacing between the
reflectors will not lead to a reduced ripple amplitude. Care
must be taken when doing simulations of the feed and reflector
system to include the feed near field effects, as it is shown that
by neglecting these erroneous sub-reflector diffraction results
are obtained which can significantly affect the gain ripple
predictions.
II. GAIN RIPPLE DUE TO SUB-REFLECTOR DIFFRACTION
A symmetry plane cut of the general configuration of a
clear aperture offset Gregorian reflector system is shown in
Fig. 1. Also shown in the figure are two geometric optic (GO)
and diffracted ray paths. The path length from the feed to the
�����
��
��
��
��
�
�
�
���
�
�
���
� �
� �
��
��
���
���
��������������
���
������
Fig. 1. Typical clear aperture offset Gregorian reflector geometry showingsome important dimensions as well as the geometric optics and main beamdiffracted ray paths at the edges of the sub-reflector.
aperture plane of the GO ray is GOn = ρSn + ρMn + ρAn,
and that of the diffracted ray is Dn = ρSn + ρDn. The path
length difference between a GO ray and a diffracted ray in the
broadside direction (θ = 0◦) is therefore
Δn = GOn −Dn = ρMn + ρAn − ρDn, (1)
with n representing any point on the edge of the GO path,
and n = 1, 2 as indicated in Fig. 1. It is therefore expected
that the interference between the GO and diffracted rays in
2172978-1-4244-9561-0/11/$26.00 ©2011 IEEE AP-S/URSI 2011
the broadside direction will give rise to a gain ripple with
the frequency dependent on the average path length difference
Δn. For most configurations the path length differences Δ1
and Δ2 are within a few percent and the average path length
can be approximated as Δn ≈ (Δ1 + Δ2)/2. The gain ripple
frequency can be approximately predicted as
fr ≈ c/Δn, (2)
with c the speed of light.
The relative amplitude of the ripple depends on the actual
diffraction pattern of the feed and sub-reflector combination
which is strongly frequency dependent. Full wave Method
of Moment (MoM) simulations of the sub-reflector were
performed in FEKO (using the Multilevel Fast Multipole
Method), with the feed given by a cosn θ radiation pattern
specified by its spherical mode expansion [4] around the
secondary focus to accurately simulate the near field behavior
of the pattern. The effect of ignoring the near field pattern of
the feed is discussed later in the paper. The radiation pattern
results of a DS = 10λ (λ indicating wavelength) sub-reflector
with the feed pattern linearly polarized in the symmetry plane
of the reflector (x-direction in Fig. 1) and with a -10 dB edge
taper is shown in Fig. 2. The θ = 0◦ angle corresponds to the
-150 -100 -50 0 50 100 150-40
-30
-20
-10
0
10
20
θ°
Dir
ecti
vit
y [
dB
i]
Fig. 2. Symmetry plane radiation pattern of a 10λ sub-reflector illuminatedby a -10 dB edge taper pattern.
z-axis direction of the dual reflector system shown in Fig. 1.
The expected diffraction pattern is clearly visible in Fig. 2 in
the area behind the sub-reflector. A plot showing the frequency
dependence of the directivity in the θ = 0◦ direction of the
same sub-reflector and feed combination is shown in Fig. 3.
The frequency dependence of the z-axis sub-reflector
diffraction pattern directivity suggests that the amplitude of
the gain ripple of the dual reflector system will also be
frequency dependent. This is confirmed in Fig. 4, where the
total efficiency [5] of the dual reflector system (DS = 10λ,
dg = 0, D = 27λ and DM = 30λ) as well as the θ = 0◦
directivity of only the sub-reflector and feed combination are
shown on the same graph. Results are shown for different edge
illuminations, showing that a softer edge illumination will have
less sub-reflector diffraction and therefore a correspondingly
1 1.2 1.4 1.6 1.8 2-24
-22
-20
-18
-16
-14
-12
-10
-8
Normalized Frequency
Dir
ecti
vit
y [
dB
i]
Fig. 3. Directivity of a sub-reflector illuminated by a -10 dB edge taperpattern in the θ = 0◦ direction. The longest dimension of the sub-reflector is10λ at the lowest frequency.
