ngVLA Antenna Memo # 4

Performance of Dual-Offset Gregorian Antennas with varying

Subreflector Angles
Sivasankaran Srikanth, Central Development Laboratory, Charlottesville, VA
June 18, 2019.
Abstract
The material presented in this memo is the result of analysis carried out during the optics design
phase of the Next Generation Very Large Array (ngVLA) project. The major science goals for the project
are described along with the chosen array configuration. Various options considered for the antenna
optics are shown and the current design that is used for costing of the antenna is illustrated. This design,
called the “Reference Design”, uses a dual-offset Gregorian geometry with shaped reflectors. The
Reference Design antenna uses a wide-angle subreflector with an opening angle (half angle) of 55º.
Performance comparison of conic section antennas with narrow-angle (≤ 20º) and wide-angle (≥ 25º)
subreflectors are shown. Efficiency and off-axis performance have been analyzed for conic section
antennas with subreflector opening angles of 41º, 46º and 55º.
1 Introduction
The initial set of science goals for the ngVLA include the study of formation of solar system analogs,
probing initial conditions for planetary systems, charting the assembly, structure and evolution of
galaxies, tests of gravity using pulsars in the galactic center, understanding the formation of black holes,
etc. [1], [2]. The science mission translates to the requirement of an array that will have 10 times the
sensitivity of the JVLA and ALMA, optimized for the 1.2 GHz to 116 GHz operating frequency range
and angular resolution of 50 mas at 1.2 GHz. The main array assumes 214 x 18 m antennas, spread over
baselines up to 1000 km located in greater New Mexico, eastern Arizona, west Texas and northern
Mexico. An additional 19 x 6 m antennas will form a short baseline array (SBA), sensitive to a portion of
the larger angular scales undetected by the main array. In mid-2018, a long baseline array (LBA)
consisting of additional 30 x 18 m antennas, was included to the full ngVLA definition [3]. The array
will be located at elevations over 2000m, which should allow for good observing conditions at the
specified frequencies.
The specification for the antenna, calls for a design goal of 75% aperture efficiency at 30 GHz. The
chosen optics for the antenna is a dual-offset Gregorian design with the subreflector/feed arm located on
the lower end of the primary reflector. The reflectors are shaped for high aperture efficiency and low
spillover temperature. The specification for the antenna surface error is < 160 μm RMS that enables
achieving 50% aperture efficiency at the highest receiver band [4]. The antenna design conceived
currently, known as the “Reference Design”, is a low technical risk costed-concept and is the technical
and cost basis of the ngVLA Astro2020 Decadal Survey proposal. The antennas will be outfitted with six
receivers covering the 1.2 to 50.5 GHz and 70 to 116 GHz frequency ranges in two cryostats [5].
2 Optics Design Options
Antenna diameters ranging from 16 to 25 m were considered initially. The final size of 18 m was chosen
for reasons mostly: (1) cost to meet point source sensitivity targets and (2) survey speed. Secondary
factors were manufacturability/transportability and viability for supporting novel concepts that may
improve cost/performance metrics. The chosen subreflector diameter is 3.2 m, nearly the same as that of
the VLBA. At 1.2 GHz, the diameter is 12.8λ resulting in diffraction loss of 7%. The subreflector
diameter is 0.18x that of the main reflector and blockage loss is 7.5% on a symmetric antenna. At the
least, six receivers are required in order to cover the specified frequency range (1.2 to 116 GHz). In
addition to the high blockage loss in case of a symmetric antenna, locating all the receivers in the shadow
of the subreflector is challenging and hence, it is necessary to gravitate towards dual-offset designs for the
ngVLA antenna. This geometry comes with additional advantages like blockage-free aperture, lower
sidelobes, minimized standing waves, and higher aperture efficiency. A design where the feed arm is
located at the bottom of the main reflector provides easy accessibility to the receivers for servicing
purposes.
Figures 1 and 2 show optics configuration for dual-offset Cassegrain and Gregorian antennas,
respectively, for various (15⁰, 18⁰ and 22⁰) subreflector opening (half) angles (θs). The main reflector
offset distance is 9.2 m for all cases. It is to be noted that the main reflector rim-to-rim distance in the
symmetric plane is larger than 18 m (19.6 m), a disadvantage of the offset geometry. Similarly, the
subreflector is also larger in the symmetric plane as shown in the figures. In case of the Cassegrain
antenna, the subreflector causes blockage even for the smallest subreflector angle. For the 15⁰ antenna, a
