Design of the tertiary optical system for the LLAMA radio telescope

Emiliano Rasztockya,c,* , Matı́as Rolf Hampelb,d , Rodrigo Reevese , Jacques R.D. Lepinef ,
Gustavo Esteban Romeroa,g
a
Instituto Argentino de Radioastronomı́a (CONICET - UNLP - CIC), Berazategui, Argentina
b
Instituto de Tecnologı́as en Detección y Astropartı́culas (CNEA - CONICET - UNSAM),Buenos Aires, Argentina
c
Universidad Nacional de San Martı́n, San Martı́n Argentina
d
Universidad Tecnológica Nacional - Facultad Regional Buenos Aires, Argentina
e
Universidad de Concepción, CePIA, Departamento de Astronomı́a, Concepción, Chile
f
Universidade de São Paulo, Instituto de Astronomia, Geofı́sica e Ciências Atmosféricas, São Paulo, Brazil
arXiv:2410.17419v1 [astro-ph.IM] 22 Oct 2024

g
Universidad Nacional de La Plata, Facultad de Ciencias Astronómicas y Geofı́sicas, La Plata, Argentina

Abstract. Many modern radio telescopes employ an observational strategy that involves maximizing the use of their
available spaces (cabins), outfitting them with various receivers at different frequencies to detect incoming signals
from the sky simultaneously or individually. The Large Latin American Millimeter Array (LLAMA), is a joint venture
between Argentina and Brazil consisting of the installation and operation of a 12-meter aperture Cassegrain telescope.
It features three available cabins for instrumentation and plans to install six single-pixel heterodyne receivers, covering
different bandwidths in the 30 to 950 GHz window of the electromagnetic spectrum, in its two lateral Nasmyth cabins
at different phases of the project. Therefore, it is crucial not only to design a tertiary optical system that couples the
antenna beam to those receivers, but also to do it in a scalable way. The primary goal for the design is to simultaneously
maximize the antenna efficiency while minimizing optical aberrations for all receivers, both fundamental aspects for
the optimal functioning of cutting-edge astronomical instruments. In this paper, we present the entire design process,
starting from the quasi-optical approach based on the propagation of a fundamental Gaussian beam mode, continuing
with the validation of the design based on physical optics simulations, and ending with a tolerance analysis of the
system. As a result of this process, a frequency independent tertiary optical system has been achieved for almost all
the receivers, which is expected to provide high optical performance for the radio telescope.

Keywords: LLAMA, Radio telescope, Quasi-optical systems, Physical optics simulations, Millimeter/sub-millimeter
wavelengths.

*Emiliano Rasztocky, [email protected]

1 Introduction
1.1 Large Latin American Millimeter Array (LLAMA) Radio Telescope
LLAMA,1–4 is a joint scientific and technological venture between Argentina and Brazil. Its aim is
to establish and operate a radio observatory at Alto Chorrillos, Argentina, at an altitude of 4800 m
above sea level, capable of making astronomical observations in the millimetre and sub-millimeter
wavelength ranges. The main scientific objectives of LLAMA are the study of spectral molecular
lines of forming stars, young stellar objects, black holes, the Sun’s chromosphere, outflows in star-
bursts, the chemistry of the intergalactic medium around galaxies and dusty Active Galactic Nuclei
(AGN), galaxy formation at high redshifts and even Cosmic Microwave Background (CMB) fluc-
tuations at small angular scales.3–5 The radio telescope is a 12-meter aperture Cassegrain-type
dual reflector antenna manufactured by Vertex Antennentechnik and shares the same architecture
and optical characteristics (Table 1) as that of the Atacama Pathfinder Experiment (APEX) radio
telescope6 (Fig. 1). The antenna has been designed to provide an absolute pointing accuracy of
2 arcsec with an offset pointing of 0.6 arcsec and a reflector surface accuracy of 25 µm. These

1
features will allow the observation of the universe in the 30 to 950 GHz window of the electro-
magnetic spectrum. The radio telescope will be able to operate as a single dish or as part of an
interferometer array such as ngEHT.7
Many modern radio telescopes maximize the use of their available cabins by outfitting them
with various receivers operating at different frequencies, allowing both rapid switching between
frequencies or even multi-frequency observations. It was decided early in the project that LLAMA,
with three cabins - the central Cassegrain cabin (Cab-Cass) and two Nasmyth cabins (Cab-A and
Cab-B) - would host six ALMA8 -like single-pixel heterodyne receivers in the Nasmyth cabins.
For the First Light (FL) phase, receivers for bands 5 (163-211 GHz), 6 (211-275 GHz), and 9
(602-720 GHz) will be installed in Cab-B to validate the technical and scientific capabilities of the
telescope. Subsequently, a significant upgrade will configure the cabins for a longer-term phase
(LT). In this phase, bands 6, 7 (275-373 GHz), and 9 will be installed in Cab-A, and bands 1
(35-52 GHz), 2+3 (67-116 GHz), and 5 in Cab-B. This configuration leaves Cab-Cass available
for future upgrades, which might include not only single-pixel detectors but also multi-detector
cameras to take advantage of the larger field of view (∼10 arcmin). Implementing this strategy
requires using tertiary optical systems to couple the detectors (receivers) with the antenna.

(a) (b)

Fig 1 (a) APEX antenna at Atacama in Chile. Credit: ESO. (b) Optical configuration of the LLAMA Cassegrain
system.

2
 Table 1 Optical parameters of the LLAMA radio telescope9 (Fig. 1).
Parabolic primary reflector:
Diameter Dp = 12 m
Focal length fp = 4.8 m
Focal ratio Fp = 0.4
Eccentricity ep = 1
Aperture illumination ψp = 64.01 °
Depth of primary reflector H= 1.875 m
Primary vertex hole diam. vh= 0.75 m

Hyperbolic secondary reflector (sub-reflector):

Diameter Ds = 0.75 m
Focal length fs = 0.30962 m
Eccentricity es = 1.10526
Secondary illumination αs = 3.58 °

Cassegrain system:
Equivalent focal length feq = 96 m
Focal aperture F#= 8
Magnification factor M= 20
Distance primary focus to sub-ref vertex ∆= 0.29414 m
Distance primary to sub-ref vertex dvv= 4.50586 m
Distance primary vertex to Cassegrain focus dvf= 1.377 m

1.2 Receivers
The receivers to be installed in the Nasmyth cabins of LLAMA are identical to those used in the
ALMA radio telescopes. These receivers are planned to be used without any modification, since
they have been validated with years of observation, avoiding in this way unnecessary delays during
the initial stage of the project, due to the development of new receivers, allowing resources to
focus on other subsystems (e.g., the tertiary optical system). In the case of ALMA, the observation
frequency band of 30-950 GHz was subdivided into 10 sub-bands, from band 1 to 10, guided
by considerations such as atmospheric conditions, astronomical requirements, available detection
technology at the time, and polarization considerations.10, 11 For each sub-band, a dedicated single-
pixel heterodyne receiver was integrated into modular cartridges, which are subsequently installed
in a cryostat (cryo) at the focal plane of the Cassegrain antenna system.12
The modular cartridges are divided into two parts: 1) the cold cartridge assembly (CCA) with
three cooled stages at 4, 15, and 90 K installed inside the cryostat, which contains the feed horn,
cold optics, and part of the front-end electronics (radio frequency amplifiers, local oscillator, and
mixer for the intermediate frequency with its cryogenic amplifiers); 2) the warm cartridge assem-
bly (WCA) at room temperature installed beneath the cryostat, which also contains part of the
front-end electronics. Each receiver detects both polarizations, with its feed horn illuminating the
sub-reflector through windows on the top of the cryostat. The receivers, equipped with reflective
(mirrors) or refractive (dielectric lenses) optics,13 are designed to achieve a 12 dB edge taper on
the sub-reflector to maximize antenna efficiency.14, 15 The ten bands observe different parts of the

3
sky, and the observing band in ALMA is selected by tilting the sub-reflector.
For LLAMA, the cartridges will be installed in two cryostats, each with a diameter of 538 mm
and a height of 550 mm. These cryostats will be placed in each Nasmyth cabin, accommodating
three receivers each (Fig. 2). The focal position (ωN ASS ) with respect to (wrt) the cryostat top
plate for each receiver is shown in Table 2.

Fig 2 CAD model of the LLAMA cryostat with cartridges installed.

1.3 LLAMA Cabins

The LLAMA radio telescope has three cabins for housing signal detection instrumentation: the
Cab-Cass, located just below the primary reflector, and the two Nasmyth cabins (Cab-A and Cab-
B) on either side. The beam can pass from Cab-Cass to both Nasmyth cabins through the Nasmyth
tubes, which have a clearance diameter of 340 mm for Cab-B and 80 mm for Cab-A due to the
presence of the antenna’s elevation encoder. This constraint determined the placement of low-
frequency receivers in Cab-B (wider beams) and high-frequency receivers in Cab-A (narrower
beams). Each cryostat is located within its respective Nasmyth cabin, with its center positioned
3800 mm from the Az axis and its reference plane (top cover) 1050 mm below the El axis of
the antenna (Fig. 3). For positioning components inside the antenna, a right-handed rectangular
coordinate system (i.e., X, Y, Z) is defined with the origin at the intersection of the azimuthal (Az)
and elevation (El) axes.
The position and orientation of each receiver with respect to the LLAMA Coordinate system is
presented in Fig. 4 and Table 2.

4
 Fig 3 LLAMA cabins configuration (dimensions in mm).

Fig 4 LLAMA Receivers position scheme. The receiver corresponding to FL phase is indicated in parentheses.

Table 2 LLAMA cartridge position and characteristics for FL and LT configurations.

