Docstoc

aaomega cod Document

Document Sample
aaomega cod Document Powered By Docstoc
					          AAOmega
           MULTI-PURPOSE
       OPTICAL SPECTROGRAPH




CONCEPTUAL DESIGN STUDY DOCUMENTATION




     ANGLO – AUSTRALIAN OBSERVATORY
              20 August, 2001
                                               AAOmega CoD




Revision History
                   Roger Haynes
Revision 1         20 August 2001   Original issue




                              2
                                                                      AAOmega CoD




Executive Summary
The AAOmega project aim is to provide dramatic improvements to the performance
of the AAT's highly successful 2dF fibre multi-object system (MOS). Essentially the
project can be split into three components:

1. Provision of a new, stable, high efficiency, bench mounted spectrograph to replace
   prime focus top end ring mounted 2dF spectrographs.

2. Improvements to the 2dF positioning system

3. Provision of an enhanced integral field capability using the SPIRAL Integral Field
   Unit (IFU)

It is expected that by developing a new bench mounted spectrograph with Volume
Phase Holographic Gratings, the modern high quantum efficiency detectors, the latest
anti-reflection and mirror coating technologies, the AAOmega system will provide
much higher throughput (3-4 times higher) and higher spectral resolution ( > 2 times)
and far better stability than with the existing 2dF spectrographs.

Most of the concept design work presented concentrates on characterising, costing
and assessing the feasibility of a number of potentially promising optical concepts in
order to select the most effective design for AAOmega. The two main competing
concepts were a single beam fully transmissive optical design and a dual beam
Schmidt system (DBSS).

A detailed comparison of the two main designs was carried out and both systems had
their pros and cons. On balance of performance and cost they came out very close.
The transmissive system won out in: overall system throughput, particularly in IFU
mode, inherent mechanical design risk, ease of maintenance/support and benefited
from double the contiguous wavelength coverage at high resolution than the DBSS.
The DBSS won out in terms of: image quality (the poorer performance of the
transmissive system noticeably degrades spectral and spatial resolution particularly in
the IFU mode), better blue throughput, marginally better low resolution wavelength
coverage and ability to perform high dispersion observations of two greatly separated
spectral regions simultaneously.

The very high risks associated with the cost, availability, complexity and manufacture
of the glass components of the transmissive system (including deep aspheric lenses
and specialised glasses) leads us to conclude that the DBSS is the preferred option for
the design of the AAOmega spectrograph. There are additional concerns that even
with further development, the transmissive optical concept would fail to meet the
required image quality specification.

It was always expected that the replacement spectrograph would constitute the bulk of
the effort and cost of the project. This is reflected very much in the cost estimates
with approximately 94% of the $3.6M project estimate being attributed to the




                                           3
                                                                       AAOmega CoD


development, manufacture (completion expected in early 2004) and commissioning
(first light expected semester 2004B) of the new spectrograph.

The 2dF corrector and positioning system now perform with commendable reliability
and efficiency. However, during operation over the past few years, a number of areas
have been highlighted where further improvements in reliability and/or efficiency
could be made with relatively small investments of effort and money. The AAOmega
project plans to address many of these. However, it will only target those areas that
are problematic and could deliver significant efficiency improvements, or simplify /
reduce operational overheads. In fact, virtually all of the current 2dF system up to the
fibre bundles and spectrographs will remain fundamentally unchanged, including the
2dF ADC.

The integral field unit (IFU) that will be mounted at the Cassegrain focus, will not
only provide a valuable 2 dimensional spectroscopic capability, but also significantly
provide for the science niche currently fulfilled by the aged RGO long slit
spectrograph, at an estimated cost of only $10K. The system will re- use most of the
current SPIRAL IFU with a small modification to the output slit, and modifications to
its mount on the Cassegrain rotator. The IFU will be able to mount on either the
straight-through Cassegrain focus, where it will be able to perform standard IFU
spectroscopy and spectropolarimetry, or at Auxiliary Cassegrain focus, without
polarimetry mode, not disturbing instruments at straight-through Cassegrain. This
should allow for much greater flexibility in scheduling of Cassegrain instrumentation.

The overall cost estimates (~$3.6M) for the AAOmega project are higher than
initially intended (~$3M). However, it is strongly believed that the spectrograph
concept and 2dF upgrades proposed would provide a very powerful and scientifically
flexible instrument that would ensure that the AAT remains competitive with the 8m
class telescopes and instrumentation well beyond 2005.




                                           4
                                  AAOmega CoD



Document authors:

Jeremy Bailey
Joss Bland-Hawthorn
Brian Boyle
Terry Bridges
Tony Farrell
Gabriella Frost
Peter Gillingham
Roger Haynes
Damien Jones (Prime Optics)
Ian Lewis
Gordon Robertson
Greg Smith
Lew Waller


Other contributors:

Scott Croom
Will Saunders
Fred Watson




                              5
                                                                                AAOmega CoD




Contents


1    Introduction ............................................................................... 9

2    Science Drivers ....................................................................... 13

3    Scientific Requirements for Software .................................. 19

4    Sky Subtraction ....................................................................... 23

5    Functional Specification ........................................................ 31

6    Fibre Selection ........................................................................ 35

7    Optical Design ......................................................................... 39

8    Mechanical Design.................................................................. 89

9    Detectors ................................................................................ 103

10 Electronics ............................................................................. 107

11 Software ................................................................................. 125

12 2dF Upgrade .......................................................................... 127

13 Risk Management ................................................................. 131

14 Costs....................................................................................... 151

15 Comparison and Conclusions ............................................ 159

16 List of Acronyms................................................................... 167




                                                 7
                                                                       AAOmega CoD




1 Introduction

1.1 The Aim of AAOmega
This document presents the results of the Concept Design Study undertaken for
AAOmega, the next generation optical spectrograph for the AAT. The aim of
AAOmega is to exploit the field of view and multiplex advantage of 2dF at as faint a
flux limit as practical, and to provide single object integral field capability, over as
wide a range of astronomical parameter space as possible.

AAOmega will consist of an innovative new spectrograph syste m fed by multi-object
fibres from a refurbished 2dF instrument and IFU fibres from an upgraded SPIRAL
instrument. AAOmega will provide much higher spectral resolution (a factor >2),
throughput (a factor of 3-4 higher), and stability than the existing 2dF instrument,
while maintaining the wide field and multiplexing advantages of 2dF, thus enabling
science that is simply not possible with 2dF, or other instruments on 4m or larger
telescopes.


1.2 AAOmega History
   May 2000: Keith Taylor had a `vision' for a successor to 2dF, fleshed out in a
    meeting with AAO staff on 2 May 2000. The main elements were:

-   dual-beam optical/IR (0.4 - 1.3 m)
-   optical: 2 spectrographs, each with 4k x 4k CCD, 20 pixels/fibre
-   IR: 1 spectrograph, 2k x 2k HgCdTe device, ~15 pixels/fibre
-   factor of ~3 gain in overall throughput compared to 2dF
-   Nod & Shuffle integral to AAOmega (tricky in IR!)
-   addition of new Prime Focus IFU on 2dF tumbler

   It was soon realised that the IR part would be too expensive, and the first
    downgrading occurred, to 2 optical spectrographs.

   September 2000: Science brainstorming session held at the AAO, to prepare a
    preliminary science case for AAOmega.

   October 2000: Preliminary User Requirements Document (PURD) compiled by
    David Lee and Terry Bridges. The PURD contained both a functional overview of
    AAOmega as envisaged at the time, and the AAOmega Science Case, which is
    still very relevant.

-   PURD sent to the Australian and UK communities at the end of October, with
    feedback received in November 2000.




                                           9                               Introduction
                                                                       AAOmega CoD


   Feb 2001: Initial costing showed that the two-spectrograph design was too
    expensive, at ~5 million AUD. A further downgrading resulted, and a Single-
    Spectrograph / fully transmissive design was pursued. A Concept Design study for
    the Single-Spectrograph design initiated (in particular the functional and technical
    specifications).

   April 2001: Update on the AAOmega Single-Spectrograph design sent to the
    community, with feedback received.

   May 2001: Will Saunders put forward the idea of a Dual- Beam Schmidt
    Spectrograph (DBSS) design, with two cameras, one optimised for the blue and
    one for the red. The DBSS design would have a single slit and collimator, a
    dichroic beam-splitter, then the two cameras, each with its own corrector; this was
    thought to be simpler and cheaper than the original two-spectrograph design
    mentioned above. Will and Peter Gillingham, with help from Joss Bland-
    Hawthorn, have pursued this concept, in parallel with the Single Beam
    Transmissive design.

   July 2001: The Prime Focus IFU was dropped during the AAOUC meeting, as it
    was considered that the upgraded SPIRAL B would fulfil the requirements for
    IFU science.

   August 2001: Concept Design phase finishes, with a choice made between the
    Single Beam Transmissive and DBSS designs, based on science, performance,
    risk and cost.

   September 2001: Approval of the AAT Board will be sought. If approval given,
    we move into the Preliminary Design phase. It is estimated that the first light for
    AAOmega would be semester 2004B.


1.3 AAOmega Outline

1.3.1 Elements:

   Multi-Object Spectroscopy (MOS) with 400 fibres, 200 with Nod & Shuffle
   Existing 2dF infrastructure, with several upgrades
   Existing SPIRAL IFU to be kept at Cassegrain, with spectropolarimetry available
   New AAOmega spectrograph: stable, bench-mounted, with higher spectral
    resolution and throughput
   Upgrades of configuration and data reduction software.




                                           10                              Introduction
                                                                       AAOmega CoD



1.3.2 Multi-Object Spectroscopy

The 2dF fibre bundles will be replaced with Heraeus STU fibre. The new fibres will
be 27m long, to reach from the AAT top-end to the new spectrograph mounted on the
south catwalk.

Nod & Shuffle will be available for 200 MOS fibres, and there will be a new CCD
autoguiding system for the MOS fibres. The existing 2dF corrector will be used in
order to maintain a 2 degree field of view (f.o.v.).

1.3.3 IFU

The SPIRAL B IFU itself would be unchanged, except for modifications to the slit
unit for use with the new AAOmega spectrograph. It could be used either at auxillary
Cassegrain port with other Cassegrain instruments remaining, or by itself at a straight
through Cassegrain focus (spectropolarimetry only at this port). The spatial sampling
and f.o.v. are currently 0.7‖/fibre and 11‖ x 22‖ respectively, but the foreoptics could
be changed to give finer spatial sampling with a reduced f.o.v. The IFU wo uld be
rotatable, and could also provide spectropolarimetry. The IFU could also be used in
Nod & Shuffle mode over its full f.o.v. Nod & shuffle is expected to be its normal
mode of operation.

1.3.4 AAOmega Spectrograph

A new spectrograph is at the heart of AAOmega. As already mentioned, during the
concept design study we have looked at two main options: a Transmissive Single
Beam Spectrograph (TSBS) and a Dual Beam Schmidt Spectrograph (DBSS) option.
Within these two main options, we have looked into a number of variations, which are
described in the following sections. The two concepts do have, however, a number of
common features:

   Thermally stable and bench- mounted on the AAT south catwalk (27m of fibre
    required)
   A total of 4k x 4k detector pixels (probably two 2k x 4k devices) with 15 micron
    pixels.
   AAO-2 CCD controllers, which will be more reliable and easier to use with nod &
    shuffle
   Articulated camera (0-90 deg), with remote control of many mechanisms
   VPH gratings for medium and high dispersion (transmission gratings, or grisms
    for low dispersion)
   Optimum anti-reflection / reflection coatings (including AR coatings for VPH
    gratings) for high efficiency
   Wavelength coverage: 370-950 nm.




                                           11                              Introduction
                                                                          AAOmega CoD



2 Science Drivers
AAOmega will have much higher throughput, spectral resolution, and stability than
the current 2dF, while retaining the wide-field and multiplex advantages of 2dF and
the integral- field capability of SPIRAL. As a result, AAOmega will be the most
efficient multi-object spectrograph on any 4m telescope, either in terms of the total
number of photons received per unit time (P), or the time required to cover a given
area of sky (T), where

       Throughput = Spectrograph Efficiency  Telescope Area,
       P  Nobj  Throughput, and
       T  1 / (F.O.V  Throughput}

Because of its large number of fibres, AAOmega will be competitive with the other
multi-object fibre spectrographs in terms of total number of photons received, even
compared to OzPoz on the 8m VLT. Because of the large field of view, the time
required to cover a given area on the sky is less than the other systems. See Table 2.1
for a tabulation of AAOmega versus the Sloan and OzPoz fibre systems, and the
VIMOS multi-slit system.

                                        AAOmega Sloan OzPoz VIMOS Formula
                                          AAT          VLT    VLT
      Telescope aperture    D       m         3.9  2.5      8     8
    Area of primary mirror  A      m^2       11.9  4.9   50.3  50.3 A = pi/4.D^2
            Spectro efficy eta                0.5  0.3    0.3   0.4 rough estimates
                      FoV          deg          2    3   0.42  0.50
                FOV area Afov     deg^2      3.14 7.07   0.14  0.20 Afov = pi/4.FoV ^2
             No of objects Nobj               400  640    132   560
   No of photons rec'd per  P               2389   942  1991 11259 P = eta.Nobj.A
                 unit time
 Time to cover given area   T                  0.053   0.096   0.486   0.253 T = 1/(eta.A fov.A)
                    of sky




Table 2.1 Comparison of AAOmega with other visible multi-object spectrographic
systems. Note, the spectrograph efficiencies are rough estimates as no hard data was
available for any system.

Thus, AAOmega will be able to do science that is simply not possible with 2dF, nor
with other multi-object spectrographs on 4m or larger telescopes. AAOmega will be a
powerful and versatile instrument, capable of carrying out a diverse range of science.
However, as well as being a flexible, general-purpose instrument, its real strength will
lie in carrying out large projects; some possibilities are described below.