3.8 m section of the subreflector blocks the aperture of the main reflector. The blockage can be avoided
by increasing the main reflector offset distance, which will result in larger rim-to-rim distance of the main
reflector. Also notice, that the subreflector size, the focal length of the hyperboloid subreflector, hence
the distance of the feed from the subreflector vertex (FI) and the amount of blockage get larger with
increasing subreflector angles. The Gregorian design provides a completely blockage-free aperture. The
size and the focal length of the ellipsoid get smaller with increasing opening angles.
3 Reference Design for the ngVLA
The Reference Design of the ngVLA antenna uses a dual-offset reflector configuration with shaped
reflectors, where θs=55º. The angle of the subreflector on AUI/NRAO/GBO centimeter wavelength radio
telescopes varies between 7.2⁰ (140-foot in Green Bank) and 15º (GBT) with the VLA at 9.3º and VLBA
at 13.3º. The MeerKAT [6] antenna (13.5 m main reflector, 3.5 m subreflector) uses a much larger angle
of 48.9º and the DVA-1 [7] (15 m main reflector, 3.6 m subreflector) has θs=55º. Linear taper corrugated
feed horn designed to provide an illumination taper of approximately -12 dB at the edge of the
subreflector, has a diameter of 15λ at the lowest frequency of operation for θs=7º, 12λ for θs=9.3º (VLA),
and 8.5λ for θs=15º (GBT). The choice of θs=55º for the ngVLA, allows close packing of the receivers as
the feeds are smaller and the band selection mechanism viable. In order to integrate the feeds into a small
number of compact dewars, quad-ridged feed horns (QRFH) [8] and axially corrugated feed horns
(ACFH) [9] will be used. QRFH is smaller in size (aperture diameter of 2λ) for a given illumination taper
and will be used for the lowest two bands that require about 3:1 bandwidth. The ridge profile and the
horn profile are optimized for obtaining aperture efficiencies of about 70%. Wide flare angle ACFH [10]
will be used for frequencies above 12.3 GHz. It has a bandwidth ratio of 2:1 and has aperture diameter of
about 2λ.
Figure 3(a) shows shaped optics design currently under consideration for the ngVLA antenna. Figure
3(b) shows the conic section dual-offset Gregorian reflector design with θs=55º, from which the shaped
optics was arrived. The main reflector is 18 m and the focal length of the parabola is 7.2 m. The offset
distance is 9.1 m. The design in Figure 3(a) was optimized for high aperture efficiency and low spillover
temperature [11]. The shape of the main reflector deviates from the parabola by +0.25 m and -1.5 m.
The aperture efficiency calculated using simulated feed patterns of an ACFH (Figure 10) is about 84% at
1.8 GHz. The calculated beam pattern of the shaped system is shown in Figure 4(a). The first sidelobe is
-19.6 dB below the peak of the beam and crosspolarized sidelobe is -32 dB. A spherical wave expansion
(SWE) representation of the fields of the feed horn was used in GRASP to calculate the antenna beam
patterns. The beam patterns of the conic section antennas are shown in Figures 4(b) and (c) at 1.8 GHz
and 17.5 GHz, respectively. The first sidelobe is -27 dB and crosspolarization is -30 dB at 1.8 GHz.
Efficiency at 1.8 GHz is 65.2% and at 17.5 GHz is 67%.
4 Comparison of conic section antennas with θs =15º, 18º and 55º at L-band
A comparison of performance of offset Gregorian antennas with subreflector angles θs =15º, 18º (Figures
2(a), 2(b)) and 55º (Figure 3(b)) at L-band is shown in this section. For the first two cases, profile/
compact corrugated horn is used, as a linear taper horn is substantially large (aperture diameter 8.5l and
length 3000 mm for θs=15º) at L-band. The profile horn is about 40% smaller in aperture and about 30%
shorter compared to a linear taper horn for a given illumination taper. While a final design of the QRFH
is not available at this time, for comparison purposes, this memo uses an ACFH for the 1.2 to 2.4 GHz
band for the θs = 55º antenna. Figure 5 shows the dimensions of corrugated feed horns for the three cases.
The aperture diameters are 4.6λ, 3.9λ and 2.0λ at the low end of the band for the three cases of 15º, 18º
and 55º, respectively. In case of the profile horns, θmax is the maximum taper angle of the inside profile
and for ACFH, θf is the flare angle. Simulated feed patterns are shown in Figures 6, 8 and 10 for the three
cases. Copolarization patterns are shown in the H- and E-planes and crosspolarization in the diagonal
plane. For the profile horns, the field patterns are calculated using mode-matching technique and for the
ACFH, Method of Moments technique is used. For the profile horns, the illumination taper at the edge of
the subreflector does not change monotonically as a function of frequency; this is due to the fact that the