Focus
Cartridge Beam Angle
Freq. Cartridge position
X;Y X;Y to
Band band diameter wrt cryo
position position cryo center
(GHz) (mm) top plate
(mm) (mm) (deg)
(mm)
B1 (LT) 35-52 170 3800 ; 150 3800 ; 190 2.48 553
B2+3 (LT) 67-116 170 3950 ; 0 3990 ; 0 2.48 131.8
B5 (FL/LT) 163-211 170 3800 ; -150 3800 ; -200 2.38 -15
B6 (FL) 211-275 170 3950 ; 0 4000 ; 0 2.38 -15
B6 (LT) 211-275 170 -3800 ; 150 3800 ; 200 2.38 -15
B7 (LT) 275-373 170 -3950 ; 0 -4000 ; 0 0.97 -50
B9 (FL) 602-720 170 3800 ; 150 3800 ; 200 0.97 -27
B9 (LT) 602-720 170 -3800 ; -150 -3800 ; -200 0.97 -27

5
2 Tertiary Optical System Design
The tertiary optical system is responsible for coupling the main beam of the telescope, focused
onto the antenna focal plane (ωCASS ), to the beam of each receiver. For LLAMA, this tertiary
optical system is referred to as the NAsmyth Cabins Optical System (NACOS).
NACOS is expected to maximize the antenna aperture efficiency for all detectors while mini-
mizing optical aberrations. It must be desigend as a flexible system, allowing the antenna user to
select between each receiver for single-band observation mode or to enable multi-band simultane-
ous observations by splitting the sky signal through the implementation of beam splitters (dichroic
filters and/or polarizing grids). These upgrades are expected to be part of a future enhancement of
the radio telescope and are not analyzed in the current work.
NACOS was designed using quasi-optics (QO),16 a precise approach for millimeter and submil-
limeter wavelength systems. At these scales, component sizes, such as mirrors, are comparable to
the wavelength of the radio waves, leading to significant diffraction effects that simpler approaches,
like geometrical optics (which treats light as rays), cannot accurately model.
In QO, beam propagation is characterized by a Gaussian electric field distribution, known as
the fundamental mode. Higher-order Gaussian modes can be superimposed on this fundamental
mode to describe more complex electric field distributions.
The propagation of the Gaussian fundamental mode approach has been extensively used in
the design of millimeter/sub-millimeter optical systems for astronomical instrumentation17–20 and
demonstrated21 to be a very accurate tool for the design of NACOS for LLAMA.
The propagation of a Gaussian beam in free space is characterized by the following parameters:
2
1 πω02

R (z) = z + , (1)
z λ
"  2 #0.5
λz
ω (z) = ω0 1 + , (2)
πω02
λz
tan (ϕ0 ) = , (3)
πω02
where R is the beam radius of curvature, ω is the beam transverse radius, ϕ0 is the phase shift and
z is the distance of propagation of the beam from the beam waist radius ω0 .
The transformation that the Gaussian beam undergoes as it propagates through an optical com-
ponent or system can be obtained from the ABCD matrix method, where each optical component
is described by a particular matrix. The complete system matrix is obtained by multiplying the
individual matrices from right to left in the order that the beam encounters them as it propagates.16
The distance dout of the output beam waist ω0out to the optical system/component can be found
with

(Adin + B)(Cdin + D) + ACzc 2

dout = − (4)
(Cdin + D)2 + C 2 zc 2
and
ω0in
ω0out = , (5)
[(Cdin + D)2 + C 2 zc 2 ]0.5

6
where din is the distance of the input beam waist ω0in to the optical system/component. zc =πω02 /λ
is the confocal distance and A, B, C and D are elements of the matrix describing either a mirror
with focal length f or a free space distance d for NACOS, and are given by
   
1 0 1 d
Mmirror = ; Mdistance = . (6)
−1/f 1 0 1
Having obtained the position and beam waist of the output beam (i.e., dout and ω0out ), its
subsequent evolution can be determined with Eqs. (1) to (3). Thus, the Gaussian beam launched
from the Nasmyth focus (or from the feed including the cartridge optics, as shown in Fig. 5 and
detailed in Table 3), can be propagated through the system. From the size of the beam (ω), it
is possible to estimate the losses22 due to beam distortion (KF i ) [Eq. (7)] and cross polarization
(KCOi ) [Eq. (8)] of each mirror, as well as the defocus loss16 (KAxiall ) [Eq. (9)]. Consequently,
the total system loss (KT otal ) [Eq. (10)] can be determined. Losses due to spillover on the optical
components are not considered at this stage of the analysis.

ωmi 2 tan2 θi
KF i = 1 − . (7)
8fi 2
ωmi 2 tan2 θi
KCOi = 1 − . (8)
4fi 2
4
KAxiall =  2  2 . (9)
ω0b ω0a λ∆z
ω0a
+ ω0b + πω0a ω0b
! !
Y Y
KT otal = KF i KCOi Kaxiall . (10)
i i

ωmi , θi and fi are the beam radius at the mirror i, the mirror inclination with respect to the
optical axis and the mirror focal length respectively. ∆z is the distance between the beam waists
of beams a and b.
Then, from the beam size at the sub-reflector (ωsubref ), the edge taper on it can be calculated,
 2
rsubref
T e = 8.696 α [dB] ; α = (11)
ωsubref
and, following the equivalent paraboloid representation, the antenna aperture efficiency16 can be
evaluated at the rim of the sub-reflector23
2  −f b2 α 2
Apef f = e − e−α . (12)
α
The antenna aperture efficiency considers spillover and illumination over the sub-reflector (α)
as well as the fractional blockage of the sub-reflector on the primary reflector, given by f b =
rsubref /rp = 0.0625. The maximum achievable aperture efficiency for LLAMA, with fundamental
Gaussian beam illumination, is 80.35%.

7
 Fig 5 Conceptual scheme of the quasi-optic system formed by the horn and the optics of the receivers.

Table 3 Quasi-optical parameters for the central frequency of the optics for each receiver.
Band B124 B2+325 B512 B612 B712 B912
Freq. (GHz) 42.50 91 187 243 324 661
a (mm) 15.63 10.69 4.50 3.54 3.00 2.54
Lh (mm) 70 43.43 60 46.54 45.68 15.52
d1 (mm) 195.80 91.69 60.05 46 38 44.48
f 1 (mm) 181 91 32.76 29.37 25.76 24.86
d2 (mm) - - 140 158 157 95.91
f 2 (mm) - - 67.19 76.61 77.07 39.41
pc (mm) 5.21 5.50 4.18 3.47 3.25 9.11
ω0 (mm) 9.38 4.80 2.80 2.19 1.86 1.06
ωa (mm) 9.38 4.80 2.90 2.28 1.93 1.63
ωout (mm) 42.54 20.70 9.60 7.45 5.59 2.73
dout (mm) 450.4 121.96 235.08 229.99 215.99 149.38

Focal plane wrt

553 131.8 -15 -15 -50 -27
cryo top plate (mm)

In Table 3, both the position of the focus relative to the top plate of the cryostat and its size
(ωout ≡ ωN ASS ) were obtained from the physical optics simulation of the horn-optics system. In
Appendix A the results for B1 and B2+3 are presented.

2.1 NACOS-FL
A reduced version of NACOS, called NACOS-FL, is planned for implementation during the FL
phase of the project. In NACOS-FL, receivers for bands 5, 6, and 9 will be installed in Cab-B. The
optical and mechanical designs were completed from 2016 to 2017 by Jacob Kooi from NASA’s Jet
Propulsion Laboratory and Astro Electro-Mechanical Engineering, LLC, respectively. NACOS-FL
was manufactured in Araraquara, Brazil, from 2018 to 2019 by the company ALFA Ferramentaria.
It is currently in the Assembly, Integration and Verification (AIV) stage at the same company.
A frequency independent optical system was design based on a Gaussian Beam Telescope
(GBT), which consists of a pair of focusing components separated by the sum of their focal
lengths.16 The key feature of the GBT is that both magnification [Eq. (13)] and output beam

8
waist location [Eq. (14)] are independent of the signal frequency, a useful and desirable feature in
systems that operates over broad bandwidths.
ω0out f2
M= = (13)
ω0in f1
 
f2 f2
dout = f1 + f2 − din . (14)
f1 f1
The GBT is formed by mirrors M3B and M4B placed in Cab-CASS and Cab-B respectively,
with M=1 as required by the LLAMA receivers (Fig. 6).

Fig 6 GBT schematics for NACOS-FL in Cab-B. Units in mm.

The optical system includes a series of flat mirrors. Some mirrors can be removed from the
optical path (OP), while others can be reoriented internally using rotary actuators. This flexibility
allows for modifications and convenient folding of the beam path. The NACOS-FL 3D model
(Fig. 7) consists of two optomechanical structures: the CASS (in Cab-Cass) and the NASS-B (in
Cab-B).

9
Fig 7 Configuration of the optical system of NACOS-FL for Cab-B. The Gaussian beam propagation at 30 dB (99.9%)
contour power is also shown for B5 (purple), B6 (blue) and B9 (read).

The beam focused by the Cassegrain system in the antenna’s focal plane (ωCASS ) is progres-
sively reflected, first, in the common optical components section of the system; the Flat D mirror,
the Flat Cass, the Flat 1, the M3B off-axis elliptical mirror and the M4B off-axis hyperbolic mir-
ror. After reflection at M4B, the system’s second section allows for the selection of the following
optical channels:

1. Band 5: By placing the Flat Grid mirror in the OP, the beam is reflected in the plane Y-
Z towards Flat 2 mirror, placed in -Y direction, and finally onto the Nasmyth focus of B5
(ωN ASS−B5 ).

10
 2. Bands 6 and 9: By removing the Flat Grid from the OP, the beam is allowed to propagate
further down until being reflected by R Flat, a rotating mirror that has two set positions in
planes X-Z and Y-Z. With R Flat oriented in plane X-Z (+X direction), the beam is reflected
towards M5-B6 mirror, and finally onto the Nasmyth focus of B6 (ωN ASS−B6 ). With R Flat
oriented in plane Y-Z (+Y direction), the beam is reflected towards M5-B9 mirror, and finally
onto the Nasmyth focus of B9 (ωN ASS−B9 ).

It is worth noting that both the Flat D and Flat Cass mirrors can be removed from the optical
path (OP), enabling the signal from the sub-reflector to propagate downstream through the CASS
structure.
The main parameters of the mirrors for NACOS-FL configuration are presented in Table 4.

Table 4 Main mirror parameters for FL configuration. Position (X; Y; Z), Surface normal (nx; ny; nz), Diameter (∅)
and Focal length (f ).
Mirror Band X(mm) Y(mm) Z(mm) nx ny nz ∅(mm) f (mm)
Flat D 5,6,9 0 0 -423 -0.3090 0 0.9510 155 ∞
Flat Cass 5,6,9 -87.22 0 -303 0.4540 0 -0.8910 176.4 ∞
Flat 1 5,6,9 670.50 0 -303 -0.7071 0 -0.7071 387.6x275 ∞
M3B 5,6,9 670.50 0 -803 0.7071 0 0.7071 519.7x369.7 1564.75
M4B 5,6,9 3800 0 -803 -0.7071 0 -0.7071 425x300 1564.75
Flat Grid 5 3800 0 -1153 0 -0.7071 0.7071 345 ∞
Flat 2 5 3800 -229.72 -1153 0 0.6923 -0.7216 222x158 ∞
R Flat 6 3800 0 -1403 0.7071 0 0.7071 295x209 ∞
M5-B6 6 4019.30 0 -1403 -0.6923 0 -0.7216 155x110 ∞
R Flat 9 3800 0 -1403 0 0.7071 0.7071 295x209 ∞
M5-B9 9 3800 208.08 -1403 0 -0.7011 -0.7131 146x103 ∞

2.2 NACOS-LT
The Long Term phase of the project involves the installation of receivers in both Nasmyth cabins
as detailed in Table 2. In Cab-A, receivers for bands 6 and 9 will be relocated from Cab-B and
installed alongside band 7. In Cab-B, receivers for bands 1 and 2+3 will be installed with band 5.
Therefore, an upgrade from NACOS-FL to NACOS-LT is necessary.

