Building A Large Affordable Optical-NIR Telescope

Building a large affordable Optical-NIR telescope (I):
An alternate way to handle segmented primary

mirror
Radhika Dharmadhikari (  radhika.dhar@iiap.res.in )
Indian Institute of Astrophysics
Padmakar Parihar
Indian Institute of Astrophysics
Annu Jacob
Inter-University Centre for Astronomy and Astrophysics
Research Article
Keywords: Segmented Mirror, Alignment, Phasing, Optical Telescopes, Simulations
Posted Date: May 16th, 2023
DOI: https://doi.org/10.21203/rs.3.rs-2060778/v1
License:   This work is licensed under a Creative Commons Attribution 4.0 International License.
Read Full License
Additional Declarations: No competing interests reported.

Springer Nature 2021 LATEX template
Building a large affordable Optical-NIR telescope (I): An

alternate way to handle segmented primary mirror
Radhika Dharmadhikari1*, Padmakar Parihar1† and Annu Jacob2†
1* Indian Institute of Astrophysics, II Block Kormangala, Bangalore, 560034, India.
2Inter-University Centre for Astronomy and Astrophysics, Pune, 411007, India.
*Corresponding author(s). E-mail(s): radhika.dhar@iiap.res.in;

Contributing authors: psp@iiap.res.in; annu@iucaa.in;
† These authors contributed equally to this work.
Abstract
The use of innovative ideas and the latest technology have undoubtedly brought down telescope
costs substantially. However, there are still ways to further reduce the cost of optical ground-based
telescopes and make them affordable to much larger and wide spread astronomical communities.
In this and subsequent papers we are presenting our studies carried out towards building afford-
able mid-size telescopes of 4.0-6.0m in size. In the present era, segmented mirror technology has
become the first choice for building moderate to large-size telescopes. In any Segmented Mirror
Telescope (SMT) the most important part is its primary mirror control system (M1CS). The con-
ventional M1CS is based on edge sensors and actuators, but such a system introduces many design
and implementation complexities. In this paper, we propose to make use of an Off-axis Alignment
and Phasing System (OAPS), which is an active mirror kind of control system working in real time
to maintain the figure of a segmented primary mirror without the use of edge-sensors. The align-
ment and phasing system which is an integral part of any segmented telescope can be used in the
real time at the off-axis. Through extensive simulations we have explored the feasibility of using
an OAPS for co-alignment, co-focusing as well as co-phasing of segmented mirror telescopes. From
our simulations we find that the co-alignment and co-focusing of the segments can be achieved
with a guide star as faint as 16-18th magnitude. This implies that seeing limited performance for
any segmented telescope can be easily accomplished without use of a complex edge sensor based
control system. Whereas, to attain diffraction limited performance, mirror segments need to be co-
phased with an accuracy of few tens of nanometers. In our simulations we have used a dispersed
fringe sensor based phasing scheme, which can effectively work up to guide stars of 14th magnitude.
Keywords: Segmented Mirror, Alignment, Phasing, Optical Telescopes, Simulations
1 Introduction mega optical-IR telescopes of 25.0-40.0m aper-

tures. A telescope of this size not only requires
Quest to explore faintest cosmic objects with high- unprecedented expense but also collective efforts
est possible spatial resolution has always been of several hundreds of scientists and engineers,
a driver to build larger and better telescopes. consistently working over more than one or two
World at present is seriously working to realize decades. A project of this scale is not possible by
1
2 Alternate way to handle segmented primary mirror
single institution and/or country and hence multi- to take a challenge and build telescope of their
institutions spread over the globe are involved. own choice. The classical telescopes built during
Undoubtedly, these future mega telescopes, com- 50-80s are found to be extremely bulky, having
bined with 8.0-10.0m class telescopes, will play very large dimensions and they usually follow the
important role in frontier research in astronomy, cost law in which price increases as 3rd power
however, importance of moderate size telescopes of the diameter (Schmidt-Kaler and Rucks, 1997;
of 4.0-6.0m can not be understated. Giving access van Belle et al, 2004). Whereas, telescopes built
to 4.0-6.0m class telescopes to a large community after 80s gained from many technological advances
would be very important for many science pro- and became lighter as well as compact. This led
grams such as transient follow-up observations, to reduction in the telescope cost and the scal-
time series studies of relatively faint objects etc. ing law became less steep i.e. power ranges from
Telescopes of this size can also serve as instrumen- 2.7-2.0 (van Belle et al, 2004; Stahl and Henrichs,
tation test bed, as well as human resource devel- 2016). By employing innovative ideas as well as
opments to larger facilities (Chanover et al, 2019). uses of latest technology, it is quite possible to
While exploring the productivity as well as impact make the ground based optical-NIR telescope of
of different size telescopes Abt (2012) finds that decent size (4.0m and above) affordable to many
in terms of average number of citations, telescopes institutions/countries.
with an aperture larger than 7.0m are marginally When 100-150 million dollars were spent to
better (30%) than those with an aperture of 2.0- get 8.0-10.0m size optical telescopes at that time
4.0m. Whereas, the construction and operation world’s two most inexpensive telescopes (Hobby
cost of 8.0-10.0m class telescopes are found to be Eberly Telescope (HET) and South African Large
4-5 times more than 4.0m class telescope. In terms Telescope (SALT)) were conceived and realized. It
of impactful productivity against telescope con- may be argued that HET/SALT are special pur-
struction and operation cost, 4.0m class telescopes pose telescopes and number of compromises such
are much more cost effective (Sagar, 2000). as image quality, sky coverage and the focal sta-
Many institutions/countries, which have been tion were made. However, we believe that all these
so far managing their observing needs with small limitations were accepted to bring down the cost
1.0-2.0m size telescopes, are now aiming to have to its lowest. By spending little more and going
four to six meter class telescope. Many four to with alternate optical configuration it is quite pos-
six meter class telescopes such as DAG, INO, sible to overcome all these limitations and come up
NRT, SEIMEI, TAO are at different stages of with a telescope which will be, by no means, infe-
realizations (Yeşilyaprak and Keskin, 2020; Khos- rior to any other good telescopes. When it comes
roshahi et al, 2016; Gutiérrez et al, 2021; Nagata to building large telescopes, the use of segmented
and Kurita, 2020; Doi et al, 2018). Telescopes primary mirrors has now become a natural choice.
larger than 4.0m size have been solely designed It has got many manufacturing as well as oper-
and built by proposing institutions with strong ational advantages over the same size monolithic
backing from industries. And there are many tele- mirror (Nelson et al, 2013). Though a serious effort
scope manufactures for the telescope apertures to make a segmented telescope started with the
up-to 2.0m. Whereas, for the mid size telescopes MMT project, the true breakthrough happened
there are only few telescope manufactures which in the segmented mirror technology when the
can provide 3.0-4.0m size telescope completely as Keck telescopes were designed and built in 1980-
a turn-key (Ninane et al, 2012; Marchiori et al, 95 (Nelson et al, 1985). Subsequently, by making
2012). The telescope design built by these private use of spherical optics, HET and the SALT were
entities are largely inspired by ESO’s VLT design designed and built by spending almost one tenth of
and are found to be complex, bulky as well as the cost of similar size telescope, such as Keck and
costly. Undoubtedly, telescope is a complex system the GTC (Buckley et al, 2004; Smith, 1989). Seg-
and for any institution with no in-house capa- mented mirror technology is making it possible to
bility, it is almost impossible to build their own realize world’s largest telescopes TMT and E-ELT
telescope. However, for many aspiring institutions having 30.0m and 39.0m size primary respec-
which have got some heritage in telescope build- tively. National Large Optical Telescope (NLOT)
ing and maintaining, it would be advantageous
Alternate way to handle segmented primary mirror 3
and Chinese Large Optic-Telescope (CLOT) are 4.0-6.0m class telescopes by about an order. In
another 10.0m class optical-near infrared tele- this and subsequent papers we are presenting the
scopes proposed in India as well as in China outcome of our extensive design and simulation
(Anupama et al, 2022; Cui and Zhu, 2016). Mono- efforts aimed at achieving this goal. We propose
lithic primary mirrors of 4.0-6.0m size are easily to make use of segmented mirror technology and
manufacturable, however, looking at cost effective- instead of using edge sensor based complex M1CS
ness, many other moderate size telescopes have for maintaining segment alignment, we plan to
also opted to go with segmented primary. Recently implement an OAPS similar to the active optics
completed Japanese 3.8m SEIMEI telescope and mechanism used to maintain the figure of an
the upcoming New Robotic Telescope (NRT) are active primary mirror. In the section 2 we describe
two examples in this regard (Nagata and Kurita, briefly, the telescope optics considered for the
2020; Gutiérrez et al, 2021). study as well as results of a comparative analysis
For any segmented mirror telescope, the align- between hexagonal and circular segment geome-
ment (co-aligning, co-focusing and co-phasing) of try. Whereas, in section 3 we explain the proposed
the segments is crucial to achieve the desired scheme as well as discuss the requirements for it.
image performance. Initially, the alignment of the In section 4 we discuss about the simulation tool
segments is carried out optically with the aid of an developed to generate the realistic alignment and
Alignment and Phasing System (APS). After the phasing related images and its validation using
segments are aligned, the primary mirror control real observations. Results related to co-aligning
system (M1CS) comprising of a large number of and co-focusing using the real time OAPS are
actuators and edge sensors, maintains the align- given in the section 5. Simulation results related
ment over several weeks to about a month under to the co-phasing of segments using the DFS tech-
varying environmental conditions. In the whole nique are presented in section 6. Section 7 and
M1CS, edge sensors are the most vital elements. section 8 respectively present the discussions and
Edge sensors are expected to sense any segment conclusions derived on the basis of results obtained
movements in multiple dimensions with very high through the rigorous simulations.
accuracy over an extended range and in addi-
tion to that, it must have high temporal stability 2 The Telescope Optics
(Minor et al, 1990; Shelton et al, 2008; Rozière
et al, 2008; Gajjar et al, 2016). Except Keck and 2.1 Design of the telescope optics
GTC (replica of Keck), all other telescopes (HET,
SALT and LAMOST) based on segmented mir- As a case study we have chosen a 4.0m class tele-
ror technology have gone through troublesome scope made of segmented primary mirror. While
period in maintaining the alignment. Even after designing the telescope optics we deliberately
commissioning, over many years, alignment of decided, not to consider aspheric primary, because
these telescopes could not be maintained, which manufacturing off-axis aspheric mirror segments
has adversely affected telescope performance and requires specialized tools and techniques, which
hence their scientific productivity (Rakoczy et al, makes it very costly. For our simulations, we con-
2003; Gajjar et al, 2006). Another challenge with sider a four mirror design, that uses three spherical
edge sensor based M1CS is the focus mode prob- mirrors and one higher order aspheric for aberra-
lem. The focus mode problem arises when all tion corrections (Figure 1). The primary mirror
mirror segments resting on metallic mirror cell go is 4.3m in diameter with 18 segments (two rings)
through uniform changes in the tilt for which edge of 1.0m size. The preliminary design of the tele-
sensors are either blind or not having required tilt scope has a f-number of 10.75 with a plate scale of
sensitivity (Oswalt and McLean, 2013). 4.42arc-sec/mm at the focal plane, other parame-
We believe that by opting number of cost ters for the design are given in Table 3. This design
effective solutions such as use of spherical pri- needs to be further optimised before finalisation.
mary mirror optics, segmented mirror technology, In the second part of this series of papers “Build-
handling mirror segments without edge sensors, ing a large affordable Optical-NIR Telescope” we
compact and light weight telescope structure etc., aim to present the telescope optical design in
possibly we can further bring down the cost of detail.
compared the image qualities for hexagonal and