1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4
70
71
72
73
Eff
icie
ncy
[%
]
-10 dB edge taper
-15 dB edge taper
1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4
-25
-20
-15
-10
Dir
ecti
vit
y [
dB
i]
Normalized Frequency1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4
-25
-20
-15
-10
Dir
ecti
vit
y [
dB
i]
Normalized Frequency
Fig. 4. Offset Gregorian system illuminated by -10 dB and -15 dB sub-reflector edge taper patterns. The total efficiency of the system and thebroadside directivity of the sub-reflector pattern is shown to illustrate thedependence of the efficiency ripple on the sub-reflector diffraction patternlevel. The longest dimension of the sub-reflector is 10λ at the lowestfrequency, and the path length difference between the broadside GO anddiffracted rays is 38.1λ.
smaller relative ripple amplitude. The total directivity of the
-15 dB edge illuminated system is, however, lower than that
of the -10 dB case since the reflector is no longer optimally
illuminated. Fig. 4 clearly shows the correspondence between
the ripple amplitude and the sub-reflector diffraction pattern
amplitude in the θ = 0◦ direction. The normalized path length
difference for this configuration is Δn ≈ 38.1188λ which,
using (2), corresponds to a normalized ripple frequency of
0.0262, which is very close to the ripple frequency found with
the full wave simulation.
It should be noted here that all the results shown in this
paper are for the symmetry plane polarization case. Similar
results are also found if the feed is polarized in the plane of
asymmetry, but with smaller sub-reflector diffraction resulting
in a smaller relative ripple.
2173
III. EFFECT OF GAP BETWEEN REFLECTORS
The dimension dg in Fig. 1 is often a free parameter
when designing offset Gregorian/Cassegrain reflector systems.
It represents the projected space between the main- and sub-
reflectors, and is typically greater than or equal to zero in a
clear aperture configuration. In such a configuration changing
dg has no effect on the theoretical GO efficiency of the
antenna since there is no aperture blockage. A larger dg can,
however, have negative mechanical effects on the antenna
structure, since, as can be seen in Fig. 5, a larger dg requires
a larger separation between the main- and sub-reflectors and
correspondingly stiffer arms to support the feed and sub-
reflector assemblies. Close investigation of Fig. 5 also shows
Fig. 5. Three offset Gregorian geometries showing the effect of changingdg . All other dimensions are kept constant (DM = 30λ and DS = 10λ)with dg = [4, 2, 0]λ corresponding, respectively, to the outer, middle andinner geometries.
that changing dg can have a notable effect on the electromag-
netic performance of the antenna system. If all the parameters
are kept constant (the Mizugutch condition is satisfied in all
configurations in this paper [6]) and only dg is varied, the
effect on the sub-reflector in Fig. 5 can be approximated as
a small rotation (change in feed angle γ) of the sub-reflector.
This rotation corresponds to a simple shift in the sub-reflector
diffraction pattern with θ, as is shown in Fig. 6 where the
directivities of the sub-reflector feed combination for different
values of dg are plotted on the same graph. The change in the
angle γ between dg = 0 and dg = 2λ is 1.293◦, and between
dg = 2λ and dg = 4λ is 1.170◦. This is very close to the shift
in diffraction patterns observed in Fig. 6, and therefore gives
a good initial estimate of the effect of changing dg .
Fig. 7 shows the comparison between the directivities of two
cases with different dg values. Note how the ripple amplitude
is similar for both cases, but the position of the peaks and
dips change. If a system is required to have an especially
flat frequency response in a certain narrow important band,
the parameter dg can be adjusted in the design to give a
sub-reflector diffraction pattern which has a dip around the
required frequency. In such a design only the sub-reflector
needs to be analyzed at the required frequency to find the
diffraction pattern with dg = 0, and then dg can be adjusted to
yield a dip in the diffraction pattern in the θ = 0◦ direction by
using the feed angle rotation approximation mentioned earlier.
-70 -60 -50 -40 -30 -20 -10 0 10 20-35
-30
-25
-20
-15
-10
-5
0
5
θ°
Dir
ecti
vit
y [
dB
i]
dg = 0λ
dg = 2λ
dg = 4λ
Fig. 6. Sub-reflector diffraction pattern for varying dg . DS = 10λ and anedge illumination of -10 dB is used in all cases.
1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4
70
71
72
73
Normalized Frequency
Eff
icie
ncy
[%
]
dg = 0λ
dg = 4λ
1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4
-20
-15
-10
Dir
ecti
vit
y [
dB
i]
Normalized Frequency1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4
-20
-15
-10
Dir
ecti
vit
y [
dB
i]
Normalized Frequency
Fig. 7. Offset Gregorian systems with different dg values illuminatedby a -10 dB sub-reflector edge taper pattern. The total efficiency of thesystem and the broadside directivity of the sub-reflector pattern is shownto illustrate the dependence of the efficiency ripple on the sub-reflectordiffraction pattern level. The longest dimension of the sub-reflector is 10λ
at the lowest frequency. The path length difference between the broadsideGO and diffracted rays is approximately 40λ for dg = 0λ and approximately48λ for dg = 4λ.