2.2.1 Cab-A
A frequency independent optical system was designed for Cab-A. Various alternatives were ex-
plored, focusing on reducing system complexity and the number of mirrors to minimize reflection
losses and optical loading. This resulted in the development of a three-focusing mirror system,
described below.
The design process began with the determination of the focal length (f ) of the sole mirror
(M1A) to be installed in Cab-Cass. This mirror was positioned with its center coinciding with the
intersection of the elevation and azimuth axes, inclined at 45º. The focal length of M1A (fM 1A ) was
optimized to minimize the loss from encoder truncation on the beam. As a design rule, truncation
in quasi-optical systems should not occur below 4ω, where ω is the beam radius in the truncation
plane. This ensures a power transmission of ∼ 35 dB ≡ 99.97%. This rule was applied to each
mirror in the system. Therefore, the output beam waist (ωout1 ) from M1A must be sized and

11
positioned to ensure that encoder truncation on the beam does not occur below 4ω on both its front
face (ff) and its back face (bf). This is illustrated in Fig. 8.

Fig 8 Gaussian beam configuration to avoid elevation encoder truncation below 4ω. Units in mm.

To ensure both ωf f and ωbf are simultaneously < 4∅enc in Eq. (2), and using Eqs. (4) and
(5), along with the ABCD elements of a single mirror [Eq. (6)], two equations can be derived
to determine fM 1A . Equation (15) addresses truncation at the front face and Eq. (16) addresses
truncation at the back face.

  din1−ff f 2 
ff f +  2  2 − 1733 
2 d πω0in1 2
2 1− fin1
  +
∅enc ω0in1  
ff λf ff

≥ 1+   (15)
 
2  2  π ω 2
4 2
 20in1
1 − dfin1 + πωλf0in1 2
 λ
  
 din1 πω0in1 2
ff ff 1− f + λf
ff ff

  din1−fbf 2 
fbf +  2  2 − 1998 
2 din1 πω0in1 2
ω0in1 2
   1− f + 
∅enc  bf λf bf
≥ 1 +  (16)

2  2 π ω 2
4
 
πω0in1 2 20in1

1 − dfin1 + 2 2
  λ
   
λfbf din1 πω0in1
bf 1− f + λf
bf bf

Each of Eqs. (15) and (16) yields two values for the focal length at any given frequency. If the
result is a complex number, the solution lacks physical significance.
Evaluating Eqs. (15) and (16) with din =803 mm and ω0in1 ≡ ωCASS = 0.216F #λT e0.5 ∼ 6λ,
a set of results, shown in Fig. 9 (a) is obtained. The yellow shaded area indicates solutions
where truncation below 4ω does not occur on either the front or back face of the encoder. At
220 GHz, M1A can refocus the beam without truncation below 4ω, with a focal length ranging
from 582 to 593 mm. A focal length of 590 mm was chosen for M1A, which is expected to produce
approximately 0.06% loss due to encoder truncation at the lowest frequency of B6 (211 GHz), a
value considered negligible. Fig. 9 (b) shows the propagation of the Gaussian beams through the
encoder.

12
 (a) (b)

Fig 9 (a) Admissible frequencies through encoder with respect to the focal length of M1A mirror. (b) Gaussian beam
propagation through encoder of central frequencies of receivers in Cab-A.

Since M = 1 for NACOS, Eqs. (13) and (14) show that the output beam waist would be
1577 mm from the center of M1A, still inside Cab-Cass (Fig. 3). Therefore, completing the fre-
quency independent system for Cab-A requires the addition of at least two more focusing mirrors.
The methodology for the implementation of a frequency independent system formed by N-
focusing mirrors is well described in.26 In an optical system formed by N components (Fig. 10),
the elements of the ABCD matrix describing the entire system (MN ) can be described in terms
of the elements of the MN −1 matrix. For the current configuration of three mirrors (N = 3), M2
is the system formed by mirrors M1A and M2A. A new focusing component with focal length f3
(i.e., mirror M3A) and propagation distance to the new target plane dN +1 (i.e., d4 ) is added. The
following equations allow the design of the system:
A2
C3 = C2 − (17)
f3
1 1 D2
= + (18)
f3 d4 B2
d4
ωout = ωin (19)
|B2 |
  
1 1 B2 B2
= 1+ A2 + (20)
Rout d4 d4 Rin
A, B, C and D are the elements of the matrix that describes the system. The sub-index corre-
sponds to the optical component number.

13
Fig 10 Schematic of the quasi-optical system to be implemented in Cab-A for LT. fM 1A , fM 2A , and fM 3A are the
focal lengths of mirrors M1A, M2A, and M3A respectively. d1 and d4 are the distances between mirrors and focal
planes. d2 and d3 are distances between mirrors.

The NACOS configuration requires that frequency independence occurs between both focal
planes (i.e., ωCASS and ωN ASS ). This implies that Rin and Rout are both infinite. Introducing these
constraints in Eq. (20) along with Eq. (19) yields:

A2 = 1. (21)

Since the beam converges towards the output beam waist, the value of B2 is negative. This
implies that |B2 | is equal to −B2 and introducing this condition in Eq. (19), the value of B2 can
be calculated as

B2 = −d4 . (22)
On the other hand, the magnification of the entire system (M3 ) must be equal to one, which
implies that the element C3 of M3 is equal to zero, i.e. an afocal system. This condition, combined
with Eq. (21), leads to the following relation:
1
C2 = . (23)
fM 3A
The matrix M2 is defined as,

     
1 d3 1 0 1 d2 1 0 1 d1
M2 = Md3 MM 2A Md2 MM 1A Md1 = ,
0 1 − fM12A 1 0 1 − fM11A 1 0 1

in which replacing d1 = 803 mm (distance from Cassegrain focal plane to center of mirror
M1A), d2 = 3800 mm (distance between the centers of mirrors M1A and M2A) and fM 1A =
590 mm (focal length of mirror M1A) gives

     
d3 d3
5.441 fM 2A − 0.0017d3 − 5.441 568.86 fM 2A − 0.361d3 − 568.86
M2 =      . (24)
5.441 fM12A − 0.0017 568.86 fM12A − 0.361

14
 Combining Eqs. (21), (22) and (23) with the elements of M2 [Eq. (24)], the design Eqs. (25),
(26) (also plotted in Fig. 11) and (27) for the system are found:
 
d3
5.441 − 0.0017d3 − 5.441 = 1 (25)
fM 2A
 
d3
568.86 − 0.361d3 − 568.86 = −d4 (26)
fM 2A
 
1 1
5.44 − 0.0017 = . (27)
fM 2A fM 3A

Fig 11 Relationship for the quasi-optical design of Cab-A for LT system: d3 and fM 2A (left axis), d3 and d4 (right
axis).

From Fig. 11 it is seen that d3 ∼ 5.5d4 + 570 mm. This means that d3 increases faster than d4 .
Therefore d4 = 150 mm is established as a good compromise between the location of the mirror
M3A outside the cryostat and the distance between mirrors M2A and M3A (d3 = 1385 mm). The
set of parameters that completes the optical characteristics of the system for Cab-A are:
Table 5 Final optical parameters of Cab-A LT system.
Parameter Value (mm)
fM 1A 590
fM 2A 857.4
fM 3A 215
d1 803
d2 3800
d3 1385
d4 150

15
 The QO propagation of the fundamental Gaussian beam from the feed horn, along the optical
system up to the sub-reflector, was performed for five spaced frequencies on each band. ( e.g., B6
for Cab-A in Fig. 12). From this analysis, the size and radii of curvature for the focusing mirrors
were obtained, as well as the evaluation of the system optical performance [Eqs. (7) to (12)]. Refer
to Appendix B for details.

Fig 12 Gaussian beam propagation from the B6 receiver’s feed horn up to the sub-reflector, for the Cab-A optical
system configuration.

To accommodate the optical system, a series of flat mirrors, some removable and others reori-
entable with motors, completes the design. This setup allows the beams to be directed such that the
same support structure used in Cab-B (i.e., NASS-B) can be reused in Cab-A as NASS-A, elimi-
nating the need for a new structure in Cab-A The 3D model of NACOS-LT for Cab-A is shown in
Fig. 13. consists of two optomechanical structures; the CASS (common for both Nasmyth cabins)
and the NASS-A (in Cab-A).

16
Fig 13 Configuration of the optical system of NACOS-LT for Cab-A. The Gaussian beam propagation at 35 dB
(99.96%) contour power is also shown for B6 (blue), B7 (green) and B9 (read).

The beam focused by the Cassegrain system in the antenna’s focal plane (ωCASS ) propagates
downstream towards mirror M1A. This is possible either by remotely removing mirror flat D from
the optical path (OP) or by replacing it with a high-pass dichroic filter (HPDF) in a future upgrade
to allow transmission of higher frequencies (i.e., above B5). The beam is progressively reflected
and refocused, first by the off-axis elliptical mirror M1A, which refocuses the beam through the
elevation encoder and the Nasmyth Tube A towards Cab-A. The beam is then reflected by the M2A
off-axis elliptical mirror, and finally by R Flat, a rotating mirror that allows the selection of the any
of the following optical channels in Cab-A:

1. Band 6: With R Flat oriented in plane Y-Z (+Y direction), the beam is reflected towards Flat
1-B6, Flat 2-B6, M3A-B6 off-axis elliptical mirror, and finally onto the Nasmyth focus of
B6 (ωN ASS−B6 ).

17
 2. Band 7: With R Flat oriented in plane X-Z (-X direction), the beam is reflected towards Flat
1-B7, Flat 2-B7, M3A-B7 off-axis elliptical mirror, and finally onto the Nasmyth focus of
B7 (ωN ASS−B7 ).

3. Band 9: With R Flat oriented in plane Y-Z (-Y direction), the beam is reflected towards Flat
1-B9, Flat 2-B9, M3A-B9 off-axis hyperbolic mirror, and finally onto the Nasmyth focus of
B9 (ωN ASS−B9 ).

Two important features of the system are:

• The M1A mirror is supported by a rotating system that allows its complete removal from the
optical path. Together with the removal of mirrors Flat D and Flat Cass, this setup clears the
path, enabling the signal to propagate inside Cab-Cass for additional receivers, cameras, or
optical systems that may be installed in this cabin.

• In a future upgrade, a high-pass dichroic filter (HPDF) could be installed between mirrors
M2A and R Flat. This filter would reflect B6 while transmitting B7 and B9, enabling si-
multaneous observations in two bands in Cab-A. The analysis and design of the HPDF are
beyond the scope of this work.

The main parameters of the mirrors for the LT configuration are presented in Table 6.