circular segments in our telescope design at dif-
ferent wavelengths. Figure 2 shows the hexagonal
and circular segment geometry for a 4.3m, two ring
segmented primary mirror.
Fig. 2 Two ring Segmented primary mirror with diame-

ter=4.3m and segment size=1m, left: hexagonal segments,
right: circular segments (Dashed red circle marks the area
obscured by the secondary)
In order to understand the effect of segment

shape on the telescope image quality, we con-
sider two cases one with telescope primary mirror
made of hexagonal and the other with circular
segments. For these two cases, we obtain the D50
Fig. 1 Optical design layout for the 4.3m four mirror and D80 values for different wavelengths within
telescope the visible-IR spectral window. Figure 3 shows the
variation of D50 values with wavelength for all
2.2 Choice of Segment Shape the cases. On comparing results for hexagonal and
circular segment geometry, we find that as com-
The mirror segments can be hexagonal, circular pared to hexagonal case, the D50 and D80 values
or petal in the shape. Most of the existing SMTs for circular segments are found to be larger by
make use of hexagonal segments due to their ≈12% and ≈30% respectively, at all wavelengths.
higher filling factor. As per established practice Nearly matching D50 but comparatively larger
first roundel mirror blanks are polished and fig- D80 for the circular segments indicates the spread
ured to the required specifications using different of energy in the wings of the point spread func-
sophisticated techniques, such as, Stress Mirror tion (PSF). The D80 over optical band is found
Polishing (Lubliner and Nelson, 1980) and/or to be within 0.07-0.15 arc-second, which may be
complex CNC based polishing machine (Walker acceptable to many telescopes that aim to work
et al, 2006). Then, the segments are cut into in seeing limited regime. From our analysis we
hexagonal shape by maintaining its surface figure find that as far as D50 is concerned, hexagonal
which is the most critical task that leads to signif- segments give marginally better image quality as
icant increase in the segment manufacturing cost. compared to circular segments, moreover, use of
There are a few telescopes such as Multiple Mirror circular segments would reduce the manufacturing
Telescope (MMT) and upcoming Giant Magellan cost, time and complexity to a large extent. Thus,
Telescope (GMT) which uses circular segments to by replacing the hexagonal segments with circu-
form the primary mirror. Since circular segment lar, the telescope design can be made much more
introduces uneven gaps which, not only leads to cost effective.
some light loss, but may also affect the image
quality when diffraction limited imaging is aimed,
therefore, using ZEMAX ray tracing software we
D50 for Hexagonal Segments the piston error) between any two adjacent seg-
0.16 D50 for Circular Segments ments. The co-alignment and co-focusing of the
0.14 segments can be easily handled by the use of a
Shack-Hartmann (SH) sensor. Whereas, for the
Encircled Energy in arcsec
0.12
co-phasing, multiple techniques are available, like
0.10 the Keck Narrow band and Broad band phas-
0.08 ing, interferometric phasing, pyramid sensor based
phasing etc. For our simulations we have consid-
0.06
ered the Dispersed Fringe Sensor (DFS) based
0.04 phasing technique, because it has got large capture
range and can have high measurement accuracy.
0.02
The details of the OAPS and its functionality
500 1000 1500 2000 2500
Wavelength in nm are further explained in section 5 and section 6
respectively.
Fig. 3 Plots showing variation of the D50 over optical and
NIR wavelengths for different segment shapes
3.1 Top level requirements for
3 Maintaining shape of a segment alignment and phasing
segmented primary mirror Allocating the error budget for the segment align-
ment is very crucial as that would decide the final
through optical feedback telescope image quality. We have attempted to
In active mirror telescopes the shape of the rela- come up with the requirements by making use of
tively thin primary mirror is maintained by large simulations as well as by referring to the prac-
number of actuators through optical feedback tically achieved values in the existing segmented
from an off-axis wavefront sensor. Any deforma- mirror telescopes. We use a python ZEMAX inter-
tion induced in the primary mirror, mainly due facing based tool to find alignment requirements
to gravity/temperature changes, is sensed by the for the four mirror telescope of 4.0m size (with
wavefront sensor working on real time and is cor- optical parameters given in the Table 3). In order
rected through use of actuators. The active optics to give the alignment errors in ZEMAX, we use a
system was first implemented on the New Tech- specially developed segmentation tool (Jacob et al,
nology Telescope (NTT) and now it has become 2020).This tool generates a grid sag surface with
an established procedure. Almost all modern mid random tip-tilt and/or piston errors. The grid sag
to large size telescopes utilize active optics tech- surface is then imported to ZEMAX where the
nology to get the best telescope performances image quality parameters such as the Strehl ratio,
(Roddier et al, 1995; Schechter et al, 2003; Neufeld D50 and D80 are derived. All the image quality
et al, 2004; Lousberg et al, 2016; Stephan et al, parameters obtained from the simulations are at
2016). For maintaining the alignment of a SMT, a wavelength of 584.95nm. The wavefront maps
we propose a similar methodology which will make corresponding to the pupil phase for the tele-
use of a dedicated real-time Off-Axis Alignment scope with different alignment errors are shown in
and Phasing System (OAPS). Here, by main- Figure 4.
taining the alignment of the segments, we mean, Figure 5 shows the variation of the image
the co-alignment, co-focusing and co-phasing of D50 and D80 values in presence of tip-tilt align-
the segments. The co-alignment of the segments, ment errors and Figure 6 shows the variation of
deals with the removal of their tip-tilt errors so the image D50 and D80 values with increasing
that the PSFs from each segment falls on the phase/piston errors. Thus from the plots, it is
same point. Co-focusing deals with the Radius clear that under seeing limited condition, a tip-
of Curvature (ROC) adjustment of each of the tilt error up to 0.04arc-sec (it would degrade the
segments, to make all the PSF sizes equal and co- seeing limited FWHM by less than 0.5%) is allow-
phasing is the removal of the height difference (or able. Whereas, under diffraction limited condition
we restrict the Strehl ratio to be >0.9 for which
the RMS tip-tilt error must be <0.02arc-sec. In
Fig. 4 Wavefront map corresponding to the pupil phase for ideal case, RMS tip-tilt error of 0.04 arc-sec, RMS defocus
error of 385 micron and RMS piston error of 100nm
case of piston errors, the D80 is strongly affected D50

as compared to D50. Both the D50 and D80 values 0.5 D80
degrade rapidly till piston error of about 0.3µm

(λ/2) after that it becomes nearly constant up to 0.4
Encircled Energy in Arcsec

a piston error of 25µm (Figure 25). Hence, for the
seeing limited case, the piston error can be toler- 0.3
ated up to 25 to 30µm. Beyond this, the piston
error falls under the domain of the focus error 0.2
which can be measured using SH mode of OAPS
and is corrected with the help of segment support 0.1
actuators. Under seeing limited condition, phas-
ing of mirror segments is not required, but for
0.0 0.2 0.4 0.6 0.8 1.0
diffraction limited imaging, since we restrict the RMS Piston Error in micron
Strehl ratio to be >0.9 hence the segments must
Fig. 6 Variation of D50 and D80 values with RMS piston
be phased with an accuracy of 40.0nm or better. errors
D50
not needed. Whereas, for the Keck telescopes,
2.00 D80 which are designed to work in the diffraction
1.75 limited mode, the tip-tilt error must be within
1.50
0.014arc-sec (Chanan et al, 1986). Keck segments
Encircled Energy in Arcsec
are phased using the device called Phasing Cam-

1.25
era System (PCS) which is capable of measuring
1.00 piston errors down to 5.0-6.0nm in narrow band
0.75 mode (Chanan et al, 2000a). From the simula-
tions, as well as by studying few of the above
0.50
mentioned cases, we have an estimate that the
0.25
tip-tilt error requirement for our system would be
0.00 approximately 0.04arc-sec for seeing limited case
0.0 0.1 0.2 0.3 0.4 0.5
RMS Tilt Error in Arcsec and close to 0.02arc-sec for the diffraction limited
case. The co-focusing requirement for the piston
Fig. 5 Variation of D50 and D80 values with RMS tip-tilt errors are of the order of 20 microns for the seeing
error
limited imaging. The diffraction limited imaging
is only possible with the warping harness and it
The HET has an alignment requirement of requires very small ROC (defocus) and other high
0.065arc-sec and requires a piston accuracy within orders segment figure errors. For co-phasing of the
25.0µm (Wolf et al, 2003). Similarly for SALT segments, under the seeing limited case, the pis-
the segment tilt error after fine alignment must ton error can be of the order of 20.0-25.0µm and
be within 0.04arc-sec with a piston requirement for diffraction limited case it would be of the order
of 20.0µm (Wirth et al, 2004). The point to be of a few tens of a nanometers. The typical require-
noted here is that, both HET and SALT tele- ment of alignment and phasing for any segmented
scopes are aimed to work in seeing limited domain
and hence precise phasing of mirror segments is
mirror telescope for seeing and diffraction limited

imaging is given in the Table 1.
Table 1 Alignment requirements for SMTs
Alignment Error Seeing Diffraction