IV. IMPORTANCE OF NEARFIELD FEED PATTERNS IN THE
ANALYSIS
As previously stated, all the simulations were performed
with a spherical mode expansion of the feed pattern as source.
This allows the near field effects of the feed to be included
in the model, since the feed is often required to be placed
relatively close to the sub-reflector. Fig. 8 shows a comparison
between the spherical mode source and a radiation pattern
point source in the directivity of the illuminated sub-reflector.
Note that in the reflected part of the pattern the results agree
closely, but in the diffracted part of the field, behind the
sub-reflector, the results differ substantially. Therefore, care
should be taken to include the feed near field effects in the
simulations of the feed sub-reflector combination to accurately
2174
-150 -100 -50 0 50 100 150-35
-30
-25
-20
-15
-10
-5
0
5
10
15
θ°
Dir
ecti
vit
y [
dB
i]Spherical Mode Source
Radiation Pattern Source
Fig. 8. Comparison of the spherical mode and radiation pattern sourcedirectivities when used to illuminate a sub-reflector of size 10λ.
model and predict the expected gain ripple of the full system.
A comparison of the full simulation results using spherical
mode and radiation pattern sources is shown in Fig. 9, where
the difference in gain ripple responses is obvious.
1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4
70
71
72
73
Eff
icie
ncy
[%
]
Spherical Mode Source
Radiation Pattern Source
1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4
-25
-20
-15
-10
Dir
ecti
vit
y [
dB
i]
Normalized Frequency1 1.05 1.1 1.15 1.2 1.25 1.3 1.35 1.4
-25
-20
-15
-10
Dir
ecti
vit
y [
dB
i]
Normalized Frequency
Fig. 9. Offset Gregorian system illuminated by a -10 dB sub-reflectoredge taper spherical mode and radiation pattern sources. The total efficiencyof the system and the broadside directivity of the sub-reflector pattern isshown to illustrate the dependence of the efficiency ripple on the sub-reflectordiffraction pattern level. The longest dimension of the sub-reflector is 10λ atthe lowest frequency. The path length difference between the broadside GOand diffracted rays is approximately 40λ, with dg = 0λ, D = 27λ andDM = 30λ.
V. CONCLUSION
Chromatic aberration in radio telescopes can be a limiting
factor when doing radio frequency spectroscopy. A reduction
in gain ripple of the reflector antenna system is required to
alleviate this problem, and clear aperture offset Gregorian an-
tennas can provide this due to the elimination of GO multipath
interference. Sub-reflector diffraction, however, can still cause
significant gain ripple in electrically smaller antennas, and this
effect was investigated in this paper. It was shown that softer
edge illumination will reduce this ripple, but possibly at a
cost of efficiency loss. This might be alleviated by shaping
the reflectors to improve the main dish illumination while still
maintaining soft sub-reflector edge illumination. The effect of
changing the projected spacing between the reflectors can be
simply approximated as a rotation of the sub-reflector, and
will not cause significant improvement of the relative ripple
amplitude. Instead, the frequency response of the ripple will
shift, and this result can possibly be used to flatted the gain
response in certain very sensitive frequency bands. Also, in
order to accurately predict gain ripple, care must be taken
to accurately simulate or measure the sub-reflector and feed
combination response - including all the near field effects.
ACKNOWLEDGMENT
The author would like to thank the South African SKA
project and EMSS Antennas in Stellenbosch, South Africa for
financial support of this work.
REFERENCES
[1] P. E. Dewdney, P. J. Hall, R. T. Schilizzi, and T. J. L. W. Lazio, “TheSquare Kilometer Array,” Proceedings of the IEEE, vol. 97, no. 8, pp.1482 – 1496, Aug. 2009.
[2] J. L. Jonas, “MeerKAT - The South African array with composite dishesand wide-band single pixel feeds,” Proceedings of the IEEE, vol. 97,no. 8, pp. 1522–1530, Aug. 2009.
[3] D. Morris, “Chromatism in radio telescopes due to blocking and feedscattering,” Astron. Astrophysics, vol. 67, pp. 221–228, 1978.
[4] J. E. Hansen, Ed., Spherical Near-Field Measurements, ser. IEE Electro-magnetic Wave Series 26. London, UK: Peter Peregrinus Ltd., 1988.
[5] P.-S. Kildal, “Factorization of the feed efficiency of paraboloids andcassegrain antennas,” IEEE Trans. Antennas and Propag., vol. AP-33,no. 8, pp. 903–908, Aug. 1985.
[6] Y. Mizugutch, M. Akagawa, and H. Yokoi, “Offset dual reflector anten-nas,” in IEEE Antennas Propag. Soc. Symp. Dig., Amherst, MA, Oct.1976, pp. 2–5.
2175