Table 6 Mirror parameters for the LT configuration in Cab-A. Position (X; Y; Z), Surface normal (nx; ny; nz), Diameter
(∅) and Focal length (f ).
Mirror Band X(mm) Y(mm) Z(mm) nx ny nz ∅(mm) f (mm)
M1A 6,7,9 0 0 -803 -0.7071 0 0.7071 277x178 590
M2A 6,7,9 -3800 0 -803 0.7071 0 -0.7071 329x223 857.4
R Flat 6 -3800 0 -1583 0 0.5000 0.8660 115 ∞
Flat 1-B6 6 -3800 199.06 -1468.09 0 -0.6534 -0.7570 67 ∞
Flat 2-B6 6 -3800 100.02 -1718.15 0 0.8271 0.5620 52 ∞
M3A-B6 6 -3800 206.25 -1718.15 0 -0.6923 -0.7216 68x47 215
R Flat 7 -3800 0 -1583 -0.5000 0 0.8660 115 ∞
Flat 1-B7 7 -3984.69 0 -1476.38 -0.6227 0 -0.7825 68 ∞
Flat 2-B7 7 -3900 0 -1753.02 0.8039 0 0.5948 52 ∞
M3A-B7 7 -4002.55 0 -1753.02 0.7011 0 -0.7131 60x42 215
R Flat 9 -3800 0 -1583 0 -0.5000 0.8660 115 ∞
Flat 1-B9 9 -3800 -195.41 -1470.18 0 0.6439 -0.7651 62 ∞
Flat 2-B9 9 -3800 -100 -1730.02 0 -0.82 0.5724 49 ∞
M3A-B9 9 -3800 -202.53 -1730.02 0 0.7011 -0.7131 50x35 215

2.2.2 Cab-B
For the LT configuration of the optical system in Cab-B, bands 9 and 6 of FL are replaced by bands
1 and 2+3 respectively. In the case of B5, the OP for LT remains the same as for FL.
Bands 1 and 2+3 were integrated into the system by reusing as many optical components as
possible from the FL version. Some flat mirrors were replaced or added to provide the required
beam path length. The reference beam waist ωN ASS for the design was based on the central fre-
quency of each band (Appendix A). It was placed at 1300.5 mm from M4B for B1 and 1299 mm
for B2+3. the optimal distance for a perfect beam radius match on the sub-reflector should have

18
been 1188 mm for B1 and 1272 mm for B2+3. The defocus losses (∼ 16λB1 and ∼ 9λB2+3 ) are
negligible compared to the truncation losses from the existing system (mirrors and Nasmyth Tube
B) when placing the ωN ASS in the ideal position.
The 3D model of NACOS-LT for Cab-B is shown in Fig. 14.

Fig 14 Configuration of the optical system of NACOS-LT for Cab-B. The Gaussian beam propagation at 30 dB contour
power is also shown for B1 (brown), B2+3 (orange) and B5 (purple).

After propagating through the common section of all system bands (Sec. 2.1), the system
allows the selection of the following optical channels:

1. Band 5: The optical system is exactly the same used for NACOS-FL (Sec. 2.1).

2. Band 1: Removing the Flat Grid from the OP allows the beam to propagate further down,
where it is progressively reflected (in the Y-Z plane) by four flat mirrors, named flat B1-1 to

19
 Flat B1-4. The signal is then focused by a lens into the feed inside the cryostat. The Nasmyth
focus of B1 (ωN ASS−B1 ) at the central frequency is located between mirrors Flat B1-2 and
Flat B1-3.

3. Band 2+3: By removing the Flat Grid (used for B5) and the Flat B1-1 (used for B1) from
the OP, the beam propagates further down. It is progressively reflected (in the plane X-Z,
towards +X direction) by two flat mirrors, Flat B23-1 first and Flat B23-2 second. The
signal is then focused by the lens into the feed inside the cryostat. The Nasmyth focus of
B2+3 (ωN ASS−B2+3 ) of the central frequency is located between mirrors Flat B23-1 and Flat
B23-2.

Both Flat D and Flat Grid are metallic mirrors that could be replaced in an upgrade phase of the
LT configuration. Flat D could be replaced by a HPDF, reflecting lower frequencies to Cab-B while
transmitting higher frequencies (above B5) to Cab-A. Flat Grid could be replaced by a polarizing
grid that reflects one polarization state and transmits the orthogonal one. The analysis of the HPDF
and the polarizing grid is beyond the scope of this work.
The parameters of the mirrors for NACOS-LT configuration for Cab-A are presented in Table
7.
Table 7 Mirror parameters for LT in Cab-B configuration. Position (X; Y; Z), Surface normal (nx; ny; nz), Diameter
(∅) and Focal length (f ).
Mirror Band X(mm) Y(mm) Z(mm) nx ny nz ∅(mm) f (mm)
Flat D 1,2+3,5 0 0 -423 -0.3090 0 0.9510 155 ∞
Flat Cass 1,2+3,5 -87.22 0 -303 0.4540 0 -0.8910 176.4 ∞
Flat 1 1,2+3,5 670.50 0 -303 -0.7071 0 -0.7071 387.6x275 ∞
M3B 1,2+3,5 670.50 0 -803 0.7071 0 0.7071 519.7x369.7 1564.75
M4B 1,2+3,5 3800 0 -803 -0.7071 0 -0.7071 425x300 1564.75
Flat Grid 5 3800 0 -1153 0 -0.7071 0.7071 345 ∞
Flat 2 5 3800 -229.72 -1153 0 0.6923 -0.7216 222x158 ∞
Flat B1-1 1 3800 0 -1453 0 0.6560 0.7547 320 ∞
Flat B1-2 1 3800 457.39 -1390.72 0 -0.6560 -0.7547 250 ∞
Flat B1-3 1 3800 457.39 -1613 0 -0.7071 0.7071 280 ∞
Flat B1-4 1 3800 200.39 -1613 0 0.6916 -0.7222 280 ∞
Flat B23-1 2+3 3800 0 -1753 0.4656 0 0.8850 195 ∞
Flat B23-2 2+3 3800 0 -1615.34 -0.1390 0 -0.9903 160 ∞

The QO Aperture efficiency and beam coupling analysis for all the bands of the system are
summarized in Fig. 15. The aperture efficiency plot shows that B1 and B2+3 do not exhibit fully
frequency-independent behavior, meaning efficiency does not remain constant across the frequency
range. This is due to variations in the receiver focus position (ωN ASS ) with the frequency (Ap-
pendix A). For B2+3, efficiency decreases because the size of ωN ASS , causes under-illumination
of the sub-reflector, resulting in an edge taper value higher than the expected 12 dB (Appendix A).
The coupling efficiency plot indicates that B9 for Cab-B has a total coupling efficiency Ktotal of
∼ 2.5% less than the others. This is due to the location of ωN ASS−B9 , which, for unknown reasons
to the author of the present work, was placed at 1285.15 mm (i.e., ∼ 34λ661 GHz ) from M4B instead
of the ideal position of 1300.50 mm.

20
 Fig 15 QO results of aperture efficiency (top) and beam coupling (bottom) for the whole system.

3 Physical Optics Simulations

The optical performance of the system for each receiver (presented in Sec. 2), was analyzed
through simulations carried out with GRASP,27 a dedicated software package for reflector systems
widely used in the design, verification and optimization of advanced reflector antenna systems.
GRASP uses the Physical Optics (PO) algorithm, which enables the prediction and analysis of the
entire far-field and near-field patterns of the system.
In this work, two sets of simulations were performed for the lower, central, and higher frequen-
cies of each receiver channel. The first set of simulations is based on propagating the fundamental
Gaussian beam from ωCASS to an evaluation plane located at ωN ASS , passing through the entire
tertiary optical system. This set of simulations is identified as reception mode. The second set
of simulations is based on propagating the fundamental Gaussian beam from ωN ASS to the sky,
passing through the entire tertiary optical system, the Nasmyth tubes, encoder and the Cassegrain
system. This set of simulations is identified as transmission mode. In each set of simulations, a
series of merit figures were evaluated.
Reception mode merit figures:

• Beam coupling, which is a measure of how much of the energy contained in the beam pro-
duced by the system (Eout ) is coupled with a fundamental Gaussian beam (EGauss ) in the
evaluation plane. It can be obtained by
RR ∗ 2
Eout EGauss dxdy
ηcoupling = RR ∗
RR ∗
, (28)
Eout Eout dxdy EGauss EGauss dxdy

where ∗ means the complex conjugate.

21
 • Gaussicity,28 which is a measure of how well the beam produced by the system (Eout ) main-
tains its Gaussian shape, indicates the aberration level in the beam. To obtain the Gaussicity,
Eq. (28) can be used by replacing EGauss with the fitted Gaussian distribution of Eout , which
can be obtained by
 2 2


−
(x−xof f set ) +
(y−yof f set ) 
ω0x 2 ω0y 2
EGauss−f it = e . (29)

• Ellipticity, which indicates how much the beam deviates from the ideal circular shape of the
fundamental Gaussian beam mode, can be determined by

min(ω0x , ω0y )
Ellipticity = 1 − , (30)
max(ω0x , ω0y )
where ω0x and ω0y are the beam waist radius obtained from the Gaussian fit [Eq. (29)].

• Cross-polarization level induced by the system.

Transmission mode merit figures:

• Antenna aperture efficiency.

• Antenna beam pattern on the sky, from which the Full Width to Half Maximum (FWHM),
first side lobe level (left and right) and pointing (in Azimuth and Elevation coordinates) were
obtained.

• Cross-polarization level of the whole system.

The simulations also allowed the determination of the level of beam spillover and clipping,
thereby enabling an assessment of the noise contribution of the optical system. The procedure
implemented follows that described in.29 Additionally, the contribution of the emission of mirrors
due to their physical temperature (Tphys ) was added to the source temperature noise as Tsource,n =
Tsource Rso + εTphys Rso , where Tsource is the receiver noise temperature for the first mirror of
the system. The receiver noise temperature for B6 is TRX = 50 K, as reported by,30 which is
practically the same temperature at the exit window of the LLAMA cryostat. Here, Rso repre-
sents the relative power from the source falling on the reflector, and ε denotes the emissivity of
the mirror, with a value of 0.005 for the NACOS mirrors, and values of 0.003 and 0.02 for the
sub-reflector and primary reflector, respectively, as stated by.31 The contribution to the noise
temperature due to the emissivity of each mirror of NACOS is ∼ 1.5 K. With a total noise
temperature contribution of ∼ 9 K for receivers in Cab-A, and ∼ 11 K for receivers in Cab-B
(∼ 18 K for B1), NACOS will increase the observing time of LLAMA to achieve the same sensi-
tivity as the receivers placed at the telescope focal plane, according to the radiometer equation,32
∆tobs = [(TRX + Tnoise−N ACOS + Tnoise−T elescope )/(TRX + Tnoise−T elescope )]2 (see Table 11).
The PO analysis of B6 for Cab-A is presented as an example. The beam pattern on the Nasmyth
focal plane (Fig. 16), the antenna beam pattern on the sky (Fig. 17), and the noise temperature
contribution 8 are presented. The results for all the bands of Cab-A and Cab-B, including both FL
and LT configurations are summarized in Tables 9, 10, and 11.

22
 (a) (b) (c)

Fig 16 Plots of the beam pattern on the Nasmyth focal plane for B6 in Cab-A: (a) Co-polar plot. (b) Cross-polar plot.
(c) Cut plot. Top row: 211 GHz. Middle row: 243 GHz. Bottom row: 275 GHz.

23
 Fig 17 Antenna beam pattern on the sky for B6 (243 GHz) in Cab-A.

Table 8 Noise temperature contribution for B6 (243 GHz) in Cab-A.