Limited mode Limited Mode Fig. 8 Sub aperture masks for 1 spot (≈100.0cm diam-
eter), 3 spots (43.0cm diameter) and 7 spots (30.0cm
Co-aligning ≤0.04 arc-sec ≤0.02 arc-sec diameter) per segment used for alignment and focusing
Co-focusing: Piston error ≤20.0µm ≤20.0µma
Co-focusing: ROC Error ≤ 40.0µmb ≤ λ/30c
Co-phasing Not required ≤40.0nm As in the case of active mirror telescopes, for
a Co-focusing the proposed real time OAPS also, the major
must reduce the piston error within the
capture range of the phasing system. requirement is the availability of suitable bright
b RMS ROC error achieved using warping harness. stars, within the off-axis field of view. We have
c RMS wavefront error due to ROC induced defocus calculated the average stellar density using the
error. “Besançon model of stellar population synthesis
of the Galaxy”1 . Figure 9 shows the logarithmic
variation of stellar density (cumulative number
of stars per square arc-minute) against stellar V
3.2 Off-axis APS and availability of magnitude. An annular field of 15.0 to 20.0arc-
bright stars minutes (about 100-150 square arc-minutes) is
The schematic of the proposed OAPS is shown normally available for wavefront sensing on the
in the Figure 7. At the telescope focal plane, currently working off-axis active alignment sys-
the on-axis light goes for science observations, tems on various telescopes. By making use of a
whereas very large annular off-axis field would movable pick-off mirror, a suitable bright star can
be available for the OAPS. The OAPS would be chosen from this annular field to feed its light
work in two different modes, namely, the align- to the OAPS device. From the Figure 9 we see that
ment and co-focusing mode and the phasing mode. for 100 arc-minutes field, we would have approx-
The off-axis light from the telescope focal plane, imately 120 stars of 13th magnitude or brighter
goes through a collimator, the collimated beam is in the galactic plane and at least 1 star towards
then further split into two, using a beam split- galactic pole. Star number density increases expo-
ter. One part of the beam enters the alignment nentially with the magnitude and hence we expect
and focusing section of the instrument, while the large number of fainter stars to be available in any
other part goes to the phasing instrument. For pointing direction for the OAPS.
the alignment and focusing mode, a SH kind of
setup with customized lenslet array is used. By 4 The Simulation Procedure
assigning 1, 3 or 7 lenslets per segment, we can and its Verification
measure the tilt error very precisely(Figure 8).
Whereas, for defocus measurement, we need to The real-time OAPS needs to perform three tasks,
have a minimum of 7 to 9 lenslets per segment to co-aligning, co-focusing and co-phasing of the seg-
get precise measurements. The phasing mode of ments. For the co-aligning and co-focusing we
the OAPS is described in detail in section 6. This consider the SH wavefront sensor while for phas-
off-axis system would continuously monitor the ing we use the Dispersed Fringe Sensor (DFS).
mis-alignment of the segments and provide feed- Since the SH wavefront sensor is nothing but
back to primary mirror control system, which in re-imaging of multiple sub-apertures of a seg-
turn will provide a direct movement instruction to ment using the collimator and lenslet array, hence
the segment support actuators. it resembles with a telescope working in imag-
ing mode. Whereas, using the DFS technique for
1
https://model.obs-besancon.fr/modele starcounts.php
Fig. 7 Schematic representation of the two modes of the Off-Axis APS
104
At Galactic North Pole
telescope + instrument optics, smearing of image
Towards Galactic Center due to atmospheric seeing, noises getting injected
Overall Avergae
102 in the image/spectrum from all different sources.
Number of stars per arcmin square
The parameters, such as the star magnitude,

exposure time, the sampling aperture diame-
100
ter, telescope focal length, telescope F-number,
re-imaging optics focal lengths, sky brightness
10−2
magnitude, object-moon separation, phase of the
moon, the atmospheric Fried parameter (ro ),
10−4 detector specifications (pixel size, quantum effi-
ciency, dark current, detector temperature, read
out noise) are given as input to the system. Using
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Stellar Magnitude these parameters, we first generate a synthetic
stellar image. The image blur due to dynamic
Fig. 9 Variation of stellar density per arc-min square with
stellar magnitude
atmosphere (seeing effect) is introduced by gener-
ating random and moving phase screen using the
python package AOtools (Townson et al, 2019).
phasing is more like dealing with stellar spectra. Finally we introduce varieties of noises such as
Therefore, to verify the results generated using photon noise, sky background noise and detector
the simulation tool on alignment and phasing, we noises. Since the atmospheric phase screen and all
first simulated images and spectra for the 2.0m the generated noises have a random nature, hence
Himalayan Chandra Telescope (HCT) and com- to get realistic values, we have used Monte Carlo
pared it with the real observations. After having Simulations.
success in reliable generation of synthetic tele-
scope images and spectra, we further proceeded to
the alignment and phasing related simulations.
4.2 Verification of simulations using
real observations
4.1 Simulation of telescope images Before using the simulation tool to study practi-
and spectra cability of OAPS for handling segmented primary
mirror, we verified it on real astronomical obser-
To test the feasibility of the proposed OAPS vations of 2.0m HCT equipped with Himalayan
scheme, the python code is developed in such Faint Object Spectrograph and Camera (HFOSC)
way that it generates realistic image of a tele- instrument of Indian Astronomical Observatory.
scope by incorporating all possible contributors, To verify the photometric simulations, we have
like light loss due to atmospheric extinction and
used V-band observations of M67 open stellar clus- y=x

ter, taken during three different nights with differ- Data for image 1
18 Data for image 2
ent atmospheric conditions. Using our python sim- Data for image 3
ulation tool, CCD images of 28 cluster members 16
Simulated Magnitude
having different magnitudes and colors are gener-
ated. The instrumental parameters used for the
14
simulations are listed in the Table 2. Other param-
eters such as the seeing, moon phase, object zenith
12
angle, exposure time are obtained from the image
files. The synthetic stellar fluxes/magnitudes very
10
well match with the observed magnitudes and
it nearly follows the relation y=x as shown in
10 12 14 16 18
the Figure 10. The difference between synthetic Observed Magnitude
and observed fluxes for all 28 stars of magnitude
Fig. 10 Plot showing the Observed stellar magnitude vs
range 10-18, observed in three nights are within simulated stellar magnitude
15% (except 2-3 little deviant points). Figure 11
Stellar Magnitude=16.68
shows the comparison of the real and simulated
images for a star of 16th magnitude. Further- 0
Real Observed Image
0
Simulated Image
more, we compared the radial profile of synthetic 5 5
and observed stellar images (Figure 12) which 10 10
remarkably matches in both x and y directions. 15 15
20 20
25 25
Table 2 Parameters for the HCT Observations 30 30
35 35
Parameter Value 40 40
0 10 20 30 40 0 10 20 30 40
Wavelength Range 470.0-700.0nm∗

Telescope Diameter 2.01m
Secondary mirror diameter 0.40m 50 75 100 125 150 50 75 100 125 150
Reflectivity of the mirror 90%

Number of Lens Elements 8
Transmittivity of lenses 95% at each element Fig. 11 Real Observed Image (left) and Simulated Image
Atmospheric Extinction∗ 0.12 airmass−1 (right) for stellar magnitude of 16.68 and exposure time of
Sky Brightness 21.28 magnitude/arc − sec2 5 seconds
Wind Speed 2.2m/s
Collimator Focal length 252.0mm
Camera Focal length 147.0mm atmospheric extinction, lens transmissivity, grism
Detector properties: response and detector response etc. Figure 13
Pixel size 15.0µm shows the observed spectra and the correspond-
Quantum Efficiency 85% @ 551.0nm
Dark Noise 0.3 e− per hour ing simulated spectra. One can clearly find that
Read out Noise 4.8 e− the continuum of the observed Fiege-34 spectrum
Gain 1.22e− /ADU very well matches with the simulated spectrum.
Image Scale 0.296arc-sec/pixel The departures seen at a few places are due to
Dispersive element Grism 8 with 600 lines/mm
the presence of absorption features, which are not
∗ V-band considered in the simulation. Hence, it can be con-
cluded that the simulations give realistic results
Similar to the photometric simulations, we which matches very well with the observations
also verify the spectroscopic simulations needed (both, photometric and spectroscopic) and thus
for DFS based phasing instrument. For this pur- confidently we use this tool to verify expected
pose spectra of a standard star Feige-34, observed performance of the proposed OAPS.
using the HFOSC instrument at HCT is used.
Few additional inputs required in the simulation
tool are wavelength dependent parameters such as

Real Image Intensity plot
Real Image Intensity plot
Table 3 Parameters for four mirror telescope design
40000 Simulated Image Intensity plot 100 Simulated Image Intensity plot
80 Parameter Value
30000
60
Telescope Diameter 4.330m
CCD counts
CCD counts
20000
40
Telescope F/number F/10.76
Primary Mirror ROC 17.480m
10000 20
Segment Diameter 1.0m
0
Plate Scale on telescope Focal Plane 4.428arc-sec/mm
0 Telescope Reflectivity 95% at each mirror
0 10 20 30 40 0 10 20 30 40
Pixels Pixels (4 mirror telescope)
Fig. 12 Intensity plots for simulated and real observed
images for two different stellar magnitudes (stellar magni-
tude=9.99 (left) and stellar magnitude=16.68 (right)) 5.1 Simulations on Co-Alignment
The lenslet array of SH device samples light
Simulated Spectra from different segments and generates an array of
Real Spectra spots on the detector plane. Therefore, using our
50000
simulation tool, we generate realistic SH images
40000 subjected to different practical conditions. All the
parameters used in the simulations are listed in
Detector Counts
30000 Table 4. The camera focal length is chosen such

that for 1.0arc-sec seeing the FWHM for the
20000 central wavelength covers 3 pixels approximately.
This ensures the critical sampling of the image
10000
required to attain accurate centering. The loca-
tion of the spots linked to segmented primary
0
mirror are compared with corresponding reference
5500 6000 6500 7000 7500 8000 8500 9000
Wavelength in Angstrom spots and the deviation of the spot locations are
used to measure the tip-tilt errors. The centroiding
Fig. 13 Plot showing the real observed spectra and the
simulated spectra for spectroscopic standard star Feige 34
error will translate directly to the alignment error,
hence we perform centroiding using the Photu-
tils package which is an Astropy affiliated python
5 Maintaining Co-alignment package for photometry (Bradley et al, 2021).
and Co-focusing of
Segments Table 4 Parameters used in the Monte Carlo Simulations
We plan to use an OAPS working in the real Parameter Value

time to maintain the figure of the segmented pri-
Wavelength Range 470.0-700.0nm (V-band)
mary of a 4.0m size telescope. The basic optical Sky Brightness 21.28 magnitude/arc − sec2
configuration of the OAPS is briefly described in Object-moon separation 45◦
subsection 2.1. One of the arms of OAPS is a Moon Phase Full Moon (ψ=0 ◦ )
standard SH, which uses lenslet array. By making Collimator Focal length 300.0mm
Lens Transmittivity∗ 0.75% per surface
use of our simulation tool we have attempted to Wind Speed 9.5m/s
explore the precision linked with measuring align- Detector:
ment errors (tip/tilt and defocus) under varying Pixel size 15.0µm
Quantum Efficiency 85%
conditions. The centroiding error of SH spots in
Temperature 243.0K
the presence of atmosphere (for various seeing con- Dark Noise 0.03 e− per pixel/second
ditions) and noises are derived for varying stellar Read out noise 2.3e− per pixel
magnitudes and aperture diameters. Detector gain 2
Camera focal length 58.840mm
Image scale on detector plane 0.339 arc-sec per pixel
Aperture Diameter 10.0-100.0cm
∗ assuming achromatic doublet lens

Stellar Magnitude=16, Exposure time=10s 0

Aperture=100cm, Single Spot 0
Aperture=43cm, Three Spots 0
Aperture=30cm, Seven Spots
Without Atmosphere 50 50 50
Seeing=1" 100 100 100
Seeing=1.5"
200
150 150 150
Seeing=2" 200 200 200
250 250 250
300 300 300
150 350 350 350
400 400 400