phase curvature of the aperture field is the sum of that of a linear taper horn and that of an open-ended
waveguide. In addition, the presence of HE12 mode further alters the shape of the far-field pattern. The
range of the illumination taper and the average value at the edge of the subreflector are indicated in the
figures. The variation of taper is substantial in case of the ACFH, particularly in the E-plane (-15.6 dB to
-28.3 dB). It is to be noted that the backlobe, in case of the profile horns, is about 10 dB lower than that
of the ACFH. The patterns in the principal planes are overlaid in Figures 7, 9 and 11. For the 55º horn,
the symmetry of the beam deteriorates above 2.0 GHz. Figures 12 and 13 show the computed far-field
patterns of the antennas at 1.8 GHz and 2.4 GHz, respectively. Beam patterns of the 15º antenna
computed for a linear taper horn are also included (Figures 12(a), 13(a)). Crosspolarization at 1.8 GHz is
dominated by the feed polarization for the 55º case. Computed aperture efficiency is shown in Figure 14.
While difference in efficiency between the 15º and 18º cases is minor, the linear taper horn clearly results
in much higher efficiency, about 10% at the center of the band. The ACFH displays about 5% lower
efficiency compared to the profile horns from 1.5 to 2 GHz, deteriorates monotonically at higher
frequencies and is lower by 14% at 2.4 GHz.
5 More conic section antenna designs
Analysis of antennas with subreflector angles of 55⁰, 46⁰ and 41⁰ shown in Figure 15 is covered in this
section. A preliminary design of a QRFH for the θs=55º optics has been completed at Caltech [12]. An
ACFH for the same optics has been prototyped in the 0.75-1.50 GHz band [10] and at Q-band and
characterized [13]. A scaled version of the 0.75-1.50 GHz ACFH at C-band was machined at the GBO
machine shop and characterized. QRFHs can be optimized to have a nominal -10 dB beamwidth in the
range of 50º to 140º. A QRFH to operate at the prime focus of the GBT having an illumination taper of
about -12 dB at 39º is being developed for the Ultra-wideband receiver [14]. In addition, efforts are
underway in the development of QRFHs for a gamut of opening angles at several institutions. ACFHs
have been developed recently at the CDL to illuminate subreflectors with θs=46º and 41º. These L-band
designs are shown in Figures 16(b) and 16(c). The aperture diameters are 2.2l and 2.5l and θf is 45⁰ and
40⁰, respectively. Figures 17 and 19 show simulated copolarization patterns in the H- and E-planes and
crosspolarization in the diagonal plane for the two horns. The change in illumination taper in the E-plane
is large compared to that in the H-plane. The change gets smaller with smaller flare angles and for the
θf=40º horn, it is 2.4 dB. Figures 18 and 20 show the circular symmetry of the beam. There is an
excellent match of the patterns in the two planes for the 41º horn all the way to 2.4 GHz (Figure 20(e)).
The antenna beams at 1.8 GHz and 2.4 GHz, are shown in Figures 21 and 22, respectively. The first side
lobe for the 55º antenna is marginally lower compared to the other cases at 1.8 GHz. Crosspolarization is
about the same for the 55º and 41º cases and 5 dB higher compared to the 46º case in the 45º-plane. This
results from the low crosspolarization of the feed horn (-35 dB) shown in Figure 18(a). At 2.4 GHz for
the 55º antenna, crosspolarization is worse by 6 dB compared to the 41⁰ antenna in the 45º-plane, caused
by the feed horn polarization. Computed aperture efficiency is shown in Figure 23. Efficiency is nearly
the same for the 41º and 46º antennas up to 1.7 GHz. At higher frequencies, 41º antenna has better
performance, primarily because of the preservation of the circular symmetry of the feed horn pattern. The
average efficiency of the 55º antenna is about 8% lower compared to the 41º antenna. This translates to a
loss of 18 antennas in collecting area out of the 244 18 m antennas.
Feed dimensions and efficiencies were also calculated for the 12.5 to 21.5 GHz frequency range. As seen
in Figure 24, the 41º horn is larger by 15 mm at the aperture compared to the 55º horn. Antenna beams
are shown at 17.5 GHz and 21.5 GHz, in Figures 25 and 26, respectively. Crosspolarization, which is
maximum in the asymmetric plane of a dual-offset antenna, is very low (<-50 dB) at these frequencies.
The level of crosspolarization in the 45º-plane is about -30 dB and is caused by the feed polarization
characteristics. Aperture efficiency is graphed in Figure 27, where the efficiency for a 15º antenna fed by
a linear taper horn (200 mm diameter and 290 mm long) is also included. The efficiency variation for the
three cases is very similar to that L-band and the loss in collecting area is equivalent to 17 antennas going
from 55º to 41º optics.
A brief analysis of the scanning properties of the three antennas was carried out. The feed was displaced