The Tphys is 296 K and 300 K inside and outside the antenna respectively and the Accum. power is the accumulated
power throughout the optical path. The sub-reflector surrounding temperature is obtained from,33 with the telescope
pointing to zenith. Temperatures with subscript n are the noise contribution.
M3A-B6 Flat 2-B6 Flat 1-B6 R Flat M2A Encoder (bf) Encoder (ff) M1A Sub-ref. Primary ref.
Tsource (K) 50 51.505 53.027 54.568 56.082 57.601 57.683 57.743 59.317 64.422
Tsurr (K) 296 296 296 296 296 296 296 296 2.798* 300
Tphys (K) 296 296 296 296 296 296 296 296 300 300
ε 0.005 0.005 0.005 0.005 0.005 0 0 0.005 0.003 0.02
Rso 0.99993 0.99988 0.99982 0.99990 0.99989 0.99977 0.99983 0.99973 0.93578 0.98586
Accum. power 0.99993 0.99981 0.99963 0.99953 0.99942 0.99919 0.99902 0.99875 0.93462 0.92140
Tsource,n (K) 51.484 52.991 54.516 56.053 57.568 57.615 57.693 59.238 64.230 71.262
Tsurr,n (K) 0.0216 0.0361 0.0518 0.0284 0.0335 0.0684 0.0501 0.0794 0.1920 4.3037
TRef l,n−out (K) 51.505 53.027 54.568 56.082 57.601 57.683 57.743 59.317 64.422 75.565
TRef l,n−add (K) 1.505 1.522 1.541 1.514 1.520 0.0820 0.060 1.575 5.105 11.143
Noise cont.(%) 5.89 5.95 6.03 5.92 5.94 0.32 0.23 6.16 19.97 43.59

24
 Table 9 Optical performance of the tertiary system.
Gaussian beam launched from CASS to NASS
Freq. Beam Gaussicity ω0x ω0y ω0expect xof f set yof f set Ellipticity Cx-pol
Band Config.
(GHz) coupling(%) (%) (mm) (mm) (mm) (mm) (mm) (%) (dB)
211 99.74 99.75 8.54 8.55 8.51 -0.0672 0.0048 0.16 -29.48
B6 243 Cab-A LT 99.80 99.81 7.42 7.42 7.39 -0.0478 0.0261 0.01 -30.20
275 99.83 99.84 6.56 6.56 6.53 -0.0339 0.0405 0.06 -30.72
275 99.81 99.81 6.57 6.56 6.53 -0.0257 0.0000 0.13 -58.02
B7 324 Cab-A LT 99.86 99.86 5.57 5.56 5.54 -0.0304 0.0000 0.16 -61.36
373 99.88 99.88 4.84 4.83 4.81 -0.0319 0.0000 0.18 -62.15
602 99.88 99.88 3.00 3.00 2.98 0.0080 0.0084 0.10 -32.25
B9 661 Cab-A LT 99.88 99.89 2.73 2.73 2.72 0.0100 0.0054 0.08 -32.36
720 99.88 99.89 2.51 2.51 2.49 0.0115 0.0031 0.10 -32.47
35 98.81 99.01 48.86 50.41 50.76 1.1284 0.0780 3.17 -26.73
B1 42.5 Cab-B LT 99.45 99.51 42.55 43.57 42.54 0.7730 0.0209 2.41 -26.90
52 99.78 99.78 35.07 35.48 34.82 0.0400 0.0026 1.17 -27.11
67 99.88 99.90 27.87 27.73 27.39 0.2416 0.0000 0.49 -27.26
B2+3 91 Cab-B LT 99.85 99.89 20.97 20.88 20.70 0.0366 0.0000 0.46 -27.41
116 99.84 99.87 16.30 16.24 16.12 0.0233 0.0000 0.39 -27.28
163 99.82 99.83 11.17 11.12 11.02 0.0252 0.0145 0.45 -26.96
B5 187 Cab-B FL/LT 99.81 99.82 9.74 9.71 9.60 0.0257 0.0150 0.31 -26.96
211 99.81 99.82 8.63 8.60 8.51 0.0260 0.0153 0.35 -26.95
211 99.60 99.62 8.66 8.63 8.51 0.0548 0.0000 0.36 -26.95
B6 243 Cab-B FL 99.51 99.53 7.53 7.51 7.39 0.0543 0.0000 0.35 -26.94
275 99.41 99.44 6.67 6.64 6.53 0.0542 0.0000 0.35 -26.94
602 98.07 98.10 3.12 3.12 2.98 0.0268 0.0051 0.19 -26.93
B9 661 Cab-B FL 97.71 97.74 2.86 2.86 2.72 0.0269 0.0050 0.17 -26.96
720 97.32 97.36 2.65 2.64 2.49 0.0271 0.0049 0.15 -26.93

Table 10 Optical performance of the radiotelescope.

Gaussian beam launched from NASS to sky
Freq. Ap. eff Illum. over FWHM (arcsec) Peak (arcsec) Side lobes (dB) Cx-pol
Band Config.
(GHz) (%) main refl.(dB) (Max ; Min) (Az ; El) (left ; rigth) (dB)
211 78.42 10.77 (28.76 ; 28.68) (0.47 ; -0.04) (-25.10 ; -24.21) -31.67
B6 243 Cab-A LT 78.70 11.05 (24.99 ; 24.95) (0.40 ; 0.00) (-25.36 ; -24.30) -32.10
275 79.04 11.17 (22.06 ; 22.03) (0.32 ; 0.04) (-25.40 ; -24.24) -32.54
275 79.01 11.15 (22.06 ; 22.06) (0.03 ; 0.00) (-24.38 ; -25.05) -66.70
B7 324 Cab-A LT 79.39 11.24 (18.70 ; 18.69) (0.03 ; 0.00) (-24.41 ; -24.92) -67.55
373 79.50 11.23 (16.23 ; 16.22) (0.04 ; 0.00) (-24.61 ; -24.98) -67.96
602 79.69 11.27 (10.04 ; 10.04) (0.09 ; 0.04) (-24.67 ; -24.79) -34.06
B9 661 Cab-A LT 79.67 11.26 (9.15 ; 9.14) (0.08 ; 0.03) (-24.70 ; -24.88) -34.16
720 79.66 11.25 (8.40 ; 8.40) (0.07 ; 0.02) (-24.72 ; -24.93) -34.23
35 67.88 6.26 (171.00 ; 163.80) (5.82 ; 0,00) (-20.32 ; -19.37) -30.97
B1 42,5 Cab-B LT 73.81 8.50 (144.00 ; 138.00) (3.80 ; 0.00) (-21.96 ; -20.35) -29.64
52 75.88 10.05 (106.80 ; 115.20) (1.78 ; 0,00) (-23.19 ; -21.51) -28.87
67 76.94 11.11 (91.52 ; 91.39) (0.18 ; 0.00) (-24.53 ; -26.28) -28.69
B2+3 91 Cab-B LT 77.29 11.93 (67.89 ; 67.79) (-0.12 ; 0.00) (-25.29 ; -26.20) -28.67
116 77.34 11.70 (52.96 ; 52.93) (0.10 ; 0.00) (-24.31 ; -25.68) -28.66
163 78.89 11.19 (37.22 ; 37.20) (-0.14 ; -0.04) (-24.86 ; -24.87) -28.47
B5 187 Cab-B FL/LT 78.87 11.16 (32.44 ; 32.43) (-0.12 ; -0.04) (-24.89 ; -24.90) -28.48
211 78.85 11.16 (28.76 ; 28.74) (-0.11 ; -0.04) (-24.90 ; -24.92) -28.48
211 78.87 11.20 (28.77 ; 28.74) (-0.18 ; 0.00) (-25.10 ; -24.78) -28.48
B6 243 Cab-B FL 78.84 11.19 (24.98 ; 24.96) (-0.16 ; 0.00) (-25.07 ; -24.71) -28.47
275 78.85 11.21 (22.06 ; 22.06) (-0.16 ; 0.00) (-25.05 ; -24.60) -28.47
602 78.58 11.25 (10.07 ; 10.06) (-0.13 ; -0.01) (-25.06 ; -23.70) -28.45
B9 661 Cab-B FL 78.46 11.26 (9.17 ; 9.17) (-0.13 ; -0.01) (-24.47 ; -23.51) -28.45
720 78.31 11.27 (8.42 ; 8.42) (-0.13 ; -0.01) (-24.34 ; -23.31) -28.44

25
Table 11 Noise temperature contribution for FL and LT configurations. Tnoise−N ACOS is the noise temperature added
by the NACOS system. Tnoise−T elescope is the noise temperature added by the primary and sub-reflector, considering
an effective sky temperature obtained from33 with the telescope pointing to zenith. ∆tobs is the increase in observing
time attributed to NACOS.
Freq. TRX Tnoise−N ACOS Tnoise−T elescope ∆tobs
Band Config.
(GHz) (K) (K) (K) (%)
B6 243 Cab-A LT 50 9.317 16.248 30.1
B7 324 Cab-A LT 72 9.102 21.893 20.3
B9 661 Cab-A LT 105 9.182 24.611 14.7
B1 42.5 Cab-B LT 28 17.637 18.503 90.2
B2+3 91 Cab-B LT 40 10.727 16.249 41.8
B5 183 Cab-B FL/LT 50 10.872 16.845 35.2
B6 243 Cab-B FL 50 10.778 16.149 35.2
B9 661 Cab-B FL 105 10.771 24.826 17.3

Based on the data presented in Tables 9, 10, and 11, the following observations can be made:

• For LT configuration, the beam coupling and Gaussicity are higher than 99.5% for all cases
except for B1. Losses and distortion for B1 are attributed to beam clipping while propagating
through mirrors Flat D and Flat Cass, with an energy loss of ∼ 2% at 35 GHz and ∼ 1.2% at
42.5 GHz, and some distortion on the beam (ellipticity ∼ 3%). The Nasmyth tube clips the
35 GHz beam at 26 dB (29 dB at 42.5 GHz). Additional losses of ∼ 5% at 35 GHz and ∼ 2%
at 42.5 GHz is produced by beam spillover along the optical train of the common mirrors.
∼ 70% of this spillover is produced at mirrors Flat D and Flat Cass (Fig. 18). Attempts to
reduce spillover in these two mirrors increase the beam clipping, keeping the energy losses
almost constant.

• For FL, the beam coupling and Gaussicity of B9 is reduced due to the defocus loss mentioned
at the end of Sec. 2.2.

• The aperture efficiency of B1 is reduced due to energy loss from beam spillover and inter-
ference along the common optics of the optical train..

• The contribution of NACOS to the system’s noise temperature increases LLAMA’s observ-
ing time by ∼ 22% for Cab-A and ∼ 32% for Cab-B (∼ 90% for B1).