SNR
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
100
103 104 105 103 104 102 103 104
50
Fig. 15 Simulated SH spot images (Logarithmic Normal-
isation) for single spot (left), 3 spots (middle) and 7 spots
0 (right) per segment. The stellar magnitude is 10 with expo-
25 50 75 100 125 150 175 200 sure time of 10 sec for an atmospheric seeing of 1 arc-sec
Sampling Aperture Diameter in cm
Fig. 14 Variation of SNR with aperture diameter for 16th

magnitude star with 10.0s exposure time
extensive studies on effect of seeing in measur-
ing the segment tip-tilt errors (Simulations were
carried out for three different seeing conditions of
5.1.1 Effect of size and number of 1.0arc-sec, 1.5arc-sec and 2.0arc-sec). In all the
sub-apertures cases (single and multiple spots), tip-tilt error
Using the simulation tool we have first explored increases as seeing degrades. For the brighter stars
effect of the size of sub-aperture on Signal to Noise (10 to 13th magnitude), SNR is always very high
Ratio (SNR) of the spots. SNR is derived with and centroiding error remains nearly constant.
and without considering the atmospheric seeing After that the error increases rapidly when we
effect. Figure 14 shows the variation of SNR with move to fainter domain. It is observed that under
increasing aperture diameter for a 16th magnitude all seeing conditions a single largest sub-aperture
star. As expected, SNR linearly increases with gives the least error (Figure 16).
the sub-aperture diameter, which would improve The 0.04arc-sec tilt measurement requirement
the centroiding accuracy for the SH sensor. From can be achieved by using 17, 16 and 15th magni-
Figure 14 it is clear that sensitivity and hence tude stars at seeing of 1, 1.5 and 2.0 arc-seconds
SNR reduces as seeing effect is introduced. For respectively. Over the off-axis field one can always
long exposure we obtain the seeing limited image find stars brighter than 15th magnitude at any
and hence increasing the sampling aperture size telescope pointing. Whereas, our simulation shows
does not change the FWHM of the PSF but that one can even use stars as faint as 20th magni-
only increases the SNR which proves to be very tude on the event of excellent seeing (<1.0 arc-sec)
useful when working with the fainter stars. For condition. As far as the tilt error measurement is
each individual segment, we can have SH lenslet concerned, the largest aperture with single spot is
arrangement in such way that light from the seg- found to be the best.
ment can be sampled using variable number of In the presence of randomly distributed seg-
sub-apertures. In order to explore the effect of ment tilt errors, the telescope image obtained is
number of sub-apertures that one can fit into a a convolution of the seeing limited PSF and the
segment, we carried out simulations with three PSF due to alignment errors, hence the resultant
different cases i) Seven spots with aperture diam- FWHM can be estimated by quadrature sum-
eter = 30.0cm each, ii) Three spots with aperture ming of the FWHM values for these two sources
diameter = 43.0cm each and iii) Single spot with of image smearing effect. Therefore, to get qual-
aperture diameter = 100.0cm, which covers the itative measure of degradation due to segment
complete segment. Figure 15 shows the simulated alignment error, we have computed the percent-
SH images for the three cases. age change in the image quality (FWHM) for the
The Atmospheric seeing is a crucial parameter largest aperture (100.0cm) on three different see-
which can very much influence the performance ing and plotted it against the stellar magnitudes
of the OAPS. Therefore, we have carried out very (Figure 17). From the plot it is clear that up to
16.5 magnitude the percentage image degradation
Seeing=1" Exposure time=10s Seeing=1.5" Exposure time=10s Seeing=2" Exposure time=10s

100 Aperture=30cm(Seven Spots) Aperture=30cm(Seven Spots) Aperture=30cm(Seven Spots)
Aperture=43.30cm(Three Spots) Aperture=43.30cm(Three Spots) Aperture=43.30cm(Three Spots)
Aperture=100cm 100 Aperture=100cm 100 Aperture=100cm
Radial Centroiding Error in Arcsec

10−1
10−1
10−1
10−2
10−2
10−2
13 14 15 16 17 18 13 14 15 16 17 18 13 14 15 16 17 18
Stellar Magnitude Stellar Magnitude Stellar Magnitude
Fig. 16 Variation of tilt error with stellar magnitude for the exposure time of 10.0s with different number of apertures (30.0,
43.0 and 100.0cm diameters) and atmospheric seeing conditions. Dashed line indicates the tilt requirement of 0.04arc-sec
for the seeing limited case
will not be more than 1% for all the seeing con- of the segment focal point, thus having a simi-
ditions. Which means, alignment error will have lar effect on the image quality degradation. If the
no impact on image quality and it is still seeing, segment support uses warping harness then the
which will supersede. For 1.0arc-sec seeing, one ROC error can be easily corrected by adjusting
can use relatively fainter star (18th magnitude) the segment figures. However, in the absence of
for the tilt measurement. However, with poor see- warping mechanism, for the seeing limited tele-
ing like 1.5arc-sec or larger, one needs to use stars scope the large piston and ROC error can be
of 17th or brighter magnitudes to get the image corrected to some extent by differential movement
which is still dominated by the atmospheric see- of the segments (same as classical telescope focus-
ing and not by segment tilt errors. Here we have ing exercise). The co-focusing of segments also
considered the allowed image degradation by the uses the SH sensor. In this case, the SH spots
alignment errors to be about 10% of the seeing. show a radial shift with respect to the reference
spots, hence multiple spots per segment would be
Aperture=100cm Exposure Time=10s
required. In order to obtain the segment defocus
Seeing=1"
from the spot shifts, first the wavefront slopes
102 Seeing=1.5" are measured and then, through modal wavefront
Seeing=2"
Percentage Degradation in Image FWHM
reconstruction method, wavefront error is decom-

posed using the Zernike polynomials. The Zernike
101
coefficient Z20 corresponds to the defocus aber-
ration and this value is then converted to the
100
longitudinal defocus of the segment (Mahajan,
2013).
10−1 5.2.1 Simulations on Co-focusing

Similar to the segment alignment errors, the accu-
13 14 15 16 17 18
Stellar Magnitude racy of defocus measurement would depend on the
spot centroiding accuracy which in turn affects the
Fig. 17 Percentage degradation in image quality due to
tip-tilt alignment error for seeing limited case Zernike fitting. The source of centering error can
trace its roots to the poor signal, due to non avail-
ability of suitable bright stars, atmospheric seeing,
increased background as well as noisy detector.
5.2 Co-focusing with the OAPS All these sources of errors can individually and/or
collectively influence accuracy of centroid mea-
The co-focusing of the segments deals with two
surement. Therefore, instead of explicitly using
aspects, the segment piston error (large displace-
all these sources of errors in our defocus related
ment of mirror segment in the range of several tens
simulation studies, we have used their combined
of microns) and inherent segment ROC errors,
effect in the form of centering error which can be
both of which lead to longitudinal displacement
as poor as 1 pixel (15.0µm) maximum. In order and phasing would be given as a feedback to the
to estimate the accuracy of defocus measurement, primary mirror control system. Hence in addition
we consider an array of 3X3 spots per segment to alignment, the OAPS should be capable of pro-
and random centroiding error is introduced for viding precise measurements of very small piston
each of these nine spots. Simulated spot positions error (of the order of wavelength of the light),
are then compared with the reference spots and which is usually termed as a phase error.
aberrated wavefront is reconstructed. By fitting In our simulations we have attempted to check
Zernike polynomials the defocus values repre- the feasibility of DFS based segment phasing using
sented by Z20 coefficient are obtained for different a real time OAPS. The dispersed fringe technique
values of the centroiding errors (over the range of is primarily based on the principle of Rayleigh
1/50 - 1.0 pixel). For each given centroiding error, Interferometry. The Rayleigh Interferometry tech-
we have the derived values of Z20 coefficient which nique states that in presence of an Optical Path
are fed to the segmentation tool that generates a Difference (OPD) between two collimated beams,
grid sag surface for the corresponding RMS defo- the PSF shows a splitting and the positions of the
cus error. Finally the grid-sag surface is imported two peaks of the PSFs are a measure of the OPD
into our 4.0m telescope ZEMAX model to derive (Wang et al, 2009). The DFS works on the same
the image quality related parameters (D50, D80 principle, however to enhance the capture range, it
and Strehl ratio). In the Table 5 we summarize uses a broadband source and a dispersive element
our simulation results carried for the defocus. (such as a grating or grism)(Smith et al, 2003; Shi
From the Table 5 we observe that all three et al, 2004a).
image quality parameters remain unchanged up
to centroiding error of 1.5µm(1/10th of pixel). 6.1 The Simulation of DFS phasing
This indicates that the telescope performance scheme
remains diffraction limited despite of individual
segments being subjected to large defocus errors The phasing mode of the OAPS is shown in
of 172.52µm (equivalently piston error of 6µm). Figure 7, it consists of a sub aperture mask that
Further increase in the centroiding error gives samples light from the inter-segment regions. Each
marginal degradation in the image quality. For sub aperture mask is followed by a prism which
1 pixel = 15µm, which is the largest centroiding is responsible for deviating the light from each
error considered in our study, the defcous value of these inter-segment regions so that the fringes
is found to be 1.68mm which degrades the image are separated on the detector plane. After this
D50 by nearly 18%. We find that, in order to a dispersion element (grating/grism) is used to
limit the D50 degradation to be less than 1%, the introduce the wavelength dependent dispersion.
defocus error must be within 350.01µm which cor- Finally an imaging camera is used to form DFS
responds to a centroiding accuracy of 1/5th of a images on the detector plane. In the simulation
pixel. In the absence of warping-harness, for a see- platform, a sampling aperture shape is created
ing limited telescope, the defocus error of 350.0µm for the given inter-segment gap and the piston
can be corrected by simply moving the segment by error δ is applied as phase error = ±2k ∗ (δ/2)
nearly 12.1µm. If we consider the smallest measur- for each of the inter-segment regions. As a next
able centroiding error is 1/30th of the pixel, then step, the atmospheric effect is incorporated on
corresponding piston error measurement accuracy to the aperture by generating the moving phase
turns out to be nearly 2.0µm. screen. Further, the monochromatic PSFs are gen-
erated for each wavelength by taking the Fourier
Transform of this aperture. In order to generate
6 Phasing the Mirror dispersed fringe spectrum, wavelength dependent
Segments PSFs are shifted spatially along the dispersion
direction by following the grating dispersion law.
As explained in section 3, the plan is to make use
The generation of a monochromatic PSF is same
of an OAPS which will continuously monitor the
as explained in subsection 4.1. The parameters
alignment and phasing of the segments along with
such as the stellar magnitude, stellar spectral type,
the science observations. Any change in alignment
sampling aperture diameter, exposure time, Fried
Table 5 Variation of Image Quality Parameters with Segment Defocus
Centroiding Centroiding Error RMS defocus RMS Piston D50 D80 Strehl
Error Fractions of a Pixel Error Error (in arc-sec) (in arc-sec) Ratio
0.3 µm 1/50 33.80 µm 1.18 µm 0.035 0.094 0.998