                                          13                              Science Drivers
                                                                        AAOmega CoD


2.1 Overview of AAOmega Science
Here we give a broad overview of the science that will be possible with AAOmega;
for more details, please see the AAOmega User Requirements Document (URD).

2.1.1 Multi-Object Spectroscopy:

   Brown dwarfs and white dwarfs
   Age-dating Open Clusters
   High Spectral Resolution Stellar Applications (see also below)
   Planetary Nebulae and Globular Clusters in Nearby Galaxies (see below)
   Targeted Follow- up of Galaxy/QSO Redshift Surveys (2dF, Sloan, etc; see below)
   Studies of Galaxy Clusters
   Follow- up of Deep Radio Surveys
   Deep K-selected Redshift Surveys
   Deep Wide-Area QSO Surveys

2.1.2 Integral Field Spectroscopy:

   Emission-line mapping
   ISM
   Objects on Complex Backgrounds
   Spectropolarimetry (stars, AGN)
   Time Series applications
   Dynamics and Population Synthesis of Nearby Galaxies
   Imaging and Kinematics of Gravitational Lensing

2.2 AAOmega Key Science Drivers
Now we describe in more detail two large projects that AAOmega would be well-
suited to carry out: (i) Targeted follow-up of the 2dFGRS/QSO and other surveys;
and (ii) A large survey of Galactic structure.

2.2.1 Survey Follow-up Science

Stellar Populations in Galaxies and CDM Archaeology: Follow-up of the 2dF
Galaxy Redshift Survey    (Joss Bland-Hawthorn)

We propose to follow up the 2dFGRS by deriving accurate light-weighted
metallicities and ages for a large fraction of the galaxies. Future CDM simulations
will be able to predict these properties for individ ual galaxies. Wide- field mapping of
galaxy ages and metallicities will usher in a new discipline of `galactic archaeology'
in order to test CDM in the non- linear regime. This is an important new area that the
Sloan project will not be able to carry out effectively for two key reasons: the Sloan
telescope has a small aperture, and the spectrographs are restricted to a resolving
power of R = 2000. With the 2dFGRS, we will be able to select galaxies by spectral
type, and we can compare ages and abundances for early and late type galaxies. We


                                          14                            Science Drivers
                                                                      AAOmega CoD


can also isolate cluster and field galaxies, and determine ages and abundances
separately for them. Hierarchical merging models make predictions about the relative
ages of field and cluster galaxies in particular, which can be tested with the large
samples that we propose.

How best to measure galaxy ages is a difficult subject. The age-sensitive aspect of a
stellar population that is the most understood and most readily quantifiable involves
the main sequence turnoff. The further you go to the blue, the more you isolate this
population in the integrated spectrum of a galaxy, i.e., you minimize the
―contamination" coming from the giant branch and the lower main sequence. This
holds true all the way into the near-UV, at which point you start to receive
contamination from UV-bright stars whose origin, and connection with age, are still
poorly understood. The key lines are the Balmer series, particularly H, H, and
higher. The high AAOmega resolution would allow us to separate the Balmer lines
from the dense thicket of metal lines, and from emission lines which occasionally fill
in the absorption lines. We would need R > 4000 and excellent spectral response near
or below 4000 Angstrom. Blue-sensitive CCDs make this proposition very attractive
for AAOmega. The same spectra would also yield velocity dispersions and projected
rotation velocities, so that the variation in the Fundamental Plane and Tully-Fisher
relation with local environment could also be investigated.

The 2dFGRS survey is based on a mean S/N ~ 12 at R = 750 in a one hour exposure.
At a minimum, AAOmega would need to achieve a S/N ~ 30 at R = 4500, which
should be possible in a statistical sense averaged over neighbouring groups of ten or
more galaxies. The new spectrograph+CCD combination, along with better
spectrograph stability, will give us a factor of >3 improvement in signal. The
exposure times can be increased to four hours. Thus, it should be possible to derive
statistical ages for two fields per night for all objects down to Bj ~18. At present,
there are ~ 33,000 2dFGRS galaxies with Bj ~ 18 and a quality index of 4-5, thus
perhaps ~40,000 galaxies when the survey is complete. Thus, ~ 60 nights would be
required for this project.

In terms of blue vs. red performance, we would argue for <400nm over >800nm. The
presence of [OII] 372.7 nm emission is by far the best diagnostic for the presence of
filling in by weak emission lines; [OIII] 500.7 nm is a poor substitute. It may also be
possible to detect the higher order Balmer series out to H8 (388.8nm) and
neighbouring CN lines (our magnitude cut lowers the median redshift of the survey to
about z = 0.03.)

QSO Programs (Brian Boyle)

Digital databases (e.g. VISTA, SDSS) will form much of the basis for input catalogue
selection in QSO science programs to be undertaken with AAOmega. For QSO work,
AAOmega should maintain or extend 2dF capability in the following key areas:

Wide-field/High Efficiency (the AAOmega advantage!): AAOmega will still offer
the widest field-of-view available of a multi-object spectrograph on a 4-6m class
telescope. A key factor here is matching the QSO surface density at B ~ 22.5 (100/sq.
deg) to the available fibre surface density. Thus, QSO surveys at B ~ 22.5 (1.75 mag
deeper than any existing large-scale QSO survey) will be done more efficiently with


                                         15                            Science Drivers
                                                                        AAOmega CoD


AAOmega than any other survey instrument. The wide field also enables AAOmega
to carry out the most efficient search over large areas for intrinsically rare objects (z
>4 QSOs). The clustering and evolution of QSOs at fainter luminosities and higher
redshifts than probed by either the current 2QZ or SDSS QSO surveys is a key
program ideally suited to the wide- field and high throughput of AAOmega.

Pan-Chromicity: Given the wide range in redshifts exhibited by QSOs, it is
important that AAOmega deliver high efficiency over as wide a range as possible.
For a spectral range limited to > 4000 Angstrom, the strong UV lines of CIV and Ly
will be inaccessible for over half the QSOs in a sample to a typical magnitude limit of
B=21-22.5. Many programs are likely to focus on the distribution of Ly absorption
lines or metal line absorbers (CIV) from systems lying in front of QSOs. In these
cases it will be important to both maximise the number of objects (and thus the cross-
correlation signal) in which these absorption lines can be studied, in addition to
increasing the reredshift range over which such systems can be studied. Extending the
range down to 3700 Angstrom would increase the surface density of objects in which
Ly is visible by almost 40%, effectively doubling the number of pairs of systems for
cross-correlation work.

Resolution: In addition to the blue response, AAOmega can explore a unique area of
parameter space for a wide- field MOS by extending to resolutions higher (R >4000)
than can currently be offered by 2dF or the SDSS. At these resolutions new science
becomes possible; detailed studies of emission line profiles in unprecedently large
samples of QSOs, and equally comprehensive studies of metal- line absorbers both
from existing catalogues of QSOs (e.g. 2QZ, SDSS).

IFU: The IFU provides the opportunity to conduct new observations of QSOs at low
redshift identified as part of the 2QZ and SDSS QSO surveys. A comprehe nsive
study of the host galaxy properties of a wide variety of low redshift QSOs (including
rare/unusual classes identified as part of the 2QZ starburst/QSOs, Type II AGN, BL
Lacs) would place powerful new contraints on the QSO phenomenon and its relation
to star formation in galaxies.

2.2.2 Stellar Science (Terry Bridges)

There are many stellar projects where the higher spectral resolution and S/N of
AAOmega will make a huge difference. The higher resolution is required to achieve
higher radial velocity precision, and for projects where line profiles are important.
With the current 2dF, a typical value for the best velocity precision available is ~ 5
km/sec. With a spectral resolution double that of 2dF, plus stable, bench-mounted
spectrographs, AAOmega should achieve a velocity precision of 2-3 km/sec.
AAOmega will also have a throughput of 3-4 times that of 2dF. AAOmega will open
up entirely new areas in the detailed study of stellar populations in Local Group
galaxies and beyond.

The Structure and Formation of the Galaxy

We propose a large survey of Galactic structure, similar in spirit to that being
currently carried out by Gilmore, Wyse, and Norris with 2dF (the Anglo-Australian


                                          16                            Science Drivers
                                                                       AAOmega CoD


Old Stellar Populations Survey, or AAOSPS), but with larger numbers, e xtending to
fainter stars and larger distances, covering a larger fraction of the sky, and with a
more homogenous input catalogue from the Sloan Digital Sky Survey (SDSS).

There are two main competing models for the formation of the Milky Way. Eggen,
Lynden-Bell, & Sandage (1962) proposed a monolithic collapse model, whereby the
protogalaxy underwent a rapid collapse, forming first the halo, then a rotationally-
supported disk. An alternative picture was put forward by Searle & Zinn (1978), who
proposed instead that the halo component of the Galaxy formed in a more chaotic
manner with accretion of subgalactic fragments over longer timescales.
Observational work since Searle & Zinn has indeed found evidence for age-spreads
amongst Galactic halo stars and globular clusters, and possible halo substructure.
Recent numerical work has shown that the accretion of smaller galaxies by the Milky
Way will lead to detectable tidal streams of accreted stars; such models for the
formation of the Galactic halo have been dubbed spaghetti models (e.g. Morrison et
al. 2000). Indeed, both the AAOSPS and the SDSS have detected possible
substructure/tidal debris (Gilmore & Wyse 2001, astroph/0104242; Yanny et al.
2000).

Our proposed survey will allow us to determine the ages, abundances, and kinematics
for large numbers of stars, allowing in turn the clear delineation between the Galactic
bulge, disk, and halo. We will also be able to determine if a Galactic thick disk really
exists. The wide- field and high multiplexing of AAOmega will also allow us to
recover the merger history of the galactic halo from the mapping of tidal streams from
accreted stars. To test the competing models, we require spectra for large samples of
representative (main sequence) stars extending out as far as possible into the Galactic
halo, as free from selection bias as possible. All studies to date have fallen short of
these criteria, having either significant selection bias, not going far enough out in
distance, or having sample sizes that are too small (for a good summary of work to
date, see Beers 2001, astroph/0104275).

The combination of AAOmega and the SDSS will allow us to improve tremendously
over previous work. The SDSS, with its excellent multi-band photometry, will isolate
tens of thousands of main sequence stars, extending out to at least 50 kpc into the
Galactic halo, to a limiting magnitude of g (~ B) ~ 21 (e.g. Yanny et al. 2000,
astroph/0004128). To illustrate what is possible, we propose to double the sample of
10,000 spectra that will be obtained with the AAOSPS 2dF study. Their magnitude
limit is V=19.2, and they are able to reach a S/N of 20 (the minimum required for
such analyses) in 20,000 sec (5.5 hours). With the improved throughput of
AAOmega, conservatively estimated at a factor of three higher than 2dF, we should
be able to go slightly fainter to V ~19.5 (B=20.0-20.5) in four hour integrations. We
can thus do two fields per night, and with ~ 350 objects/field, obtain high-quality
spectra for ~700 stars/night. To obtain spectra for 20,000 stars will then take between
25-30 nights. With a spectral resolution of ~ 6000, we will achieve velocities with a
precision of better than 5 km/sec. The wavelength range required is from 3700-5000
Angstrom.




                                         17                            Science Drivers
                                                                      AAOmega CoD


Other Projects

In addition to the above survey, there are many other stellar projects which would
make good use of AAOmega. Here we list a few:

   Galactic Globular Clusters :

-- Large samples of high-precision velocities will allow detailed mapping of the
velocity field across the whole cluster (e.g. Gebhardt et al 2000, AJ, 119, 1268).

-- The SDSS has found tidal tails around Pal 5 from commissioning photometry.
Spectroscopic followup will reveal whether there are radial velocity gradients in these
tidal tails, and tell us more about the nature of cluster mass loss, as well as to shed
light on the formation history and structure of the galactic halo.

-- The high spectral resolution of AAOmega will finally allow a search for binary
stars on the main sequence in galactic clusters, where an accuracy of better than 5
km/sec is required. Higher spectral resolution and better sky subtraction will also
make it feasible to measure abundances of weak features (e.g. Na) in main sequence
cluster stars.

   Velocity Mapping of Local Group Dwarfs:

With their low velocity dispersions (< 20 km/sec), the kinematic study of nearby
dwarf galaxies requires velocities to better than 5 km/sec. The kinematics, in
combination with numerical simulations, will reveal whether these dwarfs are dark-
matter dominated or in unsettled dynamical states.

   Planetary Nebulae and Globular Clusters as Dynamical Tracers in Exte rnal
    Galaxies

PNe and globular clusters are very useful dynamical probes of early-type galaxies,
allowing us to determine the amount of dark matter in these galaxies and the nature of
the tracer orbits (e.g. Bridges 1998, astroph/9811136). Higher spectral resolution will
help with intrinsically sharp emission lines such as the OIII doublet in PNe. These
emission lines are unresolved, but the higher resolution allows them to be seen more
easily against the sky background. With globular clusters, higher precision velocities
will be obtained, and the higher S/N will help considerably. AAOmega can be used
to get spectra of PNe and globular clusters out to the Fornax distance; the 2 degree
f.o.v. and multiplexing of AAOmega will allow study of the galaxy dynamics out to
large radius, complementing multi-slit spectroscopy closer to the galaxy centres.




                                         18                            Science Drivers
                                                                         AAOmega CoD




3 Scientific Requirements for Software

In this section we discuss the upgrades to the configuration and data reduction
software which are necessary for the optimum use of AAOmega, in particular Nod &
Shuffle and beam-s witching observing modes.