26
 Fig 18 B1 beam clipping and spillover on the optical train of common mirrors.

4 Tolerance Analysis
Tolerance analysis in optical systems is essential to evaluate how misalignment of optical compo-
nents and manufacturing tolerances can affect the overall performance of the optical system. Even
in systems like LLAMA, where the optical components are adjustable in translations and tilts, this
analysis is crucial to understand the impact of perturbations on optical characteristics.
Misalignment of individual mirrors have different effects on the performance of the system.
Flat mirrors with lateral offset have almost no effect on the optical performance, except for the
amount of energy that spills over the offset mirror. On the other hand, curved mirrors with focal
length f will produced an angular deviation in the beam of ∼ ∆/f . An axial offset ∆ in a tilted
flat mirror will result in a ∆ offset in the beam and an equal increment in its path length, leading
to defocus. An axial offset ∆ in a curved mirror will result in a defocus which also depends on f ,
according to the thin lens equation (1/din + 1/dout = 1/f ). A tilt α in any mirror results in a 2α
beam angular deviation. Misalignment of mirrors can be caused by both static factors (e.g., parts
manufacture, assembly, mounting) and dynamic factors (e.g., telescope’s EL position, temperature
variations). In the present work, we focus only on static factors.
Degradation in aperture efficiency, errors in telescope pointing, and increases in cross-polarization
levels are critical parameters for the telescope’s performance. Additionally, increases in sidelobe
levels and the FWHM size, though less critical, are also undesirable.
The decrease in aperture efficiency primarily occurs when the primary reflector is not symmet-
rically illuminated at its center, indicating a radial displacement in the aperture plane of the radio
telescope.34 This loss can be estimated using the coupling integral [Eq. (28)] onto the sub-reflector,
considering the beam from the sky with an Airy pattern produced by the radio telescope aperture,
and an offset fundamental Gaussian beam with a 12 dB edge taper representing the receiver horn
emission. The aperture efficiency loss estimation with respect to the beam offset (i.e., xof f set ) at
the sub-reflector is presented in Fig. 19 and accurately fitted with

Apef f loss estim. [%] = −1.17x10−9 (xof f set [mm])4 + 0.00052(xof f set [mm])2 . (31)

27
 Fig 19 Coupling loss at the sub-reflector due to receiver’s beam offset.

For LLAMA, the target in aperture efficiency loss is ≤ 1%, although a loss of up to 2% is con-
sidered acceptable. According to Eq. (31), ∼1% loss is produced by an illumination offset at the
sub-reflector of 41.3 mm, equivalent to 11% of its radius (Rs =375 mm). This offset represents a tilt
of the chief ray at the Cassegrain telescope focal plane of 0.4 ◦ (arctan(41.3 mm/5882.86 mm)).
The precise pointing of the telescope beam is crucial for radio astronomical observations with
high spatial resolution. While the telescope’s movement control system primarily ensures pointing
accuracy, any misalignment in the optical system can cause a deviation between the telescope’s
geometric and optical axes. In telescopes with a single receiver, this deviation can be corrected
within the pointing algorithms. However, in telescopes like LLAMA, which incorporate multiple
receivers sharing numerous optical components, especially with the ultimate goal of multi-band
observation, the situation is more complex. Ensuring the correct alignment of the optical system
is imperative to guarantee that the pointing in the sky is, within a certain margin, consistent for
all receivers simultaneously. A pointing error of FWHM/10 results in a reduction of the intensity
coupled by the telescope of ∼ 3%, representing the accepted limit in most cases.35–37
For LLAMA, B9 is the band with highest resolution (FWHM=9.15 arcsec at its central fre-
quency of 661 GHz), therefore, the maximum admissible pointing error is 0.915 arcsec, which
represents a lateral offset of the chief ray at the Cassegrain telescope focal plane of 0.43 mm
(tan(0.915 arcsec) · 96000 mm).
Misalignment of the mirrors will also degrade the optical performance of the system due to
defocusing and spillover, impacting the size and asymmetry of the telescope’s main beam (i.e.,
FWHM), the intensity level of the side lobes, and the level of cross-polarization (Cx-pol). The
LLAMA requirements for these parameters are: FWHM size increment ≤2%, sidelobe level ≤-
22 dB and Cx-pol efficiency ≥99.5% (-23 dB).

28
 There are different analysis techniques to correlate mechanical uncertainties with deviations in
the optical parameters of the system: The Worst Case Scenario (WCS), which consists of adding
all the absolute maximums uncertainties to obtain in this way the worst level of expected optical
performance; The Root Sum of Squares (RSS), which considers that the uncertainties of each
pPn are independent and can be described from a normal or Gaussian distribution (σsystem =
variable
2
i σi ). RSS provides results closer to the reality in comparison to WCS, but it does not provide
information about the statistical behavior of the system. The Monte Carlo (MC) analysis,38 in
which the uncertainties of each variable are characterized by a probability distribution, such that
the variable has an associated probability of assuming a given value within a defined interval. The
position and orientation of each component can then be modeled from a probability distribution,
with a normal distribution used in this work. The mean value is set to the design value, and
the standard deviation equals the manufacturing tolerance. By assigning a random value to each
variable of the components based on its probability distribution, a real (probable) configuration of
the system is generated, allowing the determination of optical performance. Conducting a large
number (≥ 103 ) of individual calculations enables the statistical analysis of the system’s optical
performance based on the compilation of individual results.
In the current work, we performed a tolerance analysis of NACOS to evaluate how the fabri-
cation tolerances of the mirrors impact the overall optical system performance. This was done by
implementing the MC technique with two methodologies, each varying in complexity, computing
time, and result accuracy:

1. The ray tracing (RT) through the implementation of ABCD matrix formalism, in which the
errors in position and orientation were incorporated into the matrices of individual com-
ponents. This approach allowed for rapid and reasonably accurate calculations of optical
parameters such as aperture efficiency and pointing.

2. PO simulations in GRASP, where errors in position and orientation were incorporated into
the coordinate system of each component. This method provided highly accurate results for
aperture efficiency, pointing, cross-polarization, sidelobe levels, and FWHM.

4.1 Monte Carlo Implemented in RT

The RT method using the ABCD matrices can be extended to scenarios where the system compo-
nents exhibit small linear and/or angular deviations.39 In a misaligned system confined to a plane,
the pointing error (∆rout and ∆θout ) of the beam after having propagated through a component,
can be characterized by a misalignment matrix Mmis multiplied by an error vector ϵ.40
 
    ∆X
∆rout 1 − cos (2θ) 0 sen (2θ) 
= Mmis .ϵ = ∆α  . (32)
∆θout − f1 −2 0
∆Z
The Mmis matrix incorporates the focal length of the component (f ) and the angle of incidence
concerning the nominal optical axis (θ). The error vector ϵ encompasses both linear (∆X and ∆Z)
and angular (∆α) alignment errors of the component.
The error introduced by each optical component (ϵi ) in the system can be incorporated into
the cascade multiplication of the ABCD matrix, as outlined in Sec. 2. This approach enables
tracking the path of an incoming beam (Bin = [rin ; θin ]) throughout the entire system, up to

29
its exit (Bout = [rout ; θout ]). As an example, the misaligned beam at the telescope focal plane
(BF P = [rF P ; θF P ]) resulting from the propagation of the input beam (BωN ASS = [rωN ASS ; θωN ASS ]
at the Nasmyth focal plane) along the B9 system for Cab-A configuration (Fig. 20) is described by

BF P = MB9 BωN ASS + Md7 MM 1A Md6 MM 2A Md5 MRF Md4 MF 1 Md3 MF 2 Md2 MM 3Amis ϵM 3A
+ Md7 MM 1A Md6 MM 2A Md5 MRF Md4 MF 1 Md3 MF 2mis ϵF 2
+ Md7 MM 1A Md6 MM 2A Md5 MRF Md4 MF 1mis ϵF 1
+ Md7 MM 1A Md6 MM 2A Md5 MRFmis ϵRF
+ Md7 MM 1A Md6 MM 2Amis ϵM 2A
+ Md7 MM 1Amis ϵM 1A ,
(33)

where MB9 is the matrix that describes the aligned optical system and is given by

MB9 = Md7 MM 1A Md6 MM 2A Md5 MRF Md4 MF 1 Md3 MF 2 Md2 MM 3A Md1 ,

being Mj and Mdi the matrices correspondent to each j-mirror and i-distance respectively [Eq.
(6)].
The offset and tilt values obtained at the focal plane of the telescope (i.e., BF P ) can then be
projected onto the sub-reflector plane to estimate the aperture efficiency loss using Eq. (31).

30
 Fig 20 RT schematic of B9 for Cab-A configuration.

A script was written in which every component of Eq. (33) takes random values for ∆X, ∆Z
and ∆α in each run. These values were obtained from a normal distribution with a mean of zero
and 6σ equivalent to the mechanical tolerance range. For LLAMA, each mirror has three assembly
interfaces: the mirror to its support, the support to the mirror assembly support, and the assembly
support to the NACOS structure (CASS or NASS). The manufacturing assembly tolerance for
each instance is ±0.1 mm (±0.05 º) p for linear (angular) displacements. Hence, the tolerance of
each mirror assembly is tolassy = ± 3toli2 (≡ ±3σ which guarantees that 99.73% of the values
are inside tolerance). Thus, σX = σZ = 0.058 mm and σα = 0.029º.
A total of 104 runs were performed for B6 and B9 for Cab-A LT configuration, and B9 for Cab-
B FL configuration. The aperture efficiency and pointing distributions are shown in Fig. 21. Each
plot includes the nominal value and the probability of meeting the LLAMA requirements, derived
from the best-fit distribution determined using the Generalized Akaike Information Criterion.41

31
 (a) (b)

Fig 21 Distributions of: (a) Aperture efficiency estimation and (b) pointing. Top row: B6 for Cab-A. Middle row: B9
for Cab-A. Bottom row: B9 for Cab-B.

In conclusion, the Monte Carlo analysis applied to RT shows that tolerances have a minor
impact on aperture efficiency. There is a 35% and 12% probability of degrading efficiency by
more than 1% and 2% respectively, for both B6 and B9 in Cab-A. The degradation is negligible
for B9 in Cab-B. Regarding the pointing, B9 for Cab-A has a probability of only 29% of satisfying
the acceptance criterion of being within FWHM/10. Although B6 presents the same dispersion
(σ ∼2.5 arcsec) in the pointing as B9, since FWHM of B6 is greater, the probability of satisfying
the acceptance criterion is higher (69%). The pointing of B9 for Cab-B has a probability of 11%
of satisfying the acceptance criterion.

32
4.2 Monte Carlo Implemented in GRASP
The Monte Carlo method was applied through 103 simulations using PO in GRASP for B6 and
B9 of Cab-A for LT configuration. Each simulation introduced random misalignments consistent
with those described in Sec. 4.1). Misalignment in position (∆X, ∆Y and ∆Z) and orientation
rotations around the X, Y, Z axes (∆a, ∆b and ∆c) were applied to the local coordinate systems of
each mirror, as illustrated in Fig. 22.

Fig 22 Coordinate system used for mirror misalignment on GRASP (M1A and M2A shown as example).

Aperture efficiency, pointing and Cx-pol distributions are shown in Fig. 23. Distributions for
FWHM (maximum and minimum values) and first side lobe level (left and right with respect to the
main lobe) are displayed in Fig. 24. The probability of aperture efficiency degradation exceeding
1% is 60% for B6 and ∼78% for B9. The likelihood of degradation beyond 2% is ∼23% for B6 and
∼67% for B9. The pointing shows a probability of fulfilling the acceptance criterion of ∼61% for
B6 and ∼11% for B9. The effect of the misalignment of the mirrors in the Cx-pol, FWHM and first
sidelobe level performance is negligible. Hence, B9 is the most sensitive channel to misalignment
of the optical system. In Fig. 25, the linear correlation between aperture efficiency and pointing
for B9 (the most sensitive channel of the system) with the linear and angular deviations of each
mirror is presented. Tilt with respect to Y-axis of both M1A and M2A mirrors presents the highest
correlation.