0.37 µm 1/40 42.42 µm 1.48 µm 0.035 0.094 0.998
0.5 µm 1/30 56.13 µm 1.96 µm 0.035 0.094 0.998
0.75 µm 1/20 84.07 µm 2.93 µm 0.035 0.094 0.997
1.5 µm 1/10 172.52 µm 6.01 µm 0.035 0.094 0.996
1.67 µm 1/9 190.53 µm 6.64 µm 0.035 0.094 0.996
1.87 µm 1/8 217.36 µm 7.57 µm 0.035 0.095 0.995
2.14 µm 1/7 244.53 µm 8.52 µm 0.035 0.095 0.995
2.5 µm 1/6 288.63 µm 10.06 µm 0.035 0.095 0.994
3 µm 1/5 350.01 µm 12.20 µm 0.035 0.095 0.993
Parameter, telescope parameters (including the to measure the piston error accurately, the inten-
re-imaging optics), grating dispersion and detector sity modulations due to all the other factors need
specifications are given as input to the simula- to be nullified. This is achieved by dividing the
tion code. While generating the dispersed fringe DFS fringe spectra by a calibration spectra, which
spectrum, effect of chromatic variation of the is a continuum spectra obtained from the same
atmospheric extinction, grating efficiency and the system under zero piston condition. Practically, to
detector QE are also taken into account. Further, get the zero piston spectra, the sampling aperture
all kinds of noises such as photon noise, sky back- is placed at the center of the segment, instead of
ground noise, detector readout, PRNU and dark the inter-segment gap, hence it mimics the zero
noises are also added in to spectral images of DFS. piston condition (Figure 19).
Figure 18 shows a DFS image generated using the
simulation code for a piston error of 10.0µm. It
50
0 100 200 300 400 500 600
100 200 300 400 500 600

600
Detector Counts
400
200
Fig. 19 Placement of sub-aperture masks for DFS based
0
co-phasing. Aperture diameter of 10.0cm is used for sam-
5000 5500 6000 6500 7000 7500 8000 8500 9000 pling light from inter-segment edges and three samples are
Wavelength in Angstrom taken from segment centers with 30.0cm aperture diameter
to generate the zero piston spectra for reference
Fig. 18 Simulated DFS fringe image for star magnitude =
10 and exposure time = 240.0s with piston error of 10.0µm
is important to note that the DFS fringe spec- 6.1.1 Effect of aperture size and
trum is a combination of the source spectrum exposure time on fringe visibility
(which strongly depends on the stellar tempera- In the DFS scheme, the fringe visibility strongly
ture), atmospheric extinction coefficient, detector influences the precision of the piston measure-
QE, grating efficiency and the piston error. Thus, ment. The visibility of the fringes is given as
((Imax − Imin )/(Imax + Imin )), where Imax and Stellar Magnitude=13
Imin are the maximum and minimum fringe inten- Stellar Magnitude=14
0.6 Stellar Magnitude=15
sity values. The effect of orientation of the grism Stellar Magnitude=16
Stellar Magnitude=17
with respect to inter-segment region on the visi- 0.5
bility is a well studied problem (Shi et al, 2004a,b)
and should be taken care of while designing the
Visibility
0.4
DFS optics. Whereas, effect of aperture size and
exposure time on the visibility is poorly known. 0.3
In order to boost SNR it is advantageous to use
larger aperture. Therefore, we have attempted to 0.2
explore effect of aperture size on the fringe and

0.1
its visibility. Figure 20 shows the variation of
20 40 60 80 100
fringe visibility against the aperture size. From Aperture Diameter in cm
the figure it is quite apparent that increase in
Fig. 20 Variation of DFS fringe visibility with increas-
the aperture diameter leads to rapid decrease in ing aperture size, for different stellar magnitudes
the fringe visibility. The fringe visibility which (Piston=10µm, exposure time=60.0s)
was initially 60% for 10.0cm aperture and 15th
magnitude star, becomes nearly 10% when aper-
6.1.2 Effect of Atmospheric Seeing
ture is enlarged to almost full segment. Since for
DFS the fringe pattern is formed by systematic After confirming upon the basic properties affect-
merging of the primary and secondary peaks for ing the DFS fringe visibility, we concluded that
different wavelengths, thus when the aperture size the aperture size must be kept small (diameter
increases the secondary peak intensity is reduced, = 10.0cm for our case) and exposure needs to be
resulting in to visibility degradation. The opti- longer. Thus we carry out the Monte-Carlo simu-
mum sampling aperture of diameter 10.0cm can lations for varying stellar magnitudes to estimate
be better used as it maximizes the visibility as the obtainable value of SNR with the moderate
well as SNR. Another factor which has marginal seeing of 1.0arc-sec. All the parameters chosen for
effect on visibility is the exposure time. From our the phasing related simulations are given in the
exploration we found that shorter exposure (few Table 6. Figure 21 shows the plots for DFS fringes
milli-seconds) will have 15-20% better visibility along the dispersion direction for the central row
than longer exposure. Though using shorter expo- with varying stellar magnitude and exposure time.
sure is advantageous as far as fringe visibility is From the intensity plots we see that for larger SNR
concerned, however, extracting piston errors using the fringe modulation is easily visible but as the
relatively faint star is only possible when long SNR reduces for fainter stars, the fringe intensity
exposure is used. modulation gets dominated by the noise. For 15th
magnitude the fringes are barely distinguishable,
Table 6 Parameters used in the Monte Carlo whereas, for 16th magnitude star, noise supersedes
Simulations for co-phasing using the DFS technique the signal. The simulated fringe images are for
piston=10.0µm. As the piston error increases, the
Parameter Value visibility will further reduce and extraction of pis-
Wavelength Range 530.0-900.0nm ton values will be subjected to larger errors. Thus,
Aperture Diameter 10.0cm to achieve good SNR, the DFS based co-phasing
Camera focal length 75.0mm system must work with relatively brighter stars
Image scale on detector plane 0.268 arc-sec/ pixel
Grating groove density 300 lines/mm
(brighter than 14th magnitude) with an exposure
Linear Dispersion 6.667 A◦ / pixel time of 2 to 3 minutes. To estimate the piston mea-
Stellar temperature 3500K (K type star) surement accuracy we perform curve fitting on the
Seeing 1 arc-sec @ 500.0nm DFS fringe intensity pattern which is explained in
the next section.
Magnitude=13, Exposure time=60s Magnitude=14, Exposure time=60s Magnitude=15, Exposure time=60s Magnitude=16, Exposure time=60s
200
105 95
130
180 100 90
Detector counts in ADU

120
160 95 85
110
90 80
140 100
85 75
120 90 80 70
100 80 75 65
70 70
5000 5500 6000 6500 7000 7500 8000 8500 9000 5000 5500 6000 6500 7000 7500 8000 8500 9000 5000 5500 6000 6500 7000 7500 8000 8500 9000 5000 5500 6000 6500 7000 7500 8000 8500 9000
Wavelength in Angstrom Wavelength in Angstrom Wavelength in Angstrom Wavelength in Angstrom
180
350 220 150

170
200
300 160 140
180
250 150
160 130
140
200
140 130 120
150
5000 5500 6000 6500 7000 7500 8000 8500 9000 5000 5500 6000 6500 7000 7500 8000 8500 9000 5000 5500 6000 6500 7000 7500 8000 8500 9000 5000 5500 6000 6500 7000 7500 8000 8500 9000
700
Fig. 21 DFS fringe plots along dispersion direction for stellar magnitudes of 13 to 16 with
280 exposure time=60.0s (top row)
320
and
600 120.0s (bottom row)

400 270
300
260
500 350
280 250
400 300 260 240

80 Central Row Data Central Row Data
Fitted Curve Fitted Curve 230
300 700 240 650
250 220
70 5000 5500 6000 6500 7000 7500 8000 8500 9000 5000 5500 6000 6500 7000 7500 8000 8500 9000
650
5000 5500 6000 6500 7000 7500 8000 8500 9000 5000 5500 6000 6500 7000 7500 8000 8500 9000
600
60 600
Detector Counts
Detector Counts
Detector Counts
550
550
50
500 500
40
450 450
30 400 Central Row Data

400 Fitted Curve
5500 6000 6500 7000 7500 8000 8500 9000 5500 6000 6500 7000 7500 8000 8500 9000 7000 7250 7500 7750 8000 8250 8500 8750 9000
Wavelength in Angstrom Wavelength in Angstrom Wavelength in Angstrom
Fig. 22 DFS fringe intensity plots after curve fitting with piston errors of 306nm which is little larger than λ/2 (left),
10.0µm(middle), and 30µm(right). For 30.0µm piston DFS fringe windowing is used
6.2 Extraction of Piston values from given by the Equation 1

DFS spectrum
I(ω) = Io (y)[1 + γcos(2kδ + φ(y))] (1)
The DFS is usually considered as a phasing scheme
which provides segment piston errors less pre- where Io is the amplitude, γ is the fringe visi-
cisely over a large capture range (Shi et al, 2003; bility, k=2π/λ, δ=piston error and φ(y) is a phase
Smith et al, 2003). However, recent developments term that depends on where the fringe signal is
have shown that this technique is not only appli- extracted. Here ω is the position vector on image
cable to large piston errors, but it can also be plane with (x,y) as rectangular co-ordinates. The
used to measure phase errors with an accuracy of resultant intensity pattern on the detector plane
few nanometers, which is required by any diffrac- is a continuous spectrum modulated by sinusoidal
tion limited segmented mirror telescope (Zhao fringes. The frequency of the fringe as well as
and Cao, 2011; Li et al, 2017). In the subse- its tilt strongly depends on the piston error. In
quent sub-sections we have explored two different order to obtain the piston error, the intensity
methodologies to recover piston errors (δ), one for pattern of the recorded DFS fringes is fitted for
large (δ>λ/2) and another one for small (δ < λ/2) the Equation 1, considering three variables, Io , γ
errors. and δ. The curve fitting is usually carried out for
the central row, which has got highest intensity.
6.2.1 Measurement of Medium to Table 7 gives the extracted piston errors from the
large Piston errors simulated DFS spectra. From the Table 7 it is clear
The intensity distribution for the DFS fringe spec- that the achieved phasing accuracy is not accept-
trum has been derived in Shi et al (2004a) and is able even in case of the 13th magnitude star with
240.0s exposure. Thus we apply a signal enhance-