3.1 Field Configuration Software
Configure is the software used to assign (multi-object) fibres to objects. There are
several modifications which are necessary for AAOmega (many of these would also
benefit 2dF in advance of the AAOmega spectrographs):

   Obtaining the same configuration on both plates: During a long integration,
    differential refraction over the 2 degree field will cause an ever- increasing fraction
    of light to be lost from object fibres, especially for objects near the edge of the
    field. Even a field near the meridian can only be observed for 5-6 hours before
    the losses become unacceptable. The solution is to have the same configuration
    on both field plates, so that the field can continually be reconfigured on one plate,
    whilst observing on the other. In this way, one could observe the same fie ld for
    indefinite amounts of time (e.g. several nights); with sky subtraction possible with
    Nod & Shuffle and beam-switching, such long integrations become feasible,
    since one is essentially limited only by Poisson noise. However, it is very difficult
    to obtain the same configuration on both plates with Configure, and work will be
    needed to make this a routine operation.

   Beam-switching and Nod & Shuffle: With beam-switching, one allocates a pair
    of fibres for each object, with each pair having a fixed separation (both in
    direction and magnitude of offset). During observing, one of the fibres in each
    pair (A) is on object and one (B) is on sky; after a period of time (typically ~15
    min) the telescope is offset so that fibre B is on the object, and fibre A on sky.
    Currently, it is a very laborious process to produce fibre configurations for beam-
    switching. Configure needs to be upgraded so that pairs of fibres can be assigned
    to each object, with an optimal separation determined so that as many objects as
    possible can be `beam-switched‘.

With beam-switching and Nod & Shuffle, there is a further difficulty of ensuring that
as many objects as possible have uncontaminated offset sky positions; this may in fact
be the main limitation to the use of these sky subtraction techniques. Configure
should be upgraded so that an optimum telescope offset (direction and magnitude) can
be determined, so that as many objects as possible have clean offset-sky positions.

   Configure Algorithm: There are some aspects of the Configure logical structure
    which need to be improved. In particular, the default fibre allocation procedures
    don‘t work well with fields which are centrally concentrated (e.g. Galactic
    globular clusters; globular clusters and planetary nebulae in other galaxies).




                                             19     Scientific Requirement for Software
                                                                       AAOmega CoD


3.2 Data Reduction Software
2dfdr is the 2dF data reduction software, which would be modified for use with
AAOmega. For many datasets 2dfdr can be run very easily as a nearly completely
automated pipeline. However, 2dfdr has largely been driven by the needs of the
2dFGalaxy and QSO Surveys, which are mainly aimed at determining redshifts from
data taken at low spectral resolution. It was also not written with Nod & Shuffle and
beam-switching in mind (though the current version of 2dfdr does handle Nod &
Shuffle data). 2dfdr will also have to incorporate both MOS and IFU observing
modes. We now list several improvements which are required to obtain the best
possible data quality from AAOmega; most of these apply for both MOS and IFU
modes.

   Beam-Switching/Nod & Shuffle: Reduction of beam-switched data needs to be
    incorporated into 2dfdr as a standard mode. It is not obvious what is the best way
    to handle beam-switched data, as it involves the reduction of two frames
    together. Work will be needed to modify the code to identify the two frames that
    are part of a beamswitched pair, and to determine the best way to actually reduce
    the beamswitched data. Some more work is also needed to allow fully automatic
    reduction of Nod & Shuffle data.

    It is inevitable that some objects observed in beam-switched or Nod & Shuffle
    mode will have contaminated sky positions. 2dfdr should be able to check for
    such contamination, and determine the optimal sky-subtraction mode for each
    object in a given field (for instance, objects with contaminated sky positions could
    be reduced using mean-sky sky subtraction, while the majority of objects would
    be reduced using beam-switched or Nod & Shuffle sky subtraction).

   PSF-Mapping: It has recently been found that 2dF sky subtraction can be
    compromised because of PSF variations across the detector. This is especially
    noticeable in data which have been throughput-calibrated using night sky lines,
    and for objects near the edge of the detector. New a lgorithms written by Scott
    Croom have now been added to 2dfdr to greatly improve the sky subtraction for
    such objects. However, PSF variations across the detector are still limiting the
    accuracy of sky subtraction with 2dF in the mean-sky method (as they will with
    AAOmega). What is needed is to map the PSF across the detector, and match the
    spectral resolution of the mean sky spectrum to that of the object spectra at any
    given location in the detector, before object spectra are sky-subtracted. This will
    give the optimal ―mean-sky‖ sky subtraction.

   Flux Calibration: While absolute flux calibration is nearly impossible with
    fibres, it is still possible to carry out relative flux calibration. 2dfdr could be
    upgraded to allow such flux calibration to be done automatically, either using a
    flux standard observed with the same observing setup, or using an AAOmega
    throughput calibration obtained elsewhere.

   CCD Processing: Since there may be two CCDs, there will be additional data
    reduction requirements. We are assuming that two CCDs would be handled as
    independent images by 2dfdr (treating the two CCDs butted together as a single
    image would present many problems and is probably unnecessary). However, we


                                           20      Scientific Requirement for Software
                                                                       AAOmega CoD


    will want to put together the red and blue sections of the spectrum after
    processing. If this is to be done automatically, there needs to be some way for the
    software to recognise which pair of images belong together.

   Tramline Mapping: Currently 2dfdr determines the tram line map for 2dF using
    a rather complex algorithm, which incorporates the full optical model for 2dF and
    ray traces from the slit to the CCD to predict image positions on the CCD. For
    SPIRAL and 6dF a much simpler algorithm has been used, which searches for
    peaks along the centre of the image, adds a curvature based on a simple radial
    distortion model and then rotates to match the data. Each approach has
    advantages and disadvantages. However, whichever approach is chosen, there
    will be work needed to adapt it to AAOmega. There are also improvements which
    could be done to data reduction. For instance, if the AAOmega slits contain gaps,
    these would make it easier to correctly determine the fibre numbering.

   Scattered Light Subtraction: The current 2dF algorithm makes use of ―dead‖
    fibres to provide points to which a background level can be fitted. This is
    unsatisfactory, since it means that if there are not sufficient broken fibres, good
    fibres have to be deliberately masked out for this purpose. Again, if the fibre slit
    contains a number of gaps, these can be used as background points (this has been
    done for 6dF). Modifications to the background- fitting code to use such gaps
    would be required.

   Wavelength Modelling: The software has to be able to determine the wavelength
    of every pixel to provide a starting point for the arc identification routines.
    Although the central wavelength prediction doesn't have to be accurate, a
    reasonably good prediction of the dispersion is needed such that pixel
    wavelengths will be within 1 or 2 pixels. Work will be needed to determine the
    best approach for wavelength modelling with AAOmega.

   Speed: 2dfdr is going to be a lot slower handling 2k x 4k or 4k x 4k images,
    compared to the current 1k x 1k (2dF) or 1k x 2k (SPIRAL). We need to ensure
    that we have machines which have sufficient speed and memory to cope.

3.3 Other Software Requirements
   Real-time reconstruction of SPIRAL IFU image for image acquisition, with
    telescope offsets sent directly to telescope. Note: it may be possible to use the
    AAT Cassegrain acquisition and guidance unit to provide offset autoguiding,
    reducing the ‗real time‘ element of the software to that of image acquisition
    checking.
   Improved S/N calculators for both MOS and IFU modes
   Software Effort: there is a considerable amount of software development
    required for AAOmega, in particular for data reduction, as well as ongoing effort
    required for software support and development. This is a critical issue, as
    AAOmega depends very much on software as it does on hardware



.


                                           21      Scientific Requirement for Software
                                                                     AAOmega CoD




4 Sky Subtraction

4.1 Sky Subtraction Methods
In order to improve sky subtraction performance, there are three main sky subtraction
techniques possible with AAOmega (most of the discussion below will centre on
MOS mode):

   Mean-Sky Method (MSM), where a number of fibres (typically 20-30) are
    dedicated to sky.

   Cross Beam-Switching (CBS), where the telescope is nodded between object and
    sky on timescales of typically 10-20 minutes, with CCD readout between nods.

   Nod & Shuffle, which combines nodding/beam-switching with charge shuffling
    on the CCD.

4.1.1 Mean-Sky Method:
MSM is the standard sky-subtraction technique for 2dF, and is the most
straightforward. After the sky fibres have been throughput-calibrated, they are
averaged to form a mean sky spectrum, which is then subtracted from each object
spectrum. Throughput calibration can be done either from offset sky exposures (3-5
 5 minute exposures taken with the telescope offset by a small amount), or by using
the flux in night sky lines; both methods have similar effectiveness.

Advantages:

   The Poisson sky noise is reduced to low levels by averaging over all the sky
    spectra
   MSM allows the highest number of objects (up to 370-380) to be observed.

Disadvantages:

   Sky and object are not observed through the same fibre, whic h can introduce
    systematics due to errors in fibre throughput and PSF variations across the CCD.

4.1.2 Cross Beam-Switching:
CBS involves configuring the 400 fibres into pairs. Each pair of fibres observes sky
and object simultaneously, and in addition, there is the same vector displacement
between the sky and object for each pair. After an exposure, the CCD is read out, and
the telescope is slewed (nodded) so that the sky fibres become the object fibres and
vice-versa. In other words, 200 objects are observed in each exposure, but through
different fibres in alternating exposures. Sky-subtracted spectra are calculated for
each fibre. For each pair of fibres, the spectra are averaged, which has the effect of


                                         23                           Sky Subtraction
                                                                        AAOmega CoD


removing systematic noise to first order.      An illustration of CBS is given in Figure
4.1.




Figure 4.1: Illustration of Cross Beam Switching

Advantages:

   Takes into account (slow) variations in the sky background, and sky subtraction in
    principle should be better than with MSM because the sky and object are observed
    through the same fibre..
   Objects are observed continuously, unlike Nod & Shuffle, where objects are
    observed for only half the time.

Disadvantages:

   Only ~200 objects can be observed, as half the fibres are needed for sky.
   Poisson noise higher than with MSM.
   Software becomes more complicated: to determine best fibre-pair positions, and to
    allow automated reduction of beam-switched data.
   There are significant overheads associated with the offsetting and with CCD
    readout.

4.1.3 Nod & Shuffle
Nod & Shuffle involves observing objects and sky through the same fibres. Between
successive exposures on object and sky, the telescope is nodded at the same time as
charge is shuffled on the CCD; object or sky counts are stored on different areas of
the CCD. The process of nodding and charge shuffling can be performed at much
higher rates (typically on ~60 sec timescales), since the readout only needs to be done
once at the end of the observation. Thus, temporal sky variations between sky and
object spectra can be reduced to negligible levels. Nod & Shuffle has been used
routinely for some time with other AAO instruments such as Taurus and LDSS++
with great success. In principle, with enough CCD space, all available fibres can be
used with Nod & Shuffle; in practice, AAOmega will only have 4k of spatial pixels,
allowing only 200 fibres to be used with Nod & Shuffle. For further information
about Nod & Shuffle, see these AAO Newsletter articles: Bland-Hawthorn & Barton
(No. 75, October 1995); Glazebrook (No. 87, November 1998); Cannon (No. 96,


                                          24                            Sky Subtraction
                                                                      AAOmega CoD


February 2001), and a recent article by Glazebrook & Bland-Hawthorn (PASP, 113,
197, 2001; discusses Nod & Shuffle in the LDSS++ context).

Advantages:

   In principle, Nod & Shuffle offers the best sky subtraction of all, limited only by
    object Poisson noise.
   No additional CCD readout overheads
   No special configuration software requirements

Disadvantages:

   Can only observe ~1/2 the number of objects as can be done with MSM (because
    AAOmega will have only 4k of spatial pixels)
   Half the time is spent on sky, so more inefficient than CBS or MSM
   Physical masks required to mask off half the fibres
   Sky Poisson noise higher than with MSM

4.1.4 Multiplex Summary
Table 4.1 gives a summary of the multiplex performance of the three methods with
AAOmega. MSM has about twice the multiplex advantage of CBS and four times
that of Nod & Shuffle, defined to be the ―number of objects‖ times ―the fraction of
time observing objects‖. The sky-subtraction factor, Fss, is equal to 1 + 1/N s, where
Ns is the number of sky fibres averaged for the sky subtraction. Thus, the reduction of
sky residuals must be important for CBS or Nod & Shuffle to be considered.

      Method             No. of Objects         Fraction of time     Sky-subtraction
                                               observing objects       factor (Fss)

    MSM                       380                    0.95                 1.05
     CBS                      200                     0.9                  2.0
NOD & SHUFFLE                 200                    0.45                  2.0

Table 4.1: Showing the multiplex performance of the different sky subtraction
methods.

4.2 Sky Subtraction with 2dF and SPIRAL

4.2.1 Mean Sky Method
4.2.1.1 2dF
As mentioned above, the MSM has been used for almost all 2dF observing, save for
tests of Nod & Shuffle and CBS (described below). The best sky subtraction
accuracy obtained with the MSM with 2dF is ~1% (defined in 2dfdr as the rms of the
throughput-calibrated sky fibres), with more typical values of 3-5%. It has only
recently been realized that 2dF sky subtraction using night sky lines for throughput
calibration (the method used for the 2dF Galaxy and QSO surveys for instance) was
being compromised by PSF variations across the detector, principally at the detector
edges. New routines for night sky-line throughput calibration have recently been


                                          25                          Sky Subtraction
                                                                     AAOmega CoD


implemented by Scott Croom and Jeremy Bailey, with improvements to 1-2% seen in
sky subtraction accuracy. PSF variations across the detector and uncertainties in the
relative fibre throughputs are probably the main limitations on the accuracy of sky
subtraction using the MSM. PSF- mapping to match the mean sky spectrum to that of
object spectra before sky subtraction would probably allow sky subtraction accuracy
of 1-2% routinely.