33
 (a) (b)

Fig 23 Distributions of: (a) B6 for Cab-A (b) B9 for Cab-A. Top row: Aperture efficiency. Middle row: Pointing.
Bottom row: Cx-pol level.

34
 (a) (b)

Fig 24 Distributions of FWHM and first side lobe level of: (a) B6 for Cab-A (b) B9 for Cab-A.

35
Fig 25 Linear correlation between aperture efficiency and pointing with respect to mirror misalignment for B9 in
Cab-A.

In conclusion, while misalignment has a negligible effect on FWHM, sidelobe levels, and Cx-
pol, the MC analysis using PO confirms the Ray Tracing RT prediction of a very low probability,
particularly for B9, of meeting the pointing requirement. Additionally, degradation in aperture
efficiency is expected to exceed the requirement. Therefore, a precise alignment strategy was
developed to improve the overall optical performance of NACOS-LT.
This alignment strategy consists of two stages. First, a coarse alignment is performed by prop-
agating a laser beam through each subsystem (CASS or NASS) and visually evaluating it at a
measurement plane. The mirrors are adjusted to ensure the laser beam passes through pinholes
placed along the optical path, as shown in Fig. 26 (a). NACOS powered mirrors can be manually
adjusted in X, Y, Z, ∆a, and ∆b, whereas the flat ones in Z, ∆a, and ∆b.
The second stage involves fine-tune alignment using a positioning sensor device (PSD), placed
at the measurement plane as shown in Fig. 26 (b). This device tracks the laser beam with an
accuracy of ∼0.01 mm.
This alignment strategy was implemented during the AIV stage of NACOS-FL. B9 achieved a
pointing accuracy of 0.68 arcsec and an estimated aperture efficiency loss of 0.1%.

36
 (a) (b)

Fig 26 (a) Laser beam propagation along NASS during B9 alignment (b) PSD measurement of the output laser beam
from NASS.

The proposed alignment method will be implemented again during the integration of NACOS
in the radio telescope and will also serve as tool for realigning NACOS whenever necessary.
Although NACOS may experience some dynamic misalignment due to gravitational effects
from changes in the pointing of the radio telescope and/or temperature variations in the cabins,
these are expected to have only a minor impact on system performance. The optical components
expected to be affected by gravitational effects are those located in Cab-CASS, as this cabin is
the only one that moves with changes in the elevation angle of the radio telescope. The Nasmyth
cabins are unaffected by this movement.
Structural simulations of the support assembly and mirror M1A showed maximum deforma-
tions of 0.08 mm in position and 0.01 deg in orientation. Evaluating this misalignment in the RT
model results in a pointing error of ∼0.7 arcsec (∼8% of FWHM), and an estimated aperture effi-
ciency loss of 0.2% for B9. Simulations predict a similar response for the optical components of
NACOS for Cab-B.
Regarding temperature variations, the cabins housing CASS and NASS are equipped with a
thermal control system maintaining temperatures within ±1 K. Therefore, negligible deformations
are expected.

5 Conclusions
A complete quasi-optical design has been developed for the long-term phase of the tertiary optical
system (NACOS) of the LLAMA radio telescope. This was achieved by meeting the requirement
to reuse most of the components of the optical system implemented for the first light phase of the
project. Physical optics simulations were employed to complete the validation and characterization
of the optical performance of the system. Despite a minimal frequency dependence exhibited by
the two lower-frequency receivers (i.e., B1 and B2+3), a frequency independent system has been
achieved for the remaining receivers (i.e., B5, B6, B7, B9) of the radio telescope. The antenna
aperture efficiency obtained by considering a fundamental Gaussian beam illumination was greater

37
than 77% for all the receivers, except for B1, which reached ∼68% at its lower frequency because
of a truncation in the beam due to the reuse of mirrors from the first light optical system. These
values are not taking into account surface errors on reflectors. The design ensures aberration-free
beams for all receivers, with Gaussicity levels over 99%.
Additionally, a tolerance analysis was carried out to evaluate the degradation in the optical
performance of the system caused by errors in the positioning of the optical components during
the NACOS integration process. The analysis revealed that implementing an alignment strategy
for the system is essential. Such a strategy was developed and successfully implemented during
the assembly, integration, and verification phase of NACOS-FL.

Appendix A: Determination of B1 and B2+3 focii Position and Size

The size and position of the output beam waist from the horn-lens system were determined by sim-
ulating the propagation of the fundamental Gaussian beam from the phase location center (PLC)
inside the horn (Fig. 27). This beam was traced through the lens, and the resulting electromagnetic
field was evaluated in planes perpendicular to the optical axis at various positions. This process
identified the location where the beam reached its smallest size (ωout ≡ ωN ASS ). The resulting
beam waist parameters were then used as input for designing the optical system for the B1 and
B2+3 band.

Fig 27 GRASP set up for the evaluation of the output beam waist of the horn-lens system.

Table 12 summarized the parameters used for the simulations and the position and size obtained
of the output beam waist for the extreme and central frequencies of each band respectively.

38
 Table 12 B1 and B2+3 horn-lens optical parameters and output beam characteristics.
B1 B2+3
Parameter 35 GHz 42.5 GHz 52 GHz 67 GHz 91 GHz 116 GHz
PLC (mm) 1.39 5.21 7.45 1.50 5.50 3.80
ωhorn (mm) 9.40 9.38 9.40 4.80 4.80 4.80
lens material HDPE HDPE
Refraction index n 1.5259 1.5259
Loss tangent δ 2.73−4 2.73−4
Focal length f (mm) 181 91
Lens diameter (mm) 190 92
Lens thickness (mm) 40.8 21.38
Lens shape bi-hyperbolic bi-hyperbolic
distance lens to horn aperture (mm) 175.4 81
ωN ASS (mm) 50.76 42.54 34.82 27.39 20.70 16.12
Distance wrt cryo top plate (mm) 563 553 473 90.50 131.80 80.50
T e over the primary reflector (dB) 11.74 12.16 12.20 12.53 13.20 13.01

Appendix B: QO parameters of B6 for Cab-A propagation

The QO parameters obtained from the propagation of B6 for Cab-A configuration are presented in
Table 13.
The parameters that define the conical surface of each focusing mirror are shown in Fig. 28 and
detailed in Table 14. For the common mirrors M1A and M2A, the chosen parameters correspond
to the central frequency (i.e., 452 GHz) of the entire broadband range (i.e., 211 to 720 GHz), as
this minimizes loss due to phase distortion42 in the system.

39
 Table 13 QO parameters of B6 for CAB-A configuration.
Fundamental GBP CAB-A B6
Freq. (GHz) 211 227 243 259 275
ωN ASS (mm) 8.58 7.97 7.45 6.99 6.58
dN ASS−T OP P LAT E (mm) -14.85 -14.94 -15.01 -15.07 -15.13
dT OP P LAT E−M 3AB6 (mm) 135
fM 3AB6 (mm) 215
θM 3AB6 (deg) 43.81
RIN M 3AB6 (mm) 326.27 302.27 282.88 266.99 253.79
ωM 3AB6 (mm) 11.67 11.23 10.87 10.56 10.30
ROU T M 3AB6 (mm) -630.44 -744.66 -859.96 -1104.17 -1406.55
KF M 3AB6 0.99966 0.99969 0.99971 0.99972 0.99974
KCO M 3AB6 0.99932 0.99937 0.99941 0.99944 0.99947
ω0OU T M 3AB6 (mm) 10.53 10.42 10.30 10.19 10.06
dOU T −M 3AB6 (mm) 116.84 103.92 90.64 77.08 63.33
dM 3AB6−M 2A (mm) 1385
fM 2A (mm) 857.4
θM 2A (deg) 45
RIN M 2A (mm) 1315.48 1333.06 1350.76 1368.45 1386.03
ωM 2A (mm) 55.52 52.76 50.43 48.43 46.71
ROU T M 2A (mm) -2462.21 -2402.91 -2347.47 -2295.88 -2248.03
KF M 2A 0.99948 0.99953 0.99957 0.99960 0.99963
KCO M 2A 0.99895 0.99905 0.99914 0.99920 0.99926
ω0OU T M 2A (mm) 18.88 18.01 17.20 16.44 15.74
dOU T −M 2A (mm) 2177.56 2123.05 2074.49 2031.27 1992.75
dM 2A−M 1A (mm) 3800
fM 1A (mm) 590
θM 1A (deg) 45
RIN M 1A (mm) 2004.48 2031.26 2053.69 2072.63 2088.73
ωM 1A (mm) 43.24 43.12 43.02 42.94 42.87
ROU T −M 1A (mm) -836.10 -831.52 -827.83 -824.79 -822.26
KF M 1A 0.99933 0.99933 0.99934 0.99934 0.99934
KCO M 1A 0.99866 0.99866 0.99867 0.99868 0.99868
ωCASS (mm) 8.58 7.97 7.45 6.99 6.58
dOU T −CASS (mm) 803.19 803.09 803.01 802.94 802.88
ωCASS IDEAL (mm) 8.51 7.91 7.39 6.93 6.53
KAXIALL 0.99994 0.99994 0.99994 0.99994 0.99994
dM 1A−SU BREF (mm) 6685.86
RIN SU BREF (mm) 5887.16 5886.65 5886.24 5885.90 5885.62
ωSU BREF V ERT EX (mm) 310.50 310.49 310.48 310.47 310.47
ωSU BREF RIM (mm) 316.37 316.36 316.35 316.34 316.33
KT OT AL 0.99535 0.99558 0.99577 0.99593 0.98606
Te SU BREF RIM (dB) 12.20 12.21 12.21 12.21 12.21
ηAP ERT U RE (%) 79.89 79.89 79.89 79.89 79.89

40
 (a) (b)

Fig 28 Conic sections: (a) Ellipsoid, (b) Hyperboloid.

Table 14 Geometrical parameters of mirrors for LT in Cab-A configuration. Ell.: Ellipsoid , Hyp.: Hyperboloid.
MIRROR M1A
Freq. Rin Rout f aθi A0 b θ ωm Diam.
e Mirror
(mm) (mm) (mm) (mm) (deg)
(mm) (mm) (mm) (deg) (mm) (mm)
452 809.95 2172.62 590.00 45
1491.29 2318.69 0.777411 938.01 110.45 42.88 277x178 Ell.
MIRROR M2A
Freq. Rin Rout f θi a A0 b θ ωm Diam.
e Mirror
(mm) (mm) (mm) (mm) (deg) (mm) (mm) (mm) (deg) (mm) (mm)
452 1903.26 1560.30 857.40 45 1731.78 2461.08 0.710565 1218.54 140.66 37.18 329x223 Ell.
MIRROR M3A
Freq. Rin Rout f θi a A0 b θ ωm Diam.
e Mirror
(mm) (mm) (mm) (mm) (deg) (mm) (mm) (mm) (deg) (mm) (mm)
243 939.62 278.79 215.00 43.81 609.21 968.95 0.795253 369.35 163.29 10.87 68x47 Ell.
324 6426.44 222.44 215.00 44.515 3324.44 6426.52 0.966557 852.56 178.02 9.71 60x42 Ell.
661 -755.76 167.38 215.00 44.515 294.19 776.84 1.320299 253.62 76.59 8.42 50x35 Hyp.