ment technique that may improve the DFS fringe 0
intensity signal and hence increase the measure-
50
ment accuracy. For signal enhancement, we sum 0 100 200 300 400 500 600
up the intensity data for multiple rows so that the 0
SNR can be improved. This helps to reduce the
50
measurement errors to some extent. From Table 7 0 100 200 300 400 500 600
it can also be observed that for larger piston errors 0
(30.0µm in our case), the fitting accuracy degrades
50
due to reduction in the fringe visibility, similar 0 100 200 300 400 500 600
is the case for lower piston errors (piston< λ/2),
0
where the number of fringes is only one, and hence
the fitting accuracy is very poor. To handle the 50
0 100 200 300 400 500 600
lower piston errors using DFS, few different tech-
10 20 30 40 50 60
niques are developed. One such technique that we
have used in this paper, is described in detail in Fig. 23 Iterative process of piston measurement using
the next subsection. DFS: Simulated Fringe images for i) δ0 =30µm ii)
δ1 =1.5834µm iii) δ2 =305.9172nm and iv) δ3 =69.4455nm
6.2.2 Measurement of Small Piston
Errors We have used DFA-LSR technique on simulated
dispersed fringe spectra subjected to piston error
In case of lower piston errors, i.e. δ < λ/2, only of 10.0-200.0nm. The values for the applied and
small fraction of a single fringe is available and recovered piston errors are listed in Table 8. The
the DFS signal primarily resides in the spectral maximum measurement error is approximately
continuum modulation. As we have discussed ear- 6.0nm which is well within the phasing require-
lier, in addition to piston error the DFS continuum ment of the telescope. Thus for large piston error
signal is highly affected by wavelength dependent the curve fitting technique is used which acts like
instrumentation throughput variation and hence a coarse phasing scheme that brings down the pis-
by using curve fitting method it is almost impos- ton error to the realm of DFA-LSR technique.
sible to get precise piston error measurement as Then further for (δ<λ/2) the DFA-LSR technique
required for diffraction limited telescope. To mea- provides very accurate piston measurement.
sure small piston errors there are other techniques
such as the Main Peak Position (MPP) method
6.2.3 An example case demonstration
(Zhao and Cao, 2011) and the Dispersed Fringe
for DFS
Accumulation based Left Subtract Right (DFA-
LSR), which relies on extraction of piston error The measurement error values reported in Table 7
using the information coded into DFS intensity are very pessimistic and suggest that attaining
distribution perpendicular to the dispersion direc- the required co-phasing accuracy with the pro-
tion. Since MPP method is highly dependent on posed APS will be difficult with faint stars but if
calibration and it requires higher SNR (Li et al, we employ the iterative approach for piston mea-
2017), therefore, we preferred to use the DFA-LSR surement with suitable extraction method then
scheme for our simulation which is best suited measurement accuracy of a few nanometers can
for piston errors < λ/2. In the DFA-LSR tech- be achieved. Here we consider an example case
nique, intensity of the DFS spectrum is added where the initial piston error is δ0 =30.0µm and
in the direction perpendicular to the dispersion it is brought down within a range of 5.0nm by
(columns are summed together). Depending on performing four consecutive measurements.
piston errors, from the summed intensity profile, Figure 23 shows the simulated fringe images
we find three or two peaks. Then, after following at each stage of iterative measurement (the sim-
the method described in Li et al (2017) vari- ulated fringes are for 13th magnitude star with
ous peak intensity ratios are derived which finally exposure time of 240 seconds and 1.0arc-sec see-
empirically get correlated with the piston errors. ing). For the 30.0µm piston, we apply the signal
Table 7 Piston extraction from fringes using curve fitting for star with 13th magnitude and 240.0s exposure time
Applied Measured Piston Measured Piston Error for Error for

Piston Using Central row Using Enhanced Signal Central row Enhanced Signal
0.1µm 0.284µm 0.259µm 0.184µm 0.159µm

10µm 8.951 µm 9.375µm -1.049µm -0.624µm
30µm 28.128µm 28.417µm -1.872µm -1.583µm
700 60
Central Row Data Central Row Data 80 Central Row Data 22000
Fitted Curve Fitted Curve Fitted Curve

650 20000
50 70
600
Summed Intensity Values

18000
60
Detector Counts
Detector Counts
Detector Counts
40 16000
550
50
14000
500
30 12000
40
450 10000
20
30 8000
400
6000
7000 7250 7500 7750 8000 8250 8500 8750 9000 5500 6000 6500 7000 7500 8000 8500 9000 5500 6000 6500 7000 7500 8000 8500 9000 0 10 20 30 40 50
Wavelength in Angstrom Wavelength in Angstrom Wavelength in Angstrom Pixels
Fig. 24 Iterative process of piston measurement using DFS: i) Curve fitting for δ0 =30µm ii) Curve fitting for δ1 =1.5834µm
iii) Curve fitting for δ2 =305.9172nm iv) Summed intensity values for DFA-LSR with δ0 =69.4455nm
Table 8 Piston extraction from fringes using DFA-LSR

of the segments by using the iterative measure-
for star with 10th magnitude and 240.0s exposure time
ment process. The values obtained are well within
Applied Measured Error (in nm) our requirements, however, in addition to simu-
Piston (in nm) Piston (in nm) lation we need to experimentally test and verify
it.
10 8.70 -1.30
25 21.29 -3.71
50
75
55.95
75.93
5.95
0.93
7 Discussions
100 97.87 -2.13
200 199.74 -0.25 7.1 The co-aligning and co-focusing
From our extensive simulations for segment co-
aligning we find that even under moderate see-
enhancement technique as well as DFS fringe win-
ing conditions, i.e. 1.5arc-sec, for seeing limited
dowing where, only a part of the DFS spectrum
telescope, which requires alignment accuracy of
that has higher visibility (700.0 to 900.0nm), is
0.04arc-sec (subsection 3.1), stars as faint as 17th
used. Then the curve fitting method is used which
magnitude in V-band can be used with a single
gives a residual error of δ1 =1.5834µm. Next, we
aperture and 10 second exposure. Looking at stel-
generate the fringe image for 1.58µm piston and
lar distribution in the sky, we will have several
apply curve fitting to the central row of the
thousands of stars brighter than 17th magnitude
fringe, the obtained measurement error for this
in the galactic plane and at least 5-6 stars at
stage was δ2 = 305.91nm. Again we generate the
the galactic pole in our annular guiding field of
next fringe image for δ2 and perform curve fitting
view. We find that our simulation results very well
on the central row. The obtained measurement
match with the studies carried out by Lee et al
error is now δ3 =69.44nm falling within δ < λ/2
(2010) on HET active mirror alignment control,
domain. Hence, at this stage DFA-LSR technique
which exactly works like any other monolithic thin
mentioned in subsubsection 6.2.2 is used. The
mirror active control (ignore segmented nature of
final measurement error obtained is 5.13nm which
the primary). It uses two off-axis SH wavefront
is well within our requirements. Figure 23 and
sensors, capable of working with a V-band lim-
Figure 24 show the simulated fringes as well as
iting magnitude of 18 with exposure time of 30
intensity plots obtained for all four stages of iter-
seconds. The active optics of 6.5m Magellan tele-
ation. From this case study it is clear that DFS
scope uses a 20x20 lenslet array (one lenslet covers
alone can handle the coarse as well as fine phasing
roughly 32.0cm of primary mirror aperture). Mag- the primary and secondary mirrors. Segmented
ellan active control is reported to work with stars mirror telescopes such as the SALT (Wirth et al,
as faint as 17th magnitude and has never faced 2004) and Keck Telescopes (Chanan et al, 1994)
any problems of getting suitable stars, even at have already implemented this alignment proce-
higher galactic latitude (Schechter et al, 2003). If dure where a single SH sensor is used for alignment
the segmented mirror telescope is intended to work of the segments as well as for the global alignment
in diffraction limited regime then segments need of the complete telescope.
to be aligned even more precisely(≤ 0.02 arc-sec Depending on the stiffness and stability of the
RMS). Since, diffraction limited imaging is gen- structures holding the telescope optics, it is quite
erally carried out in fairly good seeing conditions possible that active optics of monolithic mirror
and in that case the required alignment accuracy system can even work in open loop mode. In open
of 0.02arc-sec may be also achieved with similar loop mode it uses look-up tables which keep record
bright stars (16-17th magnitude) as used in the of all the corrections required for active optics to
seeing limited case (Figure 16). In our simulations function in varying temperature and the gravity.
we have used V-band which uses only a fraction There are many active telescopes which work very
of the star light. However, by using atmospheric smoothly in an open loop and deliver image qual-
dispersion corrector, it would be even possible to ity as good as a closed loop (Roddier et al, 1995;
make use of entire white-light and hence boost the Sawyer et al, 2000; Neufeld et al, 2004), (Reddy
signal further. K, ARIES 3.6m Devasthal Optical Telescope, Pri-
From the Monte Carlo simulations carried out vate Communication, 2022). If we go with this
for the estimation of SNR and centroiding error, it logic, then it is quite possible that without use
is clear that for off-axis stars of magnitudes 15 to of edge sensors, co-aligning and co-focusing of the
18 the centroiding error is minimal when the full seeing limited telescopes can probably be main-
segment aperture (≈ 100.0cm diameter) is con- tained over the night and possibly over weeks by
sidered. Thus, it can be concluded that using a just using look up tables. And this would make
larger aperture with a single spot is advantageous handling a segmented primary mirror remarkably
over using multiple smaller apertures. However, simple.
co-focusing requires multiple apertures per seg-
ment and for that we may have to use either longer 7.2 Co-phasing with the real time
exposure and/or about one magnitude brighter OAPS
stars which are still quite abundant within off-
axis guiding field over most parts of the sky, if Segmented mirror telescopes can achieve diffrac-
not all. In our co-focusing study we have con- tion limited performance only if their mirror seg-
sidered that OAPS uses a grid of 9 spots per ments are phased that means piston errors are
segment. We find that telescope imaging perfor- minimized to a small fraction of the observing
mance remains diffraction limited till centroiding wavelength (20.0-50.0nm). The twin Keck tele-
error smaller than 1.5 microns (1/10th of pixel). scopes are the only telescopes whose segmented
For one pixel (15.0µm), which is largest centroid- mirrors are not only successfully phased but its
ing error considered in our study, the change in the alignment and phasing is maintained by M1CS
D50 is about 18% (0.0349 to 0.0412arc-sec) and over several weeks. Though several methods for
Strehl Ratio becomes 0.9036 against 0.9980 for the phasing have been developed and experimented on
ideal telescope. The corresponding segment longi- the Keck (Chanan et al, 2000b; Shi et al, 2004a),
tudinal defcous value is found to be 1.68mm. By however, the phasing of the Keck telescopes is
looking at these numbers it appears that even the usually carried out using the broadband phasing
largest uncorrected segment defocus error will not technique, which can phase mirrors with nearly
degrade the imaging performance of seeing limited 30nm RMS (Chanan et al, 1998). For our sim-
telescope much. ulations we have chosen the DFS based phasing
Apart from the segment related errors, either technique and have attempted to explore the fea-
the same SH sensor or else an additional one can sibility of phasing the mirror segments using a
be used to find global alignment errors such as real time OAPS. From the simulations we find
the tip-tilt, decenter and despace errors between
case, DFS can be used by opting fringe win-