4.2.1.2 SPIRAL
We have more limited experience with SPIRAL sky sutraction. However, in
December 2000, tests of the MSM and Nod & Shuffle were carried out with SPIRAL.
The SPIRAL implementation of the MSM uses a subset of the outlying fibres as sky,
and uses them, after throughput calibration, to remove sky from the object. It was
found that scattered light and dark current structure made accurate sky subtraction
difficult. Additional problems result from PSF variations and errors in wavelength
calibration. The typical accuracy of sky subtraction with SPIRAL MSM is 2-4%, but
could be significantly improved with better PSF and wavelength mapping.

4.2.2 Cross Beam Switching
Beam-switched data with both 2dF and SPIRAL were taken in December 2000 and
January 2001. With SPIRAL, this simply entailed moving the target between the two
halves of the rectangular field of view, while with 2dF it involved creating a special
configuration with pairs of equally-spaced fibres for each target. Beam-switching has
been implemented with both instruments, but the data have not yet been fully
analysed.

In July 2001, beam-switching was done with 2dF during a AAO/Sloan collaboration,
using the 270R/316R gratings. The project involved looking for quasars in two
redshift ranges: z~3, and z~5, and was a great success, finding 3 z~5 QSOs and over
40 z~3 QSOs 1 . For each field, the following sequence was followed: Obj, Sky, Obj,
Obj, Sky, Obj, where each object/ sky exposure was 15 minutes. Thus, one hour was
obtained on objects, and half an hour on sky for each field; no object frame was more
than 15 minutes from a sky frame. Figure 4.2 shows the mean sky residuals for
beam-switched reduction and for the standard 2dfdr MSM method; it can be seen that
the sky-subtraction accuracy in the red is improved from ~4% to~1%. Thus, with
careful reduction of beam-switched data, sky subtraction accuracies of ~1% can be
obtained.




1
    AAO Newsletter No 98, September 2001, Glazebrook et al.


                                         26                           Sky Subtraction
                                                                            AAOmega CoD




Figure 4.2: Comparison of beam-switched data reduction (red, lower points) and standard
2dfdr MSM reduction (black, upper points) for a 2dF Sloan field. In the red, the sky residuals
are ~4% for MSM reduction, compared to ~1% for beam-switched reduction.


4.2.3 Nod & Shuffle
4.2.3.1 SPIRAL
Nod & Shuffle with SPIRAL involves masking half of the IFU, and nodding the
telescope between the target and sky, as the charge on the CCD is shuffled between
the masked and exposed regions. Object and sky spectra are thus observed through
the same fibres. Nod & Shuffle with SPIRAL was carried out in December 2000 and
January 2001, with excellent results. There was virtually perfect cancellation of night
sky lines across the entire CCD frame (limited by photon statistics), and problems
with scattered light were almost completely avoided. 2




2
    Cannon & Corbett, AAO Users‘ Committee Document, June 2001


                                            27                              Sky Subtraction
                                                                      AAOmega CoD



4.2.3.2 2dF
Nod & Shuffle with 2dF was first carried out in July 1999 by Glazebrook et al. (see
AAO Newsletter No. 90, August 1999). Nod & Shuffle with 2dF can only be done
using one CCD (with AAOmega and the new AAO-II controllers, Nod & Shuffle
could be done with two CCDs over all the detector real estate), and a mask has to be
installed on the fibre plate to block off alternate retractors (to create unilluminated
areas which can then be used to store the charge during a shuffle exposure). Although
no objects could be observed due to bad weather, excellent sky subtraction (~0.1%) of
moonlit cloud was achieved, as seen in Figure 4.3.




Figure 4.3: Sky Subtraction with 2dF Nod & Shuffle. (a) Raw sky spectrum, before
subtraction; (b) After standard MSM reduction; (c) after Nod & Shuffle reduction. Taken
from Glazebrook et al. (AAO Newsletter No. 90, August 1999).

These first observations showed that Nod & Shuffle was feasible with a fibre system,
but further tests were required on real objects. These were obtained in January 2001.
Data were taken for two fields; we will discuss results for one of these fields (a deep
SUMSS radio field). For this field, data were taken in both CBS and Nod & Shuffle
modes; several fibres were also placed on sky to allow standard MSM data reduction.


                                         28                            Sky Subtraction
                                                                      AAOmega CoD


Note that these data have not yet been fully analysed, and the results presented here
should be considered preliminary. Further tests of Nod & Shuffle with 2dF in dark
time will be required, as the January time was almost all bright, not allowing deep
observations of faint objects where Nod & Shuffle might be expected to come into its
own.

The observing itself went very smoothly, with no problems during Nod & Shuffle
observing. Jeremy Bailey was able to write a new version of 2dfdr which allowed
virtually fully automatic reduction of Nod & Shuffle data, a huge help. Thus, we
confirm the feasibility of Nod & Shuffle with 2dF, and by extension AAOmega. The
reduction and analysis of these data have been started with the help of a UK AAO
Winter Student (Parimal Patel). We have attempted to directly compare sky
subtraction with MSM and Nod & Shuffle, by looking at sky-subtracted sky spectra
(ie. those fibres which were dedicated to sky).

For the SUMSS field, we obtained two 900 sec Nod & Shuffle frames with the moon
down, and two more 900 sec Nod & Shuffle frames with the moon up. We have
reduced each of the four frames using both the Nod & Shuffle and MSM methods.
For each frame and for each data reduction method, we take the ratio between the sky-
subtracted sky fibres and the raw (unsubtracted) sky at each wavelength. We then
take the median over all (~25) sky fibres, again for each frame and for each data
reduction method. The results to date are somewhat puzzling, in that we have not
found that Nod & Shuffle gives better sky subtraction than the MSM, for the dataset
analysed. Further analysis to confirm these results is required; for instance, we are
not yet certain that the Nod & Shuffle data reduction is being carried out optimally
within 2dfdr. We also require additional test data from more fields. More detailed
results will be fully presented as part of the Preliminary Design Review.
4.2.4 Summary
With some upgrading of 2dfdr, mean-sky reduction should be able to achieve sky
subtraction accuracies of 1-2%. Careful reduction of beam-switching data, as shown
by the recent Sloan project, can achieve a sky subtraction accuracy of ~1%. Thus,
these two methods would probably be used for the vast majority of AAOmega
programs. Further effort on Configure will be required to make it easier to produce
beam-switched configurations. Further analysis of the January beam-switched data
will also be carried out, as beam-switching may is potentially as effective as Nod &
Shuffle for most applications.

The conclusions for Nod & Shuffle in MOS mode are not so clear. Preliminary
analysis of the January data does not show the expected sky subtraction gains with
Nod & Shuffle, for one dataset at least. However, earlier tests showed that very good
sky subtraction can be achieved with Nod & Shuffle. Further analysis of the January
data is ongoing, and further tests of Nod & Shuffle under better conditions are
required, for a larger number of fields. In theory, Nod & Shuffle should give the best
sky subtraction in the red, where there are many night sky lines, which are variable on
short timescales. Nod & Shuffle observing with AAOmega will always have the
disadvantages of only being possible with 200 fibres, and of requiring half the time to
be spent on sky. These are big overheads, and Nod & Shuffle will likely be reserved
for programs that are looking at very faint objects in the red, and which require the
best possible sky subtraction. If Nod & Shuffle observations are indeed only limited



                                         29                           Sky Subtraction
                                                                      AAOmega CoD


by object Poisson noise, very long Nod & Shuffle exposures are possible. In practice,
a significant limitation may be the contamination of offset sky positions (also a
problem for beam-switching).

With SPIRAL data reduction as it stands, Nod & Shuffle appears to give the best sky
subtraction performance, which is expected also to hold true for AAOmega IFU
observations. The current MSM used with SPIRAL does not sufficient account of
PSF variations and wavelength shifts on the detector. Improvements in PSF and
wavelength mapping could significantly improve MSM sky subtraction performance.
We expect further tests to show more clearly the gains with Nod & Shuffle for MOS
observations. Clearly, it is worth investigating Nod & Shuffle as an observation mode
for AAOmega. Aside from the additional software effort required for fully optimized
Nod & Shuffle (and beam-switching) data reduction, Nod & Shuffle capability does
not have major implications for this Concept Design study, nor is it driving either the
transmissive or DBSS optical designs. Further investigation of sky subtraction for
both MOS and IFU modes will be carried out for the Preliminary Design Review.




                                         30                           Sky Subtraction
                                                                     AAOmega CoD




5 Functional Specification

Version 1.0 David Lee and Ian Lewis 21 March 2001
Revision 1.1 Roger Haynes 07/08/01
Revision 1.1.1 Roger Haynes 09/08/01
Revision 1.1.2 Roger Haynes 09/08/01

5.1 Introduction
AAOmega is a new instrument being developed by the Anglo-Australian Observatory
for use on the 3.9 m Anglo-Australian Telescope. AAOmega will replace the existing
2dF spectrographs to provide a multiple-object observing capability with improved
efficiency and spectral resolution. In addition AAOmega will enhance 2dF‘s
capabilities by providing single-object spectroscopy with an integral field unit to be
mounted at Cassegrain focus. Thus, when the 2dF top end is not mounted on the
telescope, AAOmega will continue to provide single-object observing capability using
the existing SPIRAL integral field unit which will be permanently available at either
straight through or auxiliary Cassegrain focus. In this document we present a
description of the scientific functionality to be provided by the AAOmega instrument.

5.2 Existing 2dF infrastructure and upgrades
It is assumed that AAOmega will use the existing 2dF infrastructure including the
prime focus optical corrector, focal plane imager, and upgraded robotic fibre
positioner. In addition AAOmega will provide the following functionality:

      Ability to mask unused MOS fibres for Nod & Shuffle, mask installed by hand
       at focal plane (input to fibres).
      Software readable mask interlock to avoid damage to fibres.
      Replacement field plates to provide improved positioner performance.
      Upgraded gripper Z drive.
      Refurbished retractor units with new components, e.g. pulleys, to improve
       reliability.

5.3 Multiple-object spectroscopy (2dF mode)
The functionality provided by AAOmega in multi-object mode will be similar to 2dF
and can be summarised as follows:

      400 multiple-object fibres, of 2.1 arc-seconds diameter.
      Resolving power 1000-6000 (Goal >10,000).
      Nod & Shuffle capability with 200 fibres. Alternate retractors, containing 10
       MOS fibres, will be masked at the input end to provide blank areas on the
       CCD for charge shuffling.
      Total system throughput >15% (goal >20%) top of atmosphere to detected
       photons.



                                          31                 Functional Specification
                                                                      AAOmega CoD


      Autoguiding capability using 4 guide fibre bundles and a new CCD
       acquisition system.
      Fibre separation at detector ~ 10 pixels.
      Fibre Crosstalk: Assuming profiles of equal intensity, the crosstalk point
       between adjacent spectra should be <1.0% of the peak value.
      Sky-subtraction accuracy better than 1% [rms] with dedicated sky fibres.
      Sky-subtraction accuracy better than 0.1%[rms] with Nod & Shuffle.
      Uses existing 2 degree field of view.
      Uses existing fibre button and retractor design.
      Fibre sampling at detector of 4.7 pixels FWHM.
      New Heraeus optical fibres for optimal fibre transmission from 370 to 950 nm.

5.4 Integral Field Unit (Cassegrain)
AAOmega will provide single-object spectroscopy at Cassegrain using the existing
SPIRAL IFU fibre feed. This observing mode will only be available when the f/8
Cassegrain focus is in use. The SPIRAL IFU will normally be mounted at the
auxiliary Cassegrain focus. If spectropolarimetry is required then the IFU would need
to be mounted at the normal Cassegrain focus.

      Available at auxiliary Cassegrain port simultaneously with other Cassegrain
       instruments or at normal Cassegrain to the exclusion of other instruments.
      Uses existing SPIRAL IFU with appropriate slit modifications.
      Field of view 22 x 11 arc-seconds with 512 spatial elements.
      Spatial sampling 0.7 arc-sec.
      Spectrophotometric accuracy 1%.
      Spectropolarimetry available only when mounted at straight Cassegrain (to
       avoid the additional reflection of the auxiliary folding flat mirror).
      Nod & Shuffle available with the full field of view.
      No input filters.
      Acquisition and guiding via existing AAT infrastructure.
      Total system throughput >12% (goal 15%) from top of atmosphere to detected
       photons.
      Field rotation via Cassegrain rotator.

5.5 Spectrograph
AAOmega will consist of a new spectrograph and associated fibre feeds. The
functionality provided by the spectrograph will be as follows:
     Thermally stable enclosed environment.
     Bench mounted for stability. The location of the spectrograph will be on the
       South Catwalk of the telescope. A 27 m length of fibre will be required to
       prime focus.
     4k x 4k pixels detector (15 micron pixels assumed), possibly two 2k x 4k
       devices.
     AAO-2 CCD controller.
     Detector cooled to between 150-170K.
     Filters (clear plus up to three order sorting filters, as for 2dF, if required).
     Hartmann Shutter(s) for focusing


                                          32                 Functional Specification
                                                                      AAOmega CoD


     Articulated camera (0—90 degrees).
     Volume Phase Holographic gratings (VPHG's) for medium and high
      dispersion.
     Transmission gratings, grisms, or VHPG's for low dispersion.
     MgF2+Solgel Anti- Reflection coatings on transmissive optics.
     F/3.15 collimator
     Camera, to give full coverage of 400 MOS fibres at 10 pixel centre to centre
      spacing.
     At least a 150mm collimated beam size.
     Remotely controlled slit unit for slit interchange.
     Spectrograph fed by 400 MOS fibres (2 fieldplates = 800 fibres) or 512 IFU
      fibres.
     Cassegrain IFU.
     Wavelength coverage 370-950 nm.
     Long slit with lamp for detector flat fielding.
     Calibration facilities (arcs and fibre flat field lamps) as currently provided by
      2dF.