by 2.5 cms, 5 cms and 7.5 cms from the focus in the asymmetric plane of the antenna for the three cases
and aperture efficiency was calculated for Band 3. The displacement of 2.5 cms is equivalent to 1λ at
12.5 GHz. Figure 28 shows the aperture efficiency for the three displacement cases. Here the efficiency
is normalized with respect to the on-axis efficiency. For a displacement of 5 cms, efficiency decreases by
3%, 7% and 15% with respect to on-axis efficiency for the antennas with θs=41º, 46º and 55º respectively,
at 17.5 GHz.
6 Conclusions
All calculations of the antenna beam patterns and efficiencies shown in the memo were carried out using
TICRA-GRASP software, which is industry standard for reflector antenna analysis. The accuracy of the
simulations is further enhanced, resulting from the use of SWE coefficients for representing the fields at
the subreflector. Small angle subreflector (θs≤20º) antennas have at least 5% higher aperture efficiency at
L-band compared to the 55º subreflector antenna. This is purely academic as the feed at L-band is very
large for the small angle design on an 18m antenna. However, if the low end of the ngVLA frequency
band is limited to 8 GHz, small angle optics using narrow flare angle linear taper feed horn becomes an
option. This can increase the effective collecting area by about 15%.
Performance of antennas with subreflector angles of 41º, 45º, and 55º using ACFHs are compared at L-
and Ku-bands. The 41º antenna with an ACFH of 2.5λ diameter has about 8% higher aperture efficiency
compared to the 55º antenna using a 2λ diameter feed horn. The electromagnetic consulting contractors
on the project are of the opinion that the shaping of the reflectors can be optimized to yield aperture
efficiency of over 80%, irrespective of the subreflector angle. It is shown that scan loss is only 3% for the
41º antenna, compared to 15% for the 55º antenna for a feed offset of 2λ. This necessitates very accurate
deployment of the high frequency receivers with respect to the secondary focus in case of the 55º antenna.
Shaped reflector systems display poorer scanning performance in general [15]. The dependency of scan
loss on subreflector angle on shaped reflector systems has not been analyzed at present. The author
suggests that scan loss for a gamut of subreflector angles on shaped reflectors be studied before the final
design for the ngVLA antenna is chosen. It is very likely that spillover temperature for the three wide-
angle optics covered in Section 5 is very similar, a topic beyond the scope of this memo.
Beamwidth of the QRFH varies by a larger amount compared to that of the ACFH, even over 2:1
bandwidth. Giving up continuous coverage of the 1.2 to 12.3 GHz range and using ACFH in place of the
QRFH, to cover the range in two narrower bands (2:1) will result in more sensitive receivers.
The author acknowledges the comments and edits suggested by E. Murphy, M. Pospieszalski, R. Salina
and P. Ward.
References
[1] C. L. Carilli et.al. “Science Working Groups Project Overview,” ngVLA Memo # 5.
[2] ngvla.nrao.edu/page/science
[3] ngvla.nrao.edu/page/array-config
[4] R. Selina and E. Murphy, “ngVLA Reference Design Development & Performance Estimates,”
ngVLA Memo # 17.
[5] W. Grammer et al., “ngVLA Front End Reference Design Description,” Document #
020.30.03.01.00-0003-DSN.
[6] I. P. Theron, R. Lehmensiek and D. I. L. de Villiers, “The design of the MeerKAT dish optics,”
European Microwave Conference.
[7] G. Cortes, W. Imbriale and L. Baker, “DVA-1 Optics Design and Analysis,”
DVA1_CDR_Optics_V3_2012_06_17.
[8] Weinreb, S. and Mani, H., “Low Cost 1.2 to 116 GHz Receiver System – a Benchmark for ngVLA,”
ngVLA Science Workshop presentation, June 2017.
[9] R. Lehmensiek and D. I. L. de Villiers, “Wide Flare Angle Axially Corrugated Conical Horn Design
for a Classical Offset Dual-Reflector Antenna,” 6th European Conference on Antennas and
Propagation.
[10] L. Baker and B. Veidt, “DVA-1 Performance with an Octave Horn; From CST & GRASP
Simulations,” 2014.
[11] L. Baker, “Analysis of ngVLA Design #6 With Ideal and Actual Feed,” Document #:
020.25.01.00.00-0001-REP ngVLA Optical Reference Design, January 2017.
[12] J. Shi, S. Weinreb, W. Zhong and X. Yin, “Quadruple-Ridged Flared Horn Operating from 8 to 50
GHz,” Internal Memo, Dept. of Electrical Engineering, California Institute of Technology,
Pasadena, CA, December 1, 2016.
[13] L. Locke, L. B. G. Knee, F. Jiang, V. Reshetov and L. Baker, “ngVLA Communities Studies
Report: Feed and Receiver Development at NRC, Herzberg,” ngVLA Memo #32.
[14] S. White, GBO, Private communication.
[15] P. J. Napier, “Beam Scan Properties of Non-parabolic Reflectors,” NLSRT Memo No. 46, Jan 14,
1989, NRAO, Socorro, NM.
Figure 1. Dual-Offset Cassegrain antenna with θs = 15⁰, 18⁰, 22⁰.