2.5
dowing and visibility enhancement techniques.
Phasing Domain Co-focusing Such a large piston error can also be easily
2.0 Domain
captured by a SH based device designed for
D50 value in Arcsec
the co-aligning and co-focusing, and/or by a

1.5
spherometer.
2. The piston errors between 300.0nm (±λ/2)
1.0
to 20.0µm can be easily and precisely mea-
sured using the curve fitting method. An error
0.5
of this order is expected in the beginning of
every night when OAPS will resume diurnal
0.0
observation.
10−1 100 101 102
Log10 of RMS Piston Error in micron 3. Finally, very small piston errors < ±λ/2 can be
measured using the DFA-LSR technique. This
Fig. 25 Variation of image FWHM with increasing piston
errors (Dotted line shows the 25.0µm piston beyond which
technique is also expected to run in a closed
the SH co-focusing device can be used for measurement) loop to maintain co-phasing of the mirror seg-
ments during the course of night observations.
that although a large aperture, helps to get bet- It appears that phasing of mirror segments
ter SNR but, it reduces fringe visibility and hence using stars as faint as 14th magnitude is very much
smaller aperture is preferred. Also, the simulations possible by using real time OAPS under moder-
show that co-phasing would require longer expo- ate seeing conditions. At present we have no clear
sure time and comparatively brighter stars (≤ 14th idea about the time required to phase the primary
magnitude), as compared to the alignment mode. mirror comprising of a large number of segments.
For the simulation we have chosen late type star However, considering that all segments are phased
(K spectral type) which are quite abundant in the simultaneously (not sequentially), it is expected
sky and also help to improve SNR in the DFS that the length of the phasing control loop would
spectral window. be just a few minutes. In order to maintain co-
Figure 25 shows the variation of image FWHM phasing over a phasing cycle of a few minutes, the
with increasing piston values. Due to phasing structure holding the segments needs to be made
effect FWHM marginally degrades till nearly very stiff.
0.3µm(λ/2) RMS piston error. Then, upto 25.0µm
FWHM shows the saturation effect probably 7.3 Advantages and disadvantages
attributed to depth of focus. Beyond 25.0µm the of the proposed scheme
FWHM starts increasing again and any piston
error value larger than 25.0µm will be seen sim- From our very extensive studies we can confidently
ilar to the defocus. DFS is usually considered say that the optical feedback based OAPS would
a phasing device having a large capture range work satisfactorily to maintain the alignment of
but poor measurement accuracy. However, one of the segments without use of the edge sensors. This
the important results that has been established will reduce complexities of the edge sensor based
through our simulations is that DFS alone can primary mirror control, thus making the telescope
be used for both coarse and fine phasing by just design more economic. As mentioned earlier, to
varying the piston extraction methods. From the make the design more affordable we would favour
simulations we predict that the piston measure- the use of circular segments, in which case the
ment range for our system would be 10.0nm to inter segment gaps will be larger and non-uniform
50.0µm. Within this range we can break down the due to curved edges and hence, use of conven-
piston values into three categories, tional edge sensors would be difficult, whereas,
the working of our OAPS (which is optical feed-
1. Segment subjected to large piston errors falling back based control system) will be unaffected by
in the range 20.0 to 50.0µm, is most likely the the size or shape of inter-segment gaps. Another
case when the mirror segments are installed in important advantage is that the optical feedback
the primary mirror cell for the first time. In this
based system will be capable of responding to and Sensor with a guide star as faint as 14th mag-
measuring the focus mode problem to which the nitude. Iterative measurements of piston can
edge sensors are insensitive. This has been verified be carried out to achieve a phasing accuracy of
by the active phasing experiments at ESO VLT 6nm.
(Gonte et al, 2006). 7. Our studies show that optical feedback based
Alignment and phasing device is an inherent OAPS, would work satisfactorily and it would
part of any segmented mirror telescope and the also help to eliminate the use of edge sensors,
same can be used to monitor and maintain pri- thus reducing the overall cost as well as com-
mary mirror figure in the real time. Although very plexity of the complete telescope system.
extensive simulations were carried out to test the
feasibility of the OAPS, yet some factors are left Acknowledgments
for further analysis, one of which is the optical
aberrations occurring due to the telescope optics. This research has made use of the High Perfor-
Since the OAPS samples the off-axis light hence mance Computing (HPC) system of Computer
the off-axis aberrations will be dominant and an Center of the Indian Institute of Astrophysics,
exclusive corrector optics would be required to Bangalore. Use of Photutils, an Astropy pack-
minimise these field aberrations. age for detection and photometry of astronomical
sources (Bradley et al, 2021) is also acknowledged.
8 Conclusions
Declarations
By making use of innovative ideas as well as latest
technology, it is quite possible to reduce the cost The authors have no relevant financial or non-
of the moderate to large size telescopes so that financial interests to disclose.
it can be made affordable to many small institu-
tions/ countries. In this and subsequent papers we Funding
are presenting our idea, which is strongly backed
by extensive design and simulation. From our sim- No funding was obtained for this study.
ulation based study we conclude the following:
1. Segmented mirror technology is the key behind
Author’s Contribution
cost effectiveness. The simulation and design, which forms the
2. Instead of using hexagonal mirror segments, we core part of this paper is primarily carried out
propose to use circular segments which would by Radhika Dharmadhikari. Whereas, the main
degrade the telescope image quality marginally manuscript is prepared by Radhika Dharmad-
but will help to reduce the cost substantially. hikari and Padmakar Parihar. Annu Jacob has
3. In place of edge sensor based complex and provided a python based mirror segmentation tool
expensive primary mirror control system, we to generate a grid sag surface. All authors have
propose to utilize an OAPS which is an active reviewed the manuscript.
mirror kind of mechanism working in real time.
4. From our simulations we find that the pre-
cise co-alignment of segments is possible with Data Availability
OAPS using stars as faint as 17-18th magni- The data that supports the findings of this study
tudes. Real time co-focusing, which requires are available from the corresponding author, Rad-
multiple apertures per segment can be done hika Dharmadhikari, upon reasonable request.
with one magnitude brighter star as required
for co-alignment.
5. A seeing limited telescope requires only co- References
aligning and co-focusing which can be very
Abt HA (2012) Scientific Efficiency of Ground-
easily achieved using a real time OAPS.
based Telescopes. Astron. J.144(4). https://doi.
6. For the diffraction limited performance, co-
org/10.1088/0004-6256/144/4/91
phasing can be done using a Dispersed Fringe
Anupama GC, Maheswar G, Sriram S, et al (2022) In: Ground-based and Airborne Telescopes VI,
A 10-m class national large optical-IR telescope. https://doi.org/10.1117/12.2232263
Journal of Astrophysics and Astronomy 43.
https://doi.org/10.1007/s12036-022-09819-6 Doi M, Miyata T, Yoshii Y, et al (2018) The
University of Tokyo Atacama Observatory 6.5m
Bradley L, Sipőcz B, Robitaille T, et al telescope: project overview and current status.
(2021) astropy/photutils: 1.0.2. https://doi. In: Ground-based and Airborne Telescopes VII,
org/10.5281/zenodo.4453725 https://doi.org/10.1117/12.2313099
Buckley DAH, Meiring JG, Swiegers J, et al (2004) Gajjar H, Menzies J, Swiegers J, et al (2006)
Many segments and few dollars: SALT solutions Results from the capacitive edge sensing system
for ELTs? In: Second Backaskog Workshop on for the active alignment of the SALT primary
Extremely Large Telescopes, pp 245–256, https: mirror. In: SPIE Conf. Series, https://doi.org/
//doi.org/10.1117/12.566263 10.1117/12.672103
Chanan G, Troy M, Dekens F, et al (1998) Gajjar H, Menzies J, Buckley D, et al (2016)