5.6 Instrument control software
     Full software control of spectrograph configuration from 2dF control system.
     Spectrograph configuration calculator to determine optimum grating angles
      and articulation angle.
     Full spectrograph information in FITS header for all data.

5.7 Data reduction software
     Adaptation of existing 2dF data reduction software for AAOmega MOS data,
      incorporating existing specialised versions for IFU data, and multi-object Nod
      & Shuffle.
     Real time reconstruction of IFU image for image acquisition (telescope offsets
      sent directly to telescope). Note, real-time reconstruction is not required when
      using the AAT Acquisition and Guidance Unit for offset guiding.
     Signal to noise calculator for MOS and IFU cases.
     Software to cater for additional modes of sky subtraction, e.g. beam switching.

5.8 Configuration software
     As currently provided for 2dF with updating for Nod & Shuffle and beam
      switching observing.




                                          33                 Functional Specification
                                                                     AAOmega CoD




5.9 Instrument block diagram


           2dF Prime focus                           Cassegrain focus
                                               Cassegrain
                                                                 Auxiliary
        2dF P0         2dF P1                 and optional
                                                                 Cassegrain
        400              400                  polarimetry
        fibres          fibres                  module


                                                        SPIRAL IFU
                                                      512 element array



      27m long fibre
      bundles from
      Prime Focus
                                                    18m long fibre bundles
                                                    from Cassegrain Focus




                             Slit exhange unit
                                    Filters

                             ~F/3 Collimator


                             Grating or VPHG


                                 ~F/1.5 Camera

                                   4k x 4k
                                    CCD
                                    Pixels


                   AA spectrograph on AAT south
                              catwalk




                                        34                   Functional Specification
                                                                     AAOmega CoD




6 Fibre Selection

6.1 Introduction
The selection of fibres for AAOmega is driven by the requirement to carry out Multi-
Object Spectroscopy (MOS) using the current 2dF corrector, robot positioner and
field plate system. Essentially the MOS system will remain as 2dF with the two top
end ring mounted spectrographs being replaced with a bench mount spectrograph on
the South catwalk of the AAT. An integral field unit (IFU) is proposed, which due to
budget and scheduling constraints will use the SPIRAL B fibre bundle with a
modified output slit. The fibre for the IFU is therefore already determined as the
bundle for SPIRAL B will be re-used. However, there is the opportunity to choose
different fibre for MOS observing as the fibre bundles have to be replaced with longer
fibres (nominally 27m) to be able to reach the spectrograph on the South catwalk.
Currently they are 8m long, with a 140m diameter core, corresponding to a 2.1"
aperture on the sky, with a 7% variation across the field.

6.2 MOS fibres
Taylor and Gray3 discussed the detailed modelling used in the selection of fibre for
2dF. In this a large number of parameters must be considered. A detailed modelling of
the required fibre characteristics has not yet been performed for the AAOmega
project, however many of the parameters are unchanged from the 2dF system, so the
fibre core size used in 2dF has been assumed for all the optical designs (Section 7).
Based on the S/N figures presented, it is not expected that this would need to be
changed significantly for AAOmega.

6.2.1 Fibre core size

Considerations in the selection of fibre for AAOmega inc lude:
1. Source characteristics
    Source magnitude as a function of wavelength.
    Sky brightness as a function of wavelength.
    Intrinsic source size i.e. point source (e.g. stellar), extended source (e.g.
       diffuse galaxy).
2. Atmospheric characteristics
    Typical seeing characteristics of the telescope and site.
    Atmospheric dispersion and differential atmospheric refraction as a function
       of telescope zenith angle.
    Atmospheric transmission
3. Prime focus corrector characteristics (2dF)
    Image quality as a function of wavelength and field position.

3
  "System Modelling of the 2dF", Fibre Optics in Astromony II, A.S.P Conference
Series. Volume 37, Pages 379-391)


                                          35                          Fibre Selection
                                                                        AAOmega CoD


    Field distortion as a function of wavelength and ADC angle.
    Telecentric performance
4. Spectrograph considerations
    Limiting collimator and camera focal ratios
    Beam vignetting within the spectrograph
    Spectral resolution targets
    Detector characteristics (Format, pixel size and QE vs wavelength)
5. Fibre optical characteristics
    Transmission as a function of wavelength
    Focal Ratio Degradation (FRD) characteristics
6. Positioning errors
    Robot positioner accuracy
    Astrometric accuracy of targets
    Flexure
    Telescope guiding accuracy
    Integration time

Lewis, Cannon et. al. 4 discuss the performance of some of these system elements for
the 2dF system. It is clear that, for point sources, the image size at the focal plane is
dominated by the corrector image quality and the atmosphere. Both these will also
limit the throughput performance of AAOmega in the blue.

One significant development in optical fibres since 2dF is the development of Heraeus
STU silica fibres that give enhanced overall transmission compa red with either of the
more traditional High OH (blue optimised) and Low OH (red opitmised) silica fibres.
Attempts have been made to get some sample of STU fibre to carry out FRD tests, but
it appears to be available only as a custom draw, costing over $40K, virtually
independent of the length ordered. However, it is not expected to have significantly
different FRD characteristics to the 2dF fibres. FRD testing is currently being carried
out on a 27m length of 2dF fibre, taking into account the prime mirror obstruction.
This will allow for better estimates of vignetting and obscuration losses with the
spectrograph.

6.2.2 MOS fibres specification for AAOmega

      Fibre Type:                  STU Fibres
      Core diameter:               140m
      Cladding diameter:           168m
      Buffer diameter:             198m Polyimide
      Fibre length:                ~27m
      Fibre spacing at the slit:   ~300m.

A plot of the transmission characteristics of a 27m length of STU fibre is shown in
Figure 6.1




4
    "The Anglo-Australian Observatory's 2dF Facility”, MNRAS submission July 2001


                                             36                          Fibre Selection
                                                                                                                                                  AAOmega CoD


                                                                                    Transmission of a 27m length of STU fibre


                                                 100.0



                                                  90.0



                                                  80.0
  Percentage throughput (no reflection losses)




                                                  70.0



                                                  60.0



                                                  50.0                                                                                                   STU (27m)


                                                  40.0



                                                  30.0



                                                  20.0



                                                  10.0



                                                   0.0
                                                      350   400   450   500   550     600      650     700     750     800      850   900   950   1000
                                                                                             Wavelength (nm)



Figure 6.1 Transmission of a 27m length of the proposed MOS fibre (STU). This does not
include any reflection losses at the fibre ends.


6.3 IFU fibres
The AAOmega IFU, that is to mount at the AAT Cassegrain focus, is going to re- use
the SPIRAL B fibre module. This will remain unchanged other than new mounting
arrangements to the Cassegrain port and the re-terminating of the fibre outputs in a slit
unit better suited to the AAOmega spectrograph. The SPIRAL B IFU has 512 fibres,
each fed by the re- imaged telescope pupil from the square lenslets (0.7"x0.7") of a
Limo microlenses array at f/5.5 across the sides and f/3.9 across the diagonals.

6.3.1 IFU Fibre specification for AAOmega

The SPIRAL B IFU was originally designed to feed the f/4.8 SPIRAL spectrograph.
The fibre core size for an IFU is determined by

                                                                                               =  DtelFfibre

Where  is the required spatial sampling in radians on the sky (0.7" to adequately
sample the median seeing of 1.5"), Dtel is the telescope diameter of the telescope
primary mirror (3.9m) and Ffibre is the input focal ratio to the fibre (chosen to be f/5.5
in order to reduce coupling losses into the spectrograph due to FRD). This gives a
required fibre core size of 73m, However, to allow for alignment errors between the
fibres and the microlens images, the fibre core is typically selected to be oversized.
The AAOmega IFU fibres (i.e. the original SPIRAL B fibres) have the following
specification.



                                                                                                         37                                       Fibre Selection
                                                                                                                                                       AAOmega CoD



                                                    Fibre Type:                       High OH Polymicro fibre
                                                    Core diameter:                    85m
                                                    Cladding diameter:                102m
                                                    Buffer diameter:                  115m Polyimide, 200m Acrylate
                                                    Fibre length:                     18m

The focal ratio chosen to feed the SPIRAL B fibres will significantly under-fill the
collimator chosen to suit the AAOmega MOS fibres. The performance impacts of this
are discussed in Section 7. To provide the required fibre spacing at the slit (~115m)
the Acrylate buffer will have to be removed, using acetone, prior to assemble.

The transmission characteristics (based on the manufactures data sheet) for the
SPIRAL B fibre are shown in Figure 6.2.

                                                                                       Transmission of 18m length of High OH fibre


                                                   100.0


                                                    90.0


                                                    80.0
  Percentage transmission (no reflection losses)




                                                    70.0


                                                    60.0


                                                    50.0                                                                                                      High OH fibre


                                                    40.0


                                                    30.0


                                                    20.0


                                                    10.0


                                                     0.0
                                                        350   400   450   500   550    600      650      700      750      800       850   900   950   1000
                                                                                              Wavlength (nm)



Figure 6.2 Transmission of an 18m length of SPIRAL B fibre (High OH Silica). This does
not include any reflection losses at the fibre ends.




                                                                                                            38                                         Fibre Selection
                                                                          AAOmega CoD




7 Optical Design

7.1 Introduction

We have investigated two rather different optical systems for AAOmega:
   A transmissive system, which has the advantage of zero obscuration by the fibre
    slit and the detector, as well as having an external focus for the de tector, making it
    readily exchangeable. Higher throughput is to be expected from systems which
    avoid mirror losses.
   A Schmidt system which has internal foci and hence some obscuration and lack of
    flexibility in changing detectors, but which also has better imaging quality, and an
    expected cost advantage in both materials and simplicity of construction.
Both designs would use 140 m core diameter for the MOS fibres, as in the present
2dF, and 85 m for the IFU fibres. It is now intended that the IFU fibres would come
from the SPIRAL IFU at the Cassegrain focus, rather than a new prime focus IFU
incorporated into the 2dF tumbler.

Our intention with the AAOmega design is to achieve a major improvement over 2dF,
in the aspects listed below.

7.1.1 Improved throughput
    i) New Heraeus STU fibres will be used, with better transmission.
    ii) The use of Volume Phase Holographic (VPH) gratings allows a major source
         of light loss to be substantially reduced. VPH gratings have peak efficiencies
         of 80 to 90%, compared with typically 60% for the conventional surface relief
         reflection gratings.
    iii) The lens elements of an optical system can now be coated with a double layer
         of MgF2 plus Sol-gel. This cuts the reflection loss to ~0.5% per surface and
         makes highly efficient transmissive systems possible. Mirror coatings can also
         be made more efficient than the usual Al layer if they are to be used over a
         restricted waveband.
    iv) Improved CCD responsive quantum efficiency (RQE). Newer CCDs available
         to the AAO have substantially higher RQE than those used in 2dF.

    The total gain over 2dF is expected to be a factor of 2.5 – 3.

7.1.2 Larger detectors
The CCDs in the 2dF spectrographs are only 1K  1K devices. The AAOmega design
uses a total detector area of 4K  4K pixels, either as a single device or in the form of
two 2K  4K detectors. If two devices are used, they may be either butted together (in
the single-beam transmissive design) or in separate cameras (in the dual beam
Schmidt design). The larger size gives an increased wavelength range at a given
spectral resolution. It does not increase the number of fibres over that for 2dF – the



                                             39                            Optical Design
                                                                       AAOmega CoD


same 400 fibres will be used – but it does allow better separation of individual spectra
in the MOS mode.

7.1.3 Increased wavelength resolution
 For a number of projects the wavelength resolution of 2dF is too low even with the
1200 l/mm gratings. Second order operation has not proven successful in raising the
resolution significantly. AAOmega will offer substantially higher resolution,
especially with the smaller IFU fibres. Even with the MOS fibres, increased resolution
is possible because VPH gratings can operate efficiently at line densities up to and
beyond 2500 l/mm.

7.1.4 Improved spectrograph stability
The stable bench mounting of the spectrographs, compared with the top-end mounting
for 2dF, will give the benefit of better wavelength (velocity) stability, and better
night-sky line subtraction for Nod & Shuffle operation.


7.2 Single-beam system with transmissive camera

7.2.1 Origin of the concept
The transmissive system originated from the following considerations:
1. External camera foci allow detectors to be easily mounted and readily changed
   and upgraded.
2. External foci for both the collimator and camera prevent obstruction losses.
3. Transmissive systems should have a higher transmission, since mirror losses are
   avoided and lenses can now be coated with highly efficient MgF 2 +Sol-gel layers.
As noted elsewhere, it was intended to have two spectrographs, each with a 4K  4K
detector array, which would have allowed 20 pixels per MOS fibre a nd a total of
2000 IFU fibres with space for Nod & Shuffle. However, cost considerations led to
this being scaled back to a single spectrograph, giving 10 pixel MOS fibres spacing.
It is important to go as far to the blue as possible, and the wavelength range
considered here will be 370 – 1000 nm. (Note, the current specification is 370 to 950
nm.) The blue limit is imposed principally by the throughput of the 2dF corrector +
fibres.
The optical design for the single-beam system has been carried out by Damien Jones
(Prime Optics, Queensland), whose report is given in Section 7.2.3.


7.2.2 Outline of the design

7.2.2.1 Maksutov Collimator
The optical specification stated that a fully transmitting collimator was preferred, but
Damien found that better imaging quality was obtained with a system incorporating
one reflection. The first collimator design that he delivered is thus an off-axis
Houghton/Maksutov system, and is shown in Figures 7.1 and 7.2. It uses a spherical
mirror and five lenses. Details are given in Damien‘s report (section 7.2.3).