Figure 2. Dual-Offset Gregorian antenna with θs = 15⁰, 18⁰, 22⁰.

 Figure 3. (a) Reference Design shaped antenna, (b) Conic section antenna.

Figure 4. (a) Shaped antenna beam at 1.8 GHz, (b) Conic section antenna beam at 1.8 GHz,

(c) Conic section antenna beam at 17.5 GHz.

 Figure 5. (a) Profile Feed Horn θs=15⁰, (b) Profile Feed Horn θs=18⁰,

(c) Axially Corrugated Feed Horn θs=55⁰.

Figure 6. Simulated Copolar, Crosspolar Patterns of Profile Horn for θs=15º.

Figure 7. Simulated Copolar (H, E-planes), Crosspolar (D-plane) Patterns of Profile Horn for θs=15º.

Figure 8. Simulated Copolar, Crosspolar Patterns of Profile Horn for θs=18º.

Figure 9. Simulated Copolar (H, E-planes), Crosspolar (D-plane) Patterns of Profile Horn for θs=18º.

Figure 10. Simulated Copolar, Crosspolar Patterns of ACFH for θs=55º.

Figure 11. Simulated Copolar (H, E-planes), Crosspolar (D-plane) Patterns of ACFH for θs=55º.

Figure 12. Antenna Beam at 1.8 GHz, (a) Linear θs=15º, (b) Profile θs=15º, (c) Profile θs=18º,

(d) ACFH θs=55º.

Figure 13. Antenna Beam at 2.4 GHz, (a) Linear θs=15º, (b) Profile θs=15º, (c) Profile θs=18º,

(d) ACFH θs=55º.

Figure 14. Aperture Efficiency (1.2-2.4 GHz) for θs=15º, θs=18º, θs=55º.
 Figure 15. Dual-Offset Gregorian antenna with θs = 55⁰, 46⁰, 41⁰.

Figure 16. Axially Corrugated Feed Horn (1.2 – 2.4 GHz) for (a) θs=55º, (b) θs=46º, (c) θs=41º.
 Figure 17. Simulated Copolar, Crosspolar Patterns of ACFH for θs=46º.

Figure 18. Simulated Copolar (H, E-planes), Crosspolar (D-plane) Patterns of ACFH for θs=46º.
 Figure 19. Simulated Copolar, Crosspolar Patterns of ACFH for θs=41º.

Figure 20. Simulated Copolar (H, E-planes), Crosspolar (D-plane) Patterns of ACFH for θs=41º.
Figure 21. Antenna Beam at 1.8 GHz with ACFH (a) θs=55º, (b) θs=46º, (c) θs=41º.

Figure 22. Antenna Beam at 2.4 GHz with ACFH (a) θs=55º, (b) θs=46º, (c) θs=41º.
 Figure 23. Aperture Efficiency (1.2-2.4 GHz) for θs=55º, θs=46º, θs=41º.

Figure 24. Axially Corrugated Feed Horn (12.5 – 21.5 GHz) for (a) θs=55º, (b) θs=46º, (c) θs=41º.
Figure 25. Antenna Beam at 17.5 GHz with ACFH (a) θs=55º, (b) θs=46º, (c) θs=41º.

Figure 26. Antenna Beam at 21.5 GHz with ACFH (a) θs=55º, (b) θs=46º, (c) θs=41º.
 Figure 27. Aperture Efficiency (12.5-21.5 GHz) for θs=55º, θs=46º, θs=41º.

Figure 28. Scanning Performance (12.5-21.5 GHz) for θs=55º, θs=46º, θs=41º.