Phasing the Mirror Segments of the Keck Tele- SALT: Active control of the primary mirror
scopes: The Broadband Phasing Algorithm. with inductive edge sensors. In: Ground-based
Appl. Opt.37. https://doi.org/10.1364/AO.37. and Airborne Telescopes VI, https://doi.org/
000140 10.1117/12.2234264
Chanan G, Ohara C, Troy M (2000a) Phas- Gonte F, Yaitskova N, Derie F, et al (2006) The
ing the Mirror Segments of the Keck Tele- active phasing experiment: Part II. Design and
scopes II: The Narrow-band Phasing Algorithm. developments. In: SPIE Conf. Series, https://
Appl. Opt.Vol. 39:4706–4714. https://doi.org/ doi.org/10.1117/12.672057
10.1364/AO.39.004706
Gutiérrez CM, Torres M, Oria A, et al
Chanan GA, Nelson JE, Mast TS (1986) Seg- (2021) The 4 m New Robotic Telescope
ment alignment for the Keck Telescope pri- Project: An Updated Report. In: Revista Mex-
mary mirror. In: Advanced technology optical icana de Astronomia y Astrofisica Conference
telescopes III, pp 466–470, https://doi.org/10. Series, pp 8–13, https://doi.org/10.22201/ia.
1117/12.963566 14052059p.2021.53.03
Chanan GA, Nelson JE, Mast TS, et al (1994) Jacob A, Parihar P, James MK (2020) Cre-
W.M. Keck Telescope phasing camera system. ating a large aspheric primary mirror
In: Instrumentation in Astronomy VIII, pp using spherical segments. Experimen-
1139–1150, https://doi.org/10.1117/12.176697 tal Astronomy Vol. 50:51–71. https:
//doi.org/10.1007/s10686-020-09663-y
Chanan GA, Troy M, Ohara CM (2000b) Phas-
ing the primary mirror segments of the Keck Khosroshahi HG, Jenab H, Bidar M, et al (2016)
telescopes: a comparison of different techniques. Iranian National Observatory: project overview.
In: Optical Design, Materials, Fabrication, and In: Ground-based and Airborne Telescopes VI,
Maintenance, pp 188–202, https://doi.org/10. https://doi.org/10.1117/12.2241680
1117/12.391510
Lee H, Hart M, Hill GJ, et al (2010) Analysis of
Chanover N, Williams B, Crenshaw DM, et al active alignment control of the Hobby-Eberly
(2019) The Importance of 4m Class Observato- Telescope wide-field corrector using Shack-
ries to Astrophysics in the 2020s. In: Bulletin of Hartmann wavefront sensors. In: Modeling, Sys-
the American Astronomical Society, p 31 tems Engineering, and Project Management
for Astronomy IV, https://doi.org/10.1117/12.
Cui Xq, Zhu Yt (2016) Chinese Large Optic/IR 857121
Telescope (LOT): planning for the next decade.
Li Y, Wang S, Rao C (2017) Dispersed-fringe- Ninane N, Flebus C, Kumar B (2012) The 3,6 m
accumulation-based left-subtract-right method Indo-Belgian Devasthal Optical Telescope: gen-
for fine co-phasing of a dispersed fringe sen- eral description. In: Ground-based and Airborne
sor. Appl. Opt.Vol. 56:4267. https://doi.org/10. Telescopes IV, p 84441V, https://doi.org/10.
1364/ao.56.004267 1117/12.925921
Lousberg GP, Mudry E, Bastin C, et al (2016) Oswalt TD, McLean IS (2013) Planets, Stars
Active optics system for the 4m telescope of and Stellar Systems, Volume 1: Telescopes and
the Eastern Anatolia Observatory (DAG). In: Instrumentation. Springer, Dordrecht, https://
Advances in Optical and Mechanical Technolo- doi.org/10.1007/978-94-007-5621-2
gies for Telescopes and Instrumentation II,
https://doi.org/10.1117/12.2234261 Rakoczy JM, Hall D, Howard RT, et al (2003) Pri-
mary mirror figure maintenance of the Hobby-
Lubliner J, Nelson JE (1980) Stressed mirror Eberly Telescope using the Segment Align-
polishing. 1: A technique for producing nonax- ment Maintenance System. In: Large Ground-
isymmetric mirrors. Appl. Opt.Vol. 19. https: based Telescopes, pp 702–713, https://doi.org/
//doi.org/10.1364/AO.19.002332 10.1117/12.456734
Mahajan V (2013) Optical imaging and aberra- Roddier NA, Blanco DR, Goble LW, et al (1995)
tions, part iii: Wavefront analysis (spie, 2013), WIYN telescope active optics system. In: Tele-
vol. PM221 pp 88–90 scope Control Systems, pp 364–376, https://
doi.org/10.1117/12.211446
Marchiori G, Busatta A, De Lorenzi S, et al (2012)
A new era for the 2-4 meters class observato- Rozière D, Luong B, Fuchs B, et al (2008) Induc-
ries: an innovative integrated system telescope- tive edge sensors: an innovative solution for
dome. In: Ground-based and Airborne Tele- ELT segmented mirror alignment monitoring.
scopes IV, https://doi.org/10.1117/12.927117 In: Ground-based and Airborne Telescopes II,
https://doi.org/10.1117/12.789444
Minor RH, Arthur AA, Gabor G, et al (1990) Dis-
placement sensors for the primary mirror of the Sagar R (2000) Importance of small and mod-
W. M. Keck telescope. In: Advanced Technol- erate size optical telescopes. Current Science
ogy Optical Telescopes IV, https://doi.org/10. 78(9):1076–1081
1117/12.19270
Sawyer DG, Corson C, Saha A (2000) Optimizing
Nagata T, Kurita M (2020) Seimei 3.8-m Tele- the delivered image quality at the WIYN 3.5-
scope has been commissioned. In: SPIE Conf. m Telescope. In: Telescope Structures, Enclo-
Series, https://doi.org/10.1117/12.2561272 sures, Controls, Assembly/Integration/Valida-
tion, and Commissioning, pp 422–430, https:
Nelson J, Mast T, Chanan G (2013) Segmented //doi.org/10.1117/12.393896
Mirror Telescopes. In: Planets, Stars and Stellar
Systems. Volume 1: Telescopes and Instrumen- Schechter PL, Burley GS, Hull CL, et al (2003)
tation. Springer, Dordrecht, p 99, https://doi. Active optics on the Baade 6.5-m (Magellan I)
org/10.1007/978-94-007-5621-2 3 Telescope. In: Large Ground-based Telescopes,
pp 619–627, https://doi.org/10.1117/12.457899
Nelson JE, Mast TS, Faber SM (1985) The design
of the keck observatory and telescope. Caltech Schmidt-Kaler T, Rucks P (1997) Telescope costs
and cost reduction. In: Optical Telescopes of
Neufeld C, Zolcinski-Couet MC, Keane M, et al Today and Tomorrow, pp 635–640, https://doi.
(2004) The active primary mirror assembly for org/10.1117/12.269092
the SOAR telescope. In: Ground-based Tele-
scopes, pp 870–880, https://doi.org/10.1117/ Shelton C, Mast T, Chanan G, et al (2008)
12.551375 Advances in edge sensors for the Thirty Meter
Telescope primary mirror. In: Ground-based van Belle GT, Meinel AB, Meinel MP (2004)
and Airborne Telescopes II, https://doi.org/10. The scaling relationship between telescope cost
1117/12.790415 and aperture size for very large telescopes. In:
Ground-based Telescopes, pp 563–570, https:
Shi F, Redding DC, Lowman AE, et al (2003) Seg- //doi.org/10.1117/12.552181
mented mirror coarse phasing with a dispersed
fringe sensor: experiments on NGST’s Wave- Walker DD, Beaucamp ATH, Doubrovski V,
front Control Testbed. In: IR Space Telescopes et al (2006) Automated optical fabrication: first
and Instruments, pp 318–328, https://doi.org/ results from the new Precessions 1.2m CNC pol-
10.1117/12.461113 ishing machine. In: SPIE Conf. Series, https:
//doi.org/10.1117/12.671098
Shi F, Chanan G, Ohara C, et al (2004a) Experi-
mental Verification of Dispersed Fringe Sensing Wang S, Zhu Q, Cao G (2009) Dispersed Rayleigh
as a Segment Phasing Technique using the Keck interferometer. In: 4th International Sympo-
Telescope. Appl. Opt.Vol. 43:4474–4481. https: sium on Advanced Optical Manufacturing and
//doi.org/10.1364/AO.43.004474 Testing Technologies: Optical Test and Mea-
surement Technology and Equipment, https://
Shi F, Redding DC, Green JJ, et al (2004b) Per- doi.org/10.1117/12.828774
formance of segmented mirror coarse phasing
with a dispersed fringe sensor: modeling and Wirth A, Gonsiorowski T, Roberts J, et al (2004)
simulations. In: Optical, Infrared, and Millime- Developing and testing an optical alignment
ter Space Telescopes, pp 897–908, https://doi. system for SALT”s segmented primary mir-
org/10.1117/12.552323 ror. In: Ground-based Telescopes, pp 892–902,
https://doi.org/10.1117/12.551425
Smith EH, Vasudevan G, Reardon RD, et al
(2003) Coarse phasing of a segmented mirror Wolf MJ, Palunas P, Booth JA, et al (2003) Mir-
using a dispersed fringe sensor. In: IR Space ror Alignment Recovery System (MARS) on
Telescopes and Instruments, pp 469–477, https: the Hobby-Eberly Telescope. In: Large Ground-
//doi.org/10.1117/12.461577 based Telescopes, https://doi.org/10.1117/12.
458034
Smith HJ (1989) A Decade of Cost-Reduction
in Very Large Telescopes - the SST as Pro- Yeşilyaprak C, Keskin O (2020) Eastern Ana-
totype of Special Purpose Telescopes. Astro- tolia Observatory (DAG): the status in 2020.
phys. Space Sci.160:123–134. https://doi.org/ In: SPIE Conference Series, https://doi.org/10.
10.1007/BF00642762 1117/12.2560942
Stahl HP, Henrichs T (2016) Multivariable para- Zhao W, Cao G (2011) Active cophasing and
metric cost model for space and ground tele- aligning testbed with segmented mirrors. Optics
scopes. In: Modeling, Systems Engineering, and Express Vol. 19:8670. https://doi.org/10.1364/
Project Management for Astronomy VI, https: OE.19.008670
//doi.org/10.1117/12.2234088
Stephan C, Guisard S, Bourget P (2016) Long-

term performance of the VLT UT active optics
system. In: Ground-based and Airborne Tele-
scopes VI, https://doi.org/10.1117/12.2231892
Townson MJ, Farley OJD, Orban de Xivry G, et al

(2019) AOtools: Adaptive optics modeling and
analysis toolkit

Building A Large Affordable Optical-NIR Telescope

Uploaded by

Copyright:

Available Formats

Building A Large Affordable Optical-NIR Telescope

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Building A Large Affordable Optical-NIR Telescope

Uploaded by

Copyright:

Available Formats

Building a large affordable Optical-NIR telescope (I):

An alternate way to handle segmented primary

Keywords: Segmented Mirror, Alignment, Phasing, Optical Telescopes, Simulations

Posted Date: May 16th, 2023

Additional Declarations: No competing interests reported.

Building a large affordable Optical-NIR telescope (I): An

*Corresponding author(s). E-mail(s): radhika.dhar@iiap.res.in;

Keywords: Segmented Mirror, Alignment, Phasing, Optical Telescopes, Simulations

1 Introduction mega optical-IR telescopes of 25.0-40.0m aper-

2 Alternate way to handle segmented primary mirror

Alternate way to handle segmented primary mirror 3

4 Alternate way to handle segmented primary mirror

compared the image qualities for hexagonal and

Fig. 2 Two ring Segmented primary mirror with diame-

In order to understand the effect of segment

Alternate way to handle segmented primary mirror 5

6 Alternate way to handle segmented primary mirror

case of piston errors, the D80 is strongly affected D50

degrade rapidly till piston error of about 0.3µm

Encircled Energy in Arcsec

are phased using the device called Phasing Cam-

Alternate way to handle segmented primary mirror 7

mirror telescope for seeing and diffraction limited

Table 1 Alignment requirements for SMTs

Alignment Error Seeing Diffraction

8 Alternate way to handle segmented primary mirror

Fig. 7 Schematic representation of the two modes of the Off-Axis APS

The parameters, such as the star magnitude,

Alternate way to handle segmented primary mirror 9

used V-band observations of M67 open stellar clus- y=x

more, we compared the radial profile of synthetic 5 5

and observed stellar images (Figure 12) which 10 10

remarkably matches in both x and y directions. 15 15

Table 2 Parameters for the HCT Observations 30 30

Wavelength Range 470.0-700.0nm∗

Reflectivity of the mirror 90%

10 Alternate way to handle segmented primary mirror

30000 Table 4. The camera focal length is chosen such

We plan to use an OAPS working in the real Parameter Value

∗ assuming achromatic doublet lens

Alternate way to handle segmented primary mirror 11

Stellar Magnitude=16, Exposure time=10s 0

Seeing=1" 100 100 100

Seeing=2" 200 200 200

250 250 250

300 300 300

150 350 350 350

400 400 400

Fig. 14 Variation of SNR with aperture diameter for 16th

12 Alternate way to handle segmented primary mirror

Seeing=1" Exposure time=10s Seeing=1.5" Exposure time=10s Seeing=2" Exposure time=10s

Radial Centroiding Error in Arcsec

Radial Centroiding Error in Arcsec

reconstruction method, wavefront error is decom-

10−1 5.2.1 Simulations on Co-focusing

Alternate way to handle segmented primary mirror 13

14 Alternate way to handle segmented primary mirror

Table 5 Variation of Image Quality Parameters with Segment Defocus

0.3 µm 1/50 33.80 µm 1.18 µm 0.035 0.094 0.998