                                           40                           Optical Design
                                                                                   AAOmega CoD


The off-axis feed allows the fibre slit and the slit exchange mechanism to lie entirely
outside the beam, producing no obstruction. All surfaces are spherical, and the
transmissive elements are all of silica except one element of the doublet near the slit,
which is of CaF2 . The fibre slit is curved, but this can be accommodated since the
fibres will be in blocks of about 32 and these can be separately positioned. The focal
length of the collimator is 472.5 mm, the pupil (beam) diameter is 150 mm, and the
fibre slit length is 106.5 mm. The focal ratio is 3.15, as required to allow for Focal
Ratio Degradation in the fibres fed by the f/3.4 AAT prime focus + 2dF corrector.
(Some further tests are required to check the FRD of the required length of fibre, and
hence that f/3.15 is appropriate.)
Good imaging quality is obtained despite the off-axis operation, essentially because
the majority of the focusing power is in the (achromatic) mirror, and the lenses apply
only corrections.


           150 mm pupil dia m




           150 mm                          40 mm
                                           clearance
           pupil relief                                             Doublet lens


                                              Fibre slit (end on)




       Figure 7.1: Layout of the off-axis Houghton-Maksutov reflective/transmissive
       collimator design (H3FC301). Note that only the upper part, where examp le rays are
       shown, would be built. The lower part is shown only to clarify the off-axis nature of
       the design.
The imaging quality is specified by the rms radii of the spot diagrams, determined by
Zemax. Values are given in Table 7.1. Note that because the collimator and camera
together demagnify by a factor of 1.733 (for the camera focal length of 272.6 mm), all
the values in Table 7.1 should be divided by 1.733 in order to be interpreted as image
sizes on the CCD. For example, the value 9.83 m becomes 5.67 m which would be
the image size produced at this field position and wavelength with a perfect camera.
The specification for overall image quality was 9 m rms radius on the detector, so it
is clear that when added in quadrature to the camera aberrations, those of this
collimator will be almost negligible. The aberrations could be reduced still further by
reducing the 40 mm clearance from the slit to the edge of the beam, since this is larger
than the 25 mm minimum specified. Given that the camera was expected to be more
difficult (because it is fully transmitting, and is faster), keeping the collimator
aberrations low was a chief reason Damien initially chose this type of design.


                                                41                                 Optical Design
                                                                                      AAOmega CoD




                                                         Fibre slit




       Figure 7.2: The same collimator design as in Figure 7.1, v iewed along the
       perpendicular axis. Rays are shown from fibres at the centre and one end of the fibre
       slit, and an intermed iate position along the slit. The slit curvature is clear.


                                         RMS radius (values in m)
       (nm)             On axis                4.5739 off axis            6.4471 off axis
          370               6.63                         9.03                     11.94
          420               4.75                         7.17                      9.83
          500               4.10                         5.92                      7.86
          700               6.20                         6.40                      6.51
         1000               8.29                         7.67                      6.66
    Table 7.1: Values of RMS radius from the spot diagrams of collimator H3FC301. The off -
    axis colu mns are specified by the angle of the rays at the pupil; the extreme of 6.4471 
    corresponds to a fibre at the end of the slit. Note: these rms rad ii should be divided by 1.733
    in order to be interpreted as aberration sizes on the detector. Values have been computed in
    Zemax using dithered ray pattern of density 7, referred to the centroid.

If required, order-sorting filters can be placed between the slit and the first lens.
Provided they are tilted with respect to the axis of Figure 7.1 in such a way as to be
normal to the actual beam through them, no significant aberrations are introduced.
Regrettably, after checking the tolerances for this collimator and shortly before this
report is to be completed, Damien concluded that it would be extremely difficult to
manufacture the elements and align them in their mounting. He therefore produced a
design (described in the next section) for a fully transmitting collimator. The AAO
does not at this time regard the off-axis collimator as definitely impossible. Alignment
could be eased by making the elements sufficiently large that the optical axis can be
marked on them; and the off-axis spherical elements are equivalent to an on-axis
spherical lens with a superimposed wedge.




                                                    42                                  Optical Design
                                                                              AAOmega CoD


7.2.2.2 Fully Transmissive Collimator
As noted above, Damien Jones has provided a preliminary design for a fully
transmissive collimator. Being axisymmetric, the manufacturing concerns should be
lessened. The layout is shown in Figure 7.3.




                                                                       fibre slit




       Figure 7.3: Layout of the fully transmissive collimator (AAOM EGA -COL-204).
       Rays are shown from fibres at the centre and end of the fibre slit, and from an
       intermediate position.
This design has the same f ratio, beam diameter, pupil relief, focal length and fibre slit
length as the one above. The optics contain one aspheric surface and a variety of glass
types (details are given in Damien‘s report in section 7.2.3).

The aberrations from this collimator are given in Table 7.2. As expected, they are
larger than the aberrations from the off-axis Houghton/Maksutov design above. When
divided by 1.733 in order to be interpreted as image sizes on the detector, some of
these values would still give an image size outside the specified 9 m rms radius even
with a perfect camera. In reality the camera aberrations are a significant issue
themselves, so the poor image quality of this collimator is a major concern. Even if
wavelengths below 400 nm are ignored (we originally specified optimisation to be
from 400 – 1000 nm, with evaluation down to 370 nm), there are still ‗patches‘ of
poor performance. The baseline specification is now 370-950nm. We should note,
however, that this design has been produced just as this report is being completed, and
there has been no opportunity for further optimisation. It is possible, although by no
means certain, that substantial improvements might still be made in the imaging
quality (as happened with the initial design for the ATLAS spectrograph optics).




                                              43                                Optical Design
                                                                                    AAOmega CoD


                                        RMS radius (values in m)
       (nm)             On axis               4.5568 off axis           6.4301 off axis
          370             12.73                      17.33                       35.33
          420              4.15                         8.93                     15.63
          500              4.46                       7.87                       11.74
          700             24.55                      16.67                       17.98
        1000              16.86                      12.16                       13.54
    Table 7.2: Values of RMS radius fro m the spot diagrams of collimator AAOM EGA -COL-
    204. The off-axis colu mns are specified by the angle of the rays at the pupil; the extreme of
    6.4301 corresponds to a fibre at the end of the slit. Note: these rms radii should be divided
    by 1.733 in order to be interpreted as aberration sizes on the detector. Values have been
    computed in Zemax using dithered ray pattern of density 7, referred to the centroid.


7.2.2.3 Gratings

The AAOmega spectrograph is designed to use Volume Phase Holographic (VPH)
gratings. These gratings represent a new technology, and are rapidly gaining
acceptance for use in astronomical spectrographs, due to their high efficiency. The
gratings for AAOmega are discussed in Section 7.2.4. Here we note only a few points
which bear on the optical design.
VPH gratings are used in transmission only. This has the advantage that the collimator
and camera optics are therefore on opposite sides of the grating, and mechanical
interference between them, as occurs in reflection grating systems, is eliminated. This
enables the pupil relief (i.e. the separation between the grating and the last optical
element of the collimator or first element of the camera) to be kept relatively short.
Furthermore, the gratings can then be operated in Littrow configuration, avoiding the
beam dilation characteristic of non-Littrow reflection systems. In turn, both these
features improve the performance in terms of image quality, size and cost, especially
for the camera optics.
A feature of VPH gratings is that the collimator-(grating)-camera angle depends on
the centre wavelength selected for observation (and on the grating line density) and so
must be adjustable. We call this facility an articulated camera. It represents an
additional mechanical requirement over a reflection grating system, which uses a
fixed angle. In AAOmega we anticipate allowing articulation angles from zero
(straight through, possibly for imaging) up to a maximum of 90. Mechanically it will
not be possible to go past about 90, and therefore it is not possible to reach angles
such as 135, which would enable reflection gratings to be used (included angle of
45). The mechanical complications introduced by articulation are eased by the bench
mounting of the spectrograph. The grating itself must also be able to be tilted to
different incident angles to the beam. This does not alter the centre wavelength, but is
necessary to ‗tune‘ the optimum efficiency of the grating to the selected centre
wavelength.
In terms of imaging performance, the gratings are slabs of glass, and must meet
adequate standards of optical flatness to avoid introducing unacceptable aberrations.
The flatness requirement takes the form of a maximum allowed effective slope of the
surface, rather than a maximum absolute deviation from flatness. Such slopes are


                                                   44                                 Optical Design
                                                                        AAOmega CoD


normally expressed as a certain number of ‗wavelengths per inch‘ (reflecting the
predominantly US origin of gratings). We take this number to mean the standard
deviation of the slope changes across the substrate.
A study of the effects of surface errors was carried out as part of the ATLAS design,
and showed that for a VPH grating (with four surfaces, two with air and two with the
dichromated gelatin layer) a flatness specification of /2 per inch would be more than
adequate. In AAOmega this would result in the profile of even the 85 m IFU fibres
being broadened by only 1%. However, the effect of surface errors increases rapidly,
for example four surfaces with 2 per inch will broaden the images of IFU fibres by
20%.


7.2.2.4 Camera

The optical specification given to Damien Jones asked for a fully transmitting camera,
for the reasons given in Section 7.2.1. The detector size is to be 4K  4K pixels.
Initially the pixel size was specified as 13 m, but this was later raised to 15 m to
allow for a slower camera design. The specification asked for a collimator – camera
demagnification of 2.0. With the collimator being f/3.15, this implies a camera speed
of f/1.58. This was achieved with the first design (not shown here), for 13m pixels,
but could not be maintained for the present design with a larger detector. But this
simply means that the length of the fibre slit, and hence the relative spatial separation
of the spectra on the detector, remains the same instead of increasing.
The camera design as delivered has a focal length of 272.6 mm, which in combination
with the collimator (focal length 472.5 mm) gives a demagnification ratio of 1.733
from the fibre slit to the CCD. The speed of the camera is f/1.82. The layout is shown
in Figure 7.4.
The system uses three aspheric surfaces, and a variety of glass types. As is to be
expected with a fully transmitting design for a fast camera, the image aberrations are
not negligible. The rms radii are given in Table 7.3. The mean of all these values is
10.7 m, which is close to the target specification of 9 m rms radius. Sections 7.2.6
and 7.2.7 will discuss the effects of aberrations on spectral and spatial resolution,
including the issue of whether aberrations of greater than 9 m rms radius are
acceptable. For the moment we note that in combination with the Houghton/Maksutov
collimator, this camera design would give quite acceptable res ults, with little loss of
spectral resolution or spatial blurring.
It is important that the optical system does not image spectra as significantly tilted or
curved, since this complicates extraction of individual spectra. The specification is
that the spectra should be as straight as possible, certainly within the 10 pixel fibre-
fibre spacing. A check using Zemax shows that image distortions produced by the
camera are less than 1 pixel, i.e. well within the requirement.




                                            45                           Optical Design
                                                                                  AAOmega CoD




       Figure 7.4: Layout of the transmissive camera. Ray bundles are shown for an on-axis
       image and at off-axis angles corresponding to: a) an image at the edge of the CCD but
       along the centre line, and b) an image in the corner of the CCD.


                                       RMS radius (values in m)
       (nm)            On axis               6.4301 off axis           9.0567 off axis
          370              3.7                         10.2                     16.4
          420              13.5                        13.6                     12.0
          500               8.8                         9.4                      9.0
          700              14.7                        10.6                     13.1
        1000               10.4                        5.5                      9.1
    Table 7.3: Values of RMS radius fro m the spot diagrams of camera AAOM EGA -CAM-
    411. The off-axis colu mns correspond to the ray sets shown in Figure 7.4: 6.4301 gives the
    edge of the CCD along a centre line and 9.0567 gives the corner of the detector array.
    Values have been computed in Zemax using dithered ray pattern of density 6, referred to
    the centroid.

A significant issue for this camera design is the availability and cost of the optical
materials, and the cost and difficulty of the optical figuring of the surfaces. Gabe
Bloxham of the RSAA (Mount Stromlo) optical shop has been consulted regarding
the costs of figuring the surfaces, and has commented that he is very sceptical about
the refractive camera. It would be very difficult to build at their optical shop. Indeed,
he suspects there may be fundamental difficulties with material availability and
thermal strain mismatch at coupled surfaces. We have not yet had an opportunity to
explore these concerns in more detail.




                                                  46                                Optical Design
                                                                        AAOmega CoD


7.2.3 Report PO50REP from Prime Optics
7.2.3.1 Preamble
AAOmega is a proposed upgrade for the existing 2dF spectrographs. It is intended
that ―the final system will provide much higher spectral resolving power, throughput
and stability than the existing 2dF, RGO and SPIRAL spectrographs and is intended
to replace all three.‖ (User Requirements Document, D. Lee & T. Bridges, v1.0) It
has been made possible by the impending availability of high-efficiency volume-
phase holographic (VPH) gratings and large format detectors.
7.2.3.2 Optical principles
The AAT’s primary mirror, via the 2dF corrector, forms an image at prime focus. Up
to 400 optical fibres (or Integral Field Units (IFUs)) are placed at preset image
locations on the image surface and gather light from specific objects. The optical
fibres are gathered into a bundle and feed the light into a spectrograph. Here, the other
ends of the fibres are laid side by side to form a ―fibre slit‖. The light emerging from
the slit is collimated and a pupil is formed in the collimated beam area where
dispersing elements and filters can be located. The dispersed light is then imaged by a
camera which forms spectra on a detector.
7.2.3.3 Optical modelling

7.2.3.3.1 Coordinate Systems
The Optical Axis of each optical subsystem is usually the global z-axis, (gZ), which
points in the same general direction as the incoming light. The global y-axis (gY)
points in an "upward" or ―tangential‖ direction. The global x-axis (gX) completes a
right- handed system.

7.2.3.3.2 Specification
The optical surfaces in these systems are rotationally symmetric spheres, conicoids or
aspheres, and planes whose axes of symmetry would normally lie along gZ or the
(perhaps deflected) optical axis. However, some of them may be tipped (e.g. prisms)
to provide dispersion.