Code and Data Availability

The data utilized in this study were obtained by LLAMA Collaboration. Data are available from
the authors upon request and with permission from LLAMA Collaboration.

Acknowledgments
Emiliano Rasztocky extends his heartfelt gratitude to the staff of the Instituto Argentino de Ra-
dioastronomı́a for their invaluable support, which was fundamental in facilitating the publication
of this work.
Rodrigo Reeves acknowledges support from ANID CATA BASAL FB210003.

References
1 E. Arnal, Z. Abraham, C. Cappa, et al., “Llama: A new mm and submm observing facility,”
in Revista Mexicana de Astronomia y Astrofisica Conference Series, 49, 53–53 (2017).
2 LLAMA web page, “https://www.llamaobservatory.org/en/,” (10 de August
de 2023).

41
 3 G. E. Romero, “Large latin american millimeter array,” arXiv preprint arXiv:2010.00738
(2020). [doi: https://doi.org/10.52712/sciencereviews.v1i4.10].
4 J. R. Lepine, Z. Abraham, C. G. G. CASTRO, et al., “The llama brazilian-argentinian radiote-
lescope project: progress in brazil and brics collaboration,” Anais da Academia Brasileira de
Ciências 93 (2021). [doi: 10.1590/0001-3765202120200846].
5 M. Fernandez-Lopez, P. Benaglia, S. Cichowolski, et al., “Llama millimeter and submil-
limeter observatory. update on its science opportunities,” arXiv preprint arXiv:2312.12210
(2023). [doi:10.48550/arXiv.2312.12210].
6 R. Güsten, L. Nyman, P. Schilke, et al., “The atacama pathfinder experiment (apex)–a new
submillimeter facility for southern skies–,” Astronomy & Astrophysics 454(2), L13–L16
(2006). [doi:10.1051/0004-6361:20065420].
7 M. D. Johnson, K. Akiyama, L. Blackburn, et al., “Key science goals for the next-generation
event horizon telescope,” Galaxies 11(3), 61 (2023). [doi:10.48550/arXiv.2304.11188].
8 ALMA web page, “https://www.almaobservatory.org/en/home/,” (10 de Au-
gust de 2023).
9 V. A. GmbH, “Large Latin-American Millimetric Array Telescope LLAMA – Antenna with
Nasmyth Focus – Technical proposal.,” tech. rep., Vertex Antennentechnik GmbH, VA Proj-
No.: 21/09087 (Duisburg, 2014).
10 A. Wootten, L. Snyder, E. van Dishoeck, et al., “Frequency band considerations and recom-
mendations,” NRAO, ALMA Millimeter Array Memo Series (213) (2000).
11 R. Bachiller, G. Blake, R. Booth, et al., “Report of the alma scientific advisory committee:
September 2000 meeting,” (2000).
12 M. Carter et al., “Alma front-end optics design report,” available from ALMA project docu-
mentation server, Tech. Rep. FEND-40.02. 00.00-035-B-REP (2007).
13 M. C. Carter, A. Baryshev, M. Harman, et al., “Alma front-end optics,” in Ground-based
Telescopes, 5489, 1074–1084, SPIE (2004). [doi:10.1117/12.552538].
14 J. Lamb, A. Baryshev, M. Carter, et al., “Alma memo 362,” (2001).
15 J. W. Lamb, “Low-noise, high-efficiency optics design for alma receivers,”
IEEE Transactions on antennas and propagation 51(8), 2035–2047 (2003).
[doi:10.1109/TAP.2003.814743].
16 F. Paul, “Goldsmith quasioptical systems: Gaussian beam quasioptical propogation and ap-
plications,” (1998).
17 O. Nyström, I. Lapkin, V. Desmaris, et al., “Optics design and verification for the apex
swedish heterodyne facility instrument (shefi),” Journal of Infrared, Millimeter, and Tera-
hertz Waves 30, 746–761 (2009). [doi:10.1007/s10762-009-9493-7].
18 A. Gonzalez, T. Soma, T. Shiino, et al., “Optics characterization of a 900-ghz heb receiver
for the aste telescope: design, measurement and tolerance analysis,” Journal of Infrared,
Millimeter, and Terahertz Waves 35, 743–758 (2014). [doi:10.1007/s10762-014-0074-z].
19 M. Rioja and R. Dodson, “Precise radio astrometry and new developments for the next-
generation of instruments. aapr 2020, 28, 6,” arXiv preprint astro-ph.IM/2010.02156 .
[doi:10.1109/ICEAA.2019.8879329].
20 M. Kotiranta, K. Jacob, H. Kim, et al., “Optical design and analysis of the submillimeter-
wave instrument on juice,” IEEE Transactions on Terahertz Science and Technology 8(6),
588–595 (2018).

42
21 A. Gonzalez, “Design of optics for alma receivers at naoj,” in 36th ESA Antenna Workshop,
Noordwijk, the Netherlands, (2015).
22 J. Murphy, “Distortion of a simple gaussian beam on reflection from off-axis ellipsoidal
mirrors,” International journal of infrared and millimeter waves 8, 1165–1187 (1987).
[doi:10.1007/BF01010819].
23 K. Pontoppidan, “Electromagnetic properties and optical analysis of the alma antennas and
front ends,” Nat. Radio Astron. Observatory, Charlottesville, VA, ALMA EDM Doc. FEND-
80.04. 00.00-026-A-REP (2008).
24 V. Tapia, Design and measurements of an optical system for ALMA band 1. PhD thesis
(2015).
25 R. Finger responsable del desarrollo de B2+3, Universidad de Chile, “Comunicaciones per-
sonales,” (2022).
26 A. Gonzalez, “Frequency independent design of quasi-optical systems,” Journal of Infrared,
Millimeter, and Terahertz Waves 37(2), 147–159 (2016). [doi:10.1007/s10762-015-0205-1].
27 TICRA web page, “https://www.ticra.com/,” (10 de October de 2023).
28 J. F. Johansson, “A comparison of some feed types,” in Multi-Feed Systems for Radio Tele-
scopes, 75, 82–89 (1995).
29 M. Candotti, “Design of the alma band 10 optics with horn-to-horn local oscillator signal
power injection scheme,” tech. rep., ALMA. [doi:10.13140/RG.2.1.3579.9121].
30 ALMA, ALMA Cycle 7 Technical Handbook, https://almascience.nrao.edu/news/alma-cycle-
7-science-observations-status-update-3 (2019).
31 D. Schwan, P. A. Ade, K. Basu, et al., “Invited article: Millimeter-wave bolometer array re-
ceiver for the atacama pathfinder experiment sunyaev-zel’dovich (apex-sz) instrument,” Re-
view of Scientific Instruments 82(9) (2011). [doi:10.1063/1.3637460].
32 J. D. Kraus, Radio Astronomy, Cygnus-Quasar Books, 2nd ed. (1986).
33 ALMA web page, “https://almascience.eso.org/proposing/
sensitivity-calculator,” (12 de June de 2024).
34 M. Holdaway, “Alma memo 402 illumination taper misalignment and its calibration,” (2001).
35 B. L. Ulich, “Pointing characteristics of the 36-foot telescope,” tech. rep., National Radio
Astronomy Observatory Green Bank 1 West Virginia (1976).
36 T. Cwik and V. Jamnejad, “Beam squint due to circular polarization in a beam-waveguide
antenna,” Telecommunications and Data Acquisition Progress Report 128, 1–10 (1996).
37 C. T. Cunningham, G. H. Tan, H. Rudolf, et al., “Front-end sub-system for the 12 m-antenna
array technical specifications alma-40.00.00.00-001-a-spe,” (2007).
38 P. Prescott, “Monte carlo methods,” (1965).
39 A. E. Siegman, Lasers, University science books (1986).
40 B. Lazareff and S. Sakamoto, “Alignment tolerances for alma optics,” ALMA Memo 395
(2001).
41 R. A. Rigby, M. D. Stasinopoulos, G. Z. Heller, et al., Distributions for modeling location,
scale, and shape: Using GAMLSS in R, CRC press (2019). [doi:10.1201/9780429298547].
42 S. Withington, J. A. Murphy, A. Egan, et al., “On the design of broadband quasioptical
systems for submillimeter-wave radio-astronomy receivers,” International journal of infrared
and millimeter waves 13, 1515–1537 (1992). [doi:10.1007/BF01009233].

43
 Biographies of authors

Emiliano Rasztocky Graduated in Mechanical Engineering in 2002 from the Universidad Na-
cional de La Plata (UNLP). AIV and design engineer for the development of the tertiary optical
system for the LLAMA project. Since 2021, system engineer for the QUBIC project. From 2014
to 2024, Chief of the Mechanical Department at the Instituto Argentino de Radioastronomı́a (IAR).
Since 2024, design engineer for the opto-mechanical subsystems for PMI project at the Max Planck
Institute for Solar System Research (MPS).

Matı́as Rolf Hampel Received his Ph.D. in Engineering from Universidad Tecnológica Nacional
(UTN), Argentina, in 2018. He was an international excellence fellow at the Karlsruhe Institute of
Technology (KIT), Germany. He is a researcher at CONICET and CNEA, Argentina, and leads the
Detector Design and Construction Department at GAIDI-CNEA. He is a member of the Adviser
Committee of ITeDA, the Pierre Auger Collaboration, and the QUBIC Collaboration.

Rodrigo Reeves Received a Ph.D. in electrical engineering from Universidad de Concepcion

(UdeC), Chile, in 2009. Served as Chief Engineer at the Chajnantor Observatory from 2005
to 2009, overseeing CBI, CBI2, and QUIET projects. Postdoc in astronomical instrumentation
at Caltech, later a Senior Scientist there from 2009 to 2014. Currently an Associate Profes-
sor at UdeC, also directing CePIA, the Center for Astronomical Instrumentation. Specializes in
cryogenic coherent receivers, microwave array receivers, and sub-millimeter instrumentation with
radio-astronomy applications.

Jacques R.D. Lepine Graduated in Physics (1967), Master in Nuclear Physics (1970) at Physics
Institute of University of São Paulo. PhD in Astronomy at University of Paris VII (1977). Full Pro-
fessor at Institute of Astronomy, Geophysics and Atmospheric Sciences (IAG-USP). Was president
of the Brazilian Astronomical Society(1988-1989), Chair of Astronomy Department (1994-1988),
Director of IAG-USP(2001-2005) , member of the CNPQ Adviser Committee for Physics and
Astronomy. Member of Brazilian Academy of Sciences, and World Academy of Sciences.

Gustavo Esteban Romero Received his Ph.D. in Physics from the Universidad Nacional de La
Plata (UNLP). He was a postdoctoral fellow at the University of Sao Paulo and a research associate
at the Max Planck Institute for Nuclear Physics (Heidelberg, Germany). Currently, he is a Senior
Researcher at CONICET, Director of the Instituto Argentino de Radioastronomı́a (IAR), and a Full
Professor at UNLP. He has lectured globally and supervised around 40 doctoral and undergraduate
theses.

44