A mathematical description of optical surface specification can be found in any optics
text or the ZEMAX user manual.

7.2.3.3.3 Measurement of performance
Optical system performance is measured using spot diagrams, which are essentially 2-
dimensional plots of ray intercepts on an image surface. A collimator‘s performance
can be measured by constructing similar plots of ray bundle deviations from
collimation, usually measured in radians. In both cases an RMS deviation is
calculated.
7.2.3.4 Optical Functional Requirements

7.2.3.4.1 General

   Usage
AAOmega is to be used with the AAT's 2dF corrector.


                                            47                           Optical Design
                                                                         AAOmega CoD


   Wavelength Range
The wavelength range is from 370 nm - 1000 nm. (Note, now 370 to 950nm.)

   Pupil Size
The spectrograph pupil has a diameter of 150 mm. [AAO: Chosen to suit available
VPH gratings and give good spectral resolution while containing the size and cost of
the optics.]

   Fibre Exit Focal Ratio
This assumes a nominal value of f/3.15.

   Slit Demagnification
The slit demagnification evolved to a nominal value of 1.733. It was originally
specified as an ―approximate demagnification of 2‖.

   Detector
The detector specification evolved to a 4096 2 array of 0.015mm square pixels.

   Pupil Space
At least 150 mm of space is to be provided each side of the system pupil.

   Geometrical Imagery

This is specified at 9 m RMS radius.

7.2.3.4.2 Slit
The slit has to accommodate 400 140 m fibres in groups of about 10.
   Slit Length
The edge dimension of the detector can be calculated from 4096  0.015 = 61.44mm.
When multiplied by the slit demagnification of 1.733 we arrive at a slit length of
106.476 mm. This means that a centre-to-centre fibre spacing of more than a quarter
of a millimetre can be achieved, which should be ample.

7.2.3.4.3 Collimator

   Focal Length
This is calculated from the specification of the focal ratio, 3.15, and the pupil
diameter, 150mm. It has a nominal value of 472.5 mm.

   Field of View
This is calculated from slit length and collimator focal length. It has an approximate
value of 12.9. This quantity is ultimately determined by whether or not the slit lies on
a curved surface.

7.2.3.4.4 Camera

   Focal Length
This is calculated by dividing the focal length of the collimator by the slit
demagnification factor. It has a nominal value of 272.6 mm.




                                             48                            Optical Design
                                                                        AAOmega CoD


   Field of View
This is calculated from the dimensions of the detector diagonal. It has an approximate
value of 18.1.
7.2.3.5 Optical Development and Design

7.2.3.5.1 System
The preferred configuration is completely dioptric. Catadioptric systems, unless
offering significant advantages in throughput or resolution, are the ―back up‖.

7.2.3.5.2 Collimator

   Catadiopt ric Collimator
A catadioptric collimator (with purely spherical surfaces) seemed like the way to go
following the success of an f/5 unobstructed catadioptric collimator for the new
SOAR spectrograph. This system achieved diffraction limited collimation of a fibre
slit over the waveband from 250 – 2500 nm. And so, an f/3.15 version was designed
along the same lines. It is an off-axis system with enough clearance for a fibre slit. It
consists of 3 corrector ―plates‖ near the pupil, a large primary mirror and a doublet
field corrector near the convex-curved fibre slit (as in the current 2df collimators). It
can be thought of as a Houghton-Maksutov hybrid (HMak).

A layout is shown in Figure 7.1, spot diagrams in Figure 7.2 and the system is
tabulated in Table 7.1.

It can be seen that the imagery potentially delivered by this system is excellent.

   Catadiopt ric Collimator Issues
A tolerance list for the HMak collimator is shown in Table 7.2. It is evident that there
are going to be some rather fundamental problems with this system during
manufacture and assembly exacerbated by the off-axis positions of the plates. Some
of the tolerances are quite unreasonable. Therefore, it was decided to abandon this
approach and explore the possibilities using a fully refracting or dioptric collimator.

   Diopt ric Collimat or
The dioptric system is based on a reversed Petzva l configuration with a concave
object (slit) surface concentric with the entrance pupil. This means that, ideally, every
fibre can be aligned normal to the slit. In reality, groups of about 10 parallel fibres
will be aligned normal to the slit at the centre of the group.

An interesting feature of this configuration is the weak negative triplet 3 rd component.
This has the effect of substantially reducing secondary colour.

A layout is shown in Figure 7.3, spot diagrams in Figure 7.4 and the system is
tabulated in Table 7.3.

It can be seen that the imagery potentially delivered by this system is pretty good but
there is room for improvement. Unfortunately, there wasn‘t enough time to fully
optimise this system. I would expect some useful improvements in performance. Also,
the manufacturing tolerances have not been evaluated for the same reason. I would



                                            49                            Optical Design
                                                                         AAOmega CoD


expect the mechanical tolerances to be at least double those of the camera described
below.

7.2.3.5.3 Camera

   Camera Issues

- Achromatization
As a general rule, dioptric spectrograph cameras are almost always based on a field-
flattened Petzval-type configuration. They are characterized by two or more separated
positive groups and a field-flattening lens near the image surface. The control of
chromatic effects in this type of camera has traditionally been achieved with glasses
having closely matched dispersion characteristics (a favoured pair is comprised of the
Schott glasses UBK7 and FK54). In the last decade or so, the control of colour effects
has become increasingly difficult as the number of pixels in detector arrays has
increased and the size of the individual pixels has shrunk to less than 20 m.

Contemporary applications now routinely demand focal ratios less than 2, s ignificant
pupil relief and fields of view approaching 20 degrees. These goals force the use of
novel glass mixes and lead to some difficult fabrication challenges.

Quasi-achromatization at multiple wavelengths can be achieved by using at least 3
different glasses and by a careful choice of the power distribution within an optical
system.

Quasi-apochromatization is achieved by reducing the high order sphero-chromatic
effects with the judicious application of aspheric deformations on several surfaces
bounding different glasses. In this way, the chromatic characteristics of the ―aspheric
overlay‖ can be designed to oppose the high order chromatic effects from the ―spheric
core‖ of the system.
- Field Curvature
The residual field curvature and aberrations associated with large angular fields in
spectrograph cameras is always an issue and especially so here. Basic (spherical) field
curvature is controlled with a field-flattening lens next to the detector. Quite often this
can act as a window to the detector Dewar and both can be moved together for fine-
focusing. High order field curvature and astigmatic aberrations can also be controlled
with aspheric surfaces and the power distribution within the system.

- System Configuration
There are many Petzval-type configurations that have been developed for various
applications. This particular configuration synthesises the most useful features of
systems developed by Epps (at Lick), Wynne and others (including myself) over the
last several years.

A novel feature of this system is the mildly negative 3rd component which has a
significant influence on the overall colour balance.

- Final System and Performance
A camera layout is shown in Figure 7.5, spot diagrams in Figure 7.6 and the system is
tabulated in Table 7.4. This system is known simply as version 4.11 (this does not



                                             50                            Optical Design
                                                                                    AAOmega CoD


imply preceding versions of equivalent performance or importance but simply this
camera‘s position on the ―developmental tree‖).

In general, the RMS imagery comes in under 9 m radius. However, there are
―patches‖ of wavelength and field where this is degraded. These could be improved a
little at the expense of performance from 370 – 400 nm.

- Manufacturability
The system tolerances are listed in Table 7.5. Most are quite reasonable but a few will
require special care. In my experience all are achievable with the rider that the
manufacturability of the aspheres has not been investigated in depth. However, there
are a number of technologies in routine use that could be applied. Specifically, the
opticians at Lick have been making difficult aspheres for some time. There are also a
number of commercial operators that are equally capable.
7.2.3.6 Concluding Remarks
Some optical systems have been designed for the AAOmega spectrograph. In general,
the combined systems, in their final forms, should meet the imaging requirements.




                                                                 Field Corrector

                                                               Slit




                                                                Corrector Group




                                    Pupil




Figure 7.1 Layout AAOmega Catadiopt ric Collimat or, version H3FC3.01. Corrector Group
components oversize so as to include vertices for alignment. Slit is on convex surface




                                                   51                                    Optical Design
                                                                                   AAOmega CoD




                              400 nm             540 nm            1000 nm




                          RMS radius 0.006   RMS radius 0.004   RMS radius 0.004
                                               53.090,0.019




                          RMS radius 0.004   RMS radius 0.003   RMS radius 0.005
                                               37.693,0.014




                          RMS radius 0.003   RMS radius 0.003   RMS radius 0.005
                                                0.000,0.009




Figure 7.2 Spot Diag rams. AAOmega Catadioptric Collimator, version H3FC3.01. Bo xes are
equivalent to 4 x 4 pixels on the detector




                                                  52                               Optical Design
                                                                                    AAOmega CoD


Table 7.1 CATADIOPTRIC COLLIMATOR SPECIFICATIONS, H3FC, v3.01

   Surf Ident              z        Separation           Aperture/
                                                        Radius          Next
                          mm            mm                Offset
                                                          mm           medium
                                                            OR
                                                          Height x
                                                           Width
  =================================================================

    -1   EnP        0.000      -       plane      AIR
  -----------------------------------------------------------------
     0   Sce     -inf          -       plane      AIR
  -----------------------------------------------------------------
     1 Pupil        0.000    inf       plane      AIR      150.0
  -----------------------------------------------------------------
     2   CG1     160.000    160.000 12315.417 SILICA       230.0/
                                                           105.0
      3           190.000     30.000 -3767.765     AIR      230.0/
                                                           105.0
  -----------------------------------------------------------------
     4   CG2     280.000     90.000   -321.531 SILICA      240.0/
                                                           105.0
     5           310.000     30.000   -322.647    AIR      250.0/
                                                           105.0
  -----------------------------------------------------------------
     6   CG3     390.000     80.000   -370.322 SILICA      240.0/
                                                           110.0
     7           415.000     25.000   -454.123    AIR      250.0/
                                                           110.0
  -----------------------------------------------------------------
     8   PM     1050.000    635.000   -952.105    AIR      300.0/
                                                           120.0
  -----------------------------------------------------------------
     9           632.988    417.012 -2661.706    CAF2        66.0x
                                                           120.0
    10   FC      622.988     10.000    789.993 SILICA       66.0x
                                                           120.0
    11           612.988     10.000    plane      AIR       66.0x
                                                           120.0
  -----------------------------------------------------------------
    12 Slit      562.988     50.000    435.406    AIR         1.0
  -----------------------------------------------------------------

CG1, 2 & 3 are circu lar elements. They overlap the optical axis for alignment purposes.

FC is a rectangular component.




                                                   53                                Optical Design
                                                                                      AAOmega CoD


Table 7.2 H3FC COLLIMATOR TOLERANCES, v 3.01

  Surf Ident    dec    dec     RoC   T’ness   Figure    Axial
              (surf) (lens)                   Fringe     Sep
                mm     mm       %      mm     @600 nm     mm
  ===============================================================
    2          0.5     0.5     0.5               ½
    3   CG1    0.5             0.5     0.2       ½       0.2
  ---------------------------------------------------------------
    4          0.05           <0.1               ½
    5   CG2    0.05    0.1    <0.1     0.1       ½       0.2
  ---------------------------------------------------------------
    6          0.05    0.1    <0.1               ½
    7   CG3    0.1             0.2     0.1       ½       0.2
  ---------------------------------------------------------------
    8   PM     0.1     0.1     0.2      -        ½       0.2
  ---------------------------------------------------------------
    9          0.5             0.2               2
   10 FC/1     0.5     0.2     0.5     0.1       4       0.2
   11 FC/2      -               -      0.1       2
  ---------------------------------------------------------------

Where the radius of curvature tolerance is indicated as <0.1 it is assumed that a refit would be made
when the actual radius of curvature of the testplates became known.
This off-axis system would most likely be aligned by cocentering axial retro reflections fro m the
corrector group and mirror surfaces. Therefore I have specified a decenter, rather than a tilt, tolerance.
The corrector group lenses would need at least 4 degrees of freedom in their mounts.
Generally speaking, the design would be ―refitted‖ progressively as the tooling was made.
This is a fast, off-axis, high-resolution spectrograph collimator and some of the tolerances are
commensurate with this function.




                                                    54                                 Optical Design
                                                                             AAOmega CoD




                                                            Slit


                                                             FC




                                                                   CL4




                                                                    CL3

                                                                    CL2




                                                                   CL1




                                                               Pupil




Figure 7.3 Layout AAOmega Dioptric Collimator, version 2.04. Slit is on concave surface




                                               55                              Optical Design
                                                                                              AAOmega CoD




                   400 nm             420 nm             500 nm             700 nm            1000 nm




                RMS radius 0.012   RMS radius 0.009   RMS radius 0.007   RMS radius 0.010   RMS radius 0.010
                                                       0.000,53.188




                RMS radius 0.005   RMS radius 0.005   RMS radius 0.005   RMS radius 0.009   RMS radius 0.008
                                                       0.000,37.633




                RMS radius 0.001   RMS radius 0.002   RMS radius 0.003   RMS radius 0.013   RMS radius 0.008
                                                        0.000,0.000




Figure 7.4 Spot Diag rams - AAOmega Dioptric Collimator, version 2.04. Bo xes are equivalent to 4
x 4 pixels on the detector




                                                        56                                     Optical Design
                                                         AAOmega CoD




                             Table 7.3
             DIOPTRIC COLLIMATOR SPECIFICATIONS, v2.04

Surf Ident       z     Separation   Radius     Next    Aperture/
                mm         mm         mm      medium     Term
=================================================================

  -1   EnP       0.000      -        plane      AIR
-----------------------------------------------------------------
   0   Sce     -inf         -        plane      AIR
-----------------------------------------------------------------
   1 Pupil       0.000     inf       plane      AIR      150.0
-----------------------------------------------------------------
   2           150.000    150.000    563.880   LAL7      196.0
                                    0.000000              cc
       CL1                        -2.880E-09              a 4
   3           159.000      9.000    301.983   CAF2      196.0
   4           189.000     30.000 -1796.156     AIR      196.0
-----------------------------------------------------------------
   5   CL2     358.093    169.093    657.672   CAF2      216.0
   6           386.093     28.000   -802.921    AIR      216.0
-----------------------------------------------------------------
   7           387.093      1.000    358.808   BSM14     214.0
   8   CL3     394.093      7.000    206.275   CAF2      202.0
   9           422.093     28.000    585.177   LLF6      202.0
  10           429.093      7.000    367.554    AIR      196.0
-----------------------------------------------------------------
  11           580.127    151.034    608.695   BSM14     194.0
  12   CL4     587.127      7.000    248.062   CAF2      186.0
  13           631.127     44.000   -331.728    AIR      186.0
-----------------------------------------------------------------
  14    FC     803.111    171.984   -240.855 SILICA      122.0
  15           813.111     10.000 -1525.022     AIR      122.0
-----------------------------------------------------------------
  16 Slit      853.111     40.000   -757.819    AIR      110.0
-----------------------------------------------------------------




                                 57                      Optical Design
                                                                         AAOmega CoD




                                                     Detector
                                                     FF




                                                            CM4




                                                                CM3



                                                                  CM2




                                                                CM1




                                                          Pupil




Figure 7.5 Layout. AAOmega Camera, version 4.11. 4K x 4K x 15 m pixel detector




                                            58                             Optical Design
                                                                                              AAOmega CoD




                370 nm             420 nm             500 nm             700 nm            1000 nm




             RMS radius 0.018   RMS radius 0.012   RMS radius 0.009   RMS radius 0.013   RMS radius 0.009
                                                    0.000,43.443




             RMS radius 0.011   RMS radius 0.014   RMS radius 0.009   RMS radius 0.010   RMS radius 0.006
                                                    0.000,30.720




             RMS radius 0.004   RMS radius 0.014   RMS radius 0.009   RMS radius 0.015   RMS radius 0.011
                                                     0.000,0.000




Figure 7.6 Spot Diagrams - AAOmega Camera, version 4.11. 4K x 4K x 15 m p ixel detector.
Boxes are 4 x 4 pixels




                                                       59                                       Optical Design
                                                      AAOmega CoD




                             Table 7.4
                   CAMERA SPECIFICATIONS, v4.11

Surf Ident       z     Separation   Radius     Next    Aperture/
                mm         mm         mm      medium     Term
=================================================================
  -1   EnP       0.000      -        plane      AIR
-----------------------------------------------------------------
   0   Sce     -inf         -        plane      AIR
-----------------------------------------------------------------
   1 Pupil       0.000     inf       plane      AIR      150.0
-----------------------------------------------------------------
   2           150.000    150.000    481.449   LAL7      214.0
                                    0.000000              cc
                                  -1.939E-08              a 4
       CM1                        -2.469E-13              a 6
                                  -4.268E-18              a 8
   3           159.000      9.000    267.819   CAF2      214.0
   4           201.000     42.000   -678.578    AIR      214.0
-----------------------------------------------------------------
   5   CM2     291.830     90.830   1030.649   CAF2      238.0
   6           342.830     51.000   -248.993    AIR      238.0
-----------------------------------------------------------------
   7           343.830      1.000    337.905   BSM14     222.0
   8   CM3     350.830      7.000    109.673   CAF2      194.0
   9           415.830     65.000   1060.900   PBL26     194.0
  10           422.830      7.000    491.223    AIR      190.0
-----------------------------------------------------------------
  11           456.069     33.238    224.435   BSM14     194.0
                                    0.000000              cc
                                   2.205E-08              a 4
                                   1.065E-12              a 6
                                   2.234E-17              a 8
  12   CM4     463.069      7.000    106.516   FPL51     174.0
  13           530.069     67.000   -309.746    AIR      174.0
                                    0.000000              cc
                                  -6.663E-09              a 4
                                  -3.288E-13              a 6
                                  -1.123E-16              a 8
-----------------------------------------------------------------
  14    FF     606.898     76.830   -128.967   LAL7       98.0
  15           611.898      5.000    329.975    AIR       98.0
-----------------------------------------------------------------
  16    D      625.898     14.000    plane      AIR       90.0
-----------------------------------------------------------------




                                 60                    Optical Design
                                                                                  AAOmega CoD




                                            Table 7.5
                                 CAMERA TOLERANCES , v4 .11

  Surf Ident    TIR    TIR     RoC   T’ness   Figure    Axial
              (lens) (mount)                  Fringe     Sep
                mm     mm       %      mm     @600 nm     mm
  ===============================================================
    2                          0.2               ½
    3 CM1/1    0.05    0.05    0.5     0.2       4       0.15
    4 CM1/2    0.05            0.2     0.2       ½
  ---------------------------------------------------------------
    5                          0.2               1
    6 CM2      0.05    0.05   <0.1     0.1       1       0.1
  ---------------------------------------------------------------
    7                          0.2               1
    8 CM3/1    0.04    0.05   <0.1     0.1       4       0.2
    9 CM3/2    0.05            0.5     0.1       4
   10 CM3/3    0.07            0.2     0.1       2
  ---------------------------------------------------------------
   11                          0.1               2
   12 CM4/1    0.05    0.025   0.1     0.05      4       0.1
   13 CM4/2    0.05            0.1     0.05      2
  ---------------------------------------------------------------
   14                          0.2               4
   15 FF       0.05    0.04    0.2     0.1       4       0.1
  ---------------------------------------------------------------

Where the radius of curvature tolerance is indicated as <0.1 it is assumed that a refit would be made
when the actual radius of curvature of the test plates became known.
Generally speaking, the design would be ―refitted‖ progressively as the tooling was made.
This is a fast, wide-field, h igh-resolution spectrograph camera and the tolerances are commensurate
with this function.




                                                 61                                Optical Design
                                                                                   AAOmega CoD


7.2.4 VPH diffraction gratings
7.2.4.1 Volume Phase Holographic (VPH) gratings – in brief
A VPH grating consists of a layer of dichromated gelatin (DCG) sandwiched between
suitable glass substrates. The layer is about 3 – 20 m thick, and is quite transparent.
The grating action is achieved not by surface relief features as in a conventional
grating, but by modulations of the refractive index. These impress periodic phase
variations on the incoming wavefront, resulting in diffraction. Figure 7.5 illus trates
the sinusoidal modulation of the DCG layer. Very high efficiency (~90% at peak) can
be obtained at the design wavelength, and close to this efficiency within a range
around the optimum wavelength. This represents a marked improvement over the
~60% typical of a conventional reflection grating. The high efficiency, and high
resolution obtainable, are the reasons why we expect VPH gratings to form the basis
of the next generation of spectrographs. The construction and use of VPH gratings is
described by Barden et al. 5,6 and Clemens et al. 7




     Figure 7.5: Schematic illustration (not to scale) of a VPH grating. The upper panel represents the
     dichromated gelatin layer, where darker shading implies higher refractive index, not absorption.
     The lower panel shows the corresponding graph of refractive index vs position along the layer.
     Only two ‗lines‘ of the grating are shown, covering a few m.

A key point is that the depth of the grating (3 - 20m) is considerably greater than the
wavelength of light. It is for this reason that the gratings are referred to as Volume
Phase Holographic. The significant depth of the gratings means that the grating
efficiency is high only in directions obeying the Bragg relation (effectively specular
reflection from the refractive index modulations). This is an important point because
it leads to the requirement for a general purpose VPH spectrograph to have an
additional degree of freedom: as well as rotating the grating, it must be possible to
position the camera to accept a wide range of collimator- grating-camera angles. This
is the concept of an articulated camera. The gratings are tunable in that tilting the
grating with respect to the incident beam changes the wavelength which obeys the

5
  S.C. Barden, J.A. Arns and W.S. Co lburn, ―Volu me-phase holographic gratings and their potential
for astronomical applications‖, Proc. SPIE 3355, 866-876, 1998.
6
  S.C. Barden, J.A. Arns, W.S. Colburn and J.B. Williams, ―Vo lu me-Phase Holographic gratings and
the efficiency of three simple VPH gratings‖, PASP 112, 809, 2000.
7
  J.C. Clemens, H.W. Epps and S. Seagroves, ―Optics for a Volu me Holographic Grat ing Spectrograph
for the Southern Astrophysical Research (SOA R) Telescope‖, Proc. SPIE 4008, 2000


                                                  62                                 Optical Design
                                                                        AAOmega CoD


Bragg condition, and so effectively tunes the grating to peak efficiency at a different
wavelength. It is this process which replaces the concept of blaze for a conventional
surface grating: for VPH there is no fixed ‗blaze wavelength‘. A VPH grating does,
however, have a design wavelength at which its performance is optimised. This is the
wavelength at which the product of the thickness and the amplitude of refractive
index modulations gives full modulation of the phase of the wavefront for the
appropriate angle of incidence. When tuned to this wavelength, the grating‘s peak
efficiency will take its highest value. When tuned to other wavelengths the peak
efficiency will be somewhat lower. The envelope of the peak efficiencies as a grating
is tuned is referred to as the ‗superblaze‘ curve. Figure 7.6 illustrates that VPH
gratings may be tuned by changing the angle of incidence, as well as showing their
high peak efficiency.

The VPH gratings that we are proposing for use in AAOmega are transmission
gratings. It is possible to produce VPH reflection gratings, but they are effective over
only a narrow wavelength range at any one time. Using transmission gratings also
means that it is straightforward to achieve Littrow operation of the spectrograph
(incident and diffracted beams make equal angles with the grating normal). In
contrast, with reflection gratings, Littrow operation creates the difficulty that the
collimator and camera must be effectively in the same place, perhaps sharing the same
optics. Use of Littrow configuration gives the advantage that there is no anamorphic
beam expansion, hence the camera aperture can be substantially smaller than for a
non-Littrow design. This significantly eases the optical design, avoiding compromises
necessary to facilitate a range of beam dilations, and thus reducing light losses and
costs. Use in transmission also allows shorter pupil relief between the grating and
both the collimator and camera. This again reduces the size of the required camera
aperture, and even more importantly, increases the field of view.

Because the refractive index modulations are produced by a holographic process (i.e.
interference of two large collimated laser beams), it is possible to have high line
densities, up to about 6000 lines/mm. This contrasts with conventional ruled
reflection gratings, for which the maximum generally available is 1200 l/mm. A high
line density enables high dispersion to be obtained in first order, and is another
significant advantage of VPH gratings. For high line densities immersion prisms may
be used on both sides of the grating, to reduce the angles of incidence and diffraction,
and indeed to avoid total internal reflection in extreme cases. The provision of such
prisms is not a problem, but in AAOmega we expect that suitably specified ant i-
reflection coatings will be sufficient.

Other advantages of VPH gratings are that each grating is made to order, rather than
being selected from a limited catalogue, costs are less than for large ruled gratings,
and large sizes are feasible. Moreover the DCG layer containing the hologram is
protected between a substrate and a superstrate, both made of glass. The exposed
surfaces can be anti-reflection coated for the optimum angle and wavelength.

The application of VPH gratings does have some problems, however. Firstly, VPH
gratings are not immune from the general trend that it is hard to maintain good
efficiency at high spectral resolution. Thus although resolution can be readily
increased by using a high line density grating, such high resolutions are inevitably
associated with a large angular deviation of the beam by the grating. It is a property of


                                            63                           Optical Design
                                                                        AAOmega CoD


VPH gratings that light in one of the polarization planes is progressively lost at larger
deviation angles, thus losing overall efficiency. At 90 deviation (within the grating),
the p-plane light is lost entirely. In the AAOmega concept we are able to obtain
spectral resolution R (=/) up to 7600 for MOS fibres and 12,500 for IFU fibres
without entering the regime in which serious p-plane loss occurs.

The second VPH property which can potentially cause trouble is the ‗bandwidth‘ of
the gratings. This refers to the fact that the Bragg condition is rather strongly
dependent on diffracted angle and hence on wavelength. Thus when a VPH grating
has been tilted to tune a certain wavelength to peak efficiency (and the camera moved
to receive that wavelength in the centre of the detector), the rest of the wavelength
range which is diffracted to the camera and detector is not at the optimum Bragg
condition. This can result in a significant drop of efficiency towards the two ends of
the wavelength range on the detector. In this context the term ‗bandwidth‘ is used to
mean the wavelength range within which the efficiency is at least half of the peak
value. The bandwidth improves if the grating thickness can be reduced, but in order to
do this (while maintaining good peak efficiency and a fixed design wavelength) it is
necessary to increase the amplitude of the refractive index modulations. Recent work
at grating manufacturers Ralcon Development Corporation and Kaiser Optical
Systems Inc. (KOSI) has shown that effective thicknesses as small as ~3m are
possible, which enables a grating to maintain good efficiency over the entire
wavelength range which falls on the detector.

Other problems associated with the use of VPH gratings at the present time are:
 The extra mechanical complication of the articulated camera is a significant issue,
   though eased by the horizontal bench mounting of the AAOmega spectrograph.
   The grating tilt relative to the incoming beam must also be variable (as in a
   reflection grating spectrograph).
 VPH gratings are as yet unproven in a large intermediate-dispersion astronomical
   spectrograph, although one is already successfully in use for low-dispersion
   survey spectroscopy at the AAT 8 , and others in the FORS instrument at the VLT.
   Several spectrographs using VPH gratings are under design or construction. We
   have taken delivery of several full-size gratings for tests, which are described in
   the following section.
7.2.4.2 Efficiency and flatness tests of Ralcon 1516 prototype grating
During the ATLAS design study phase, an order was placed with Ralcon for two
gratings:
 Both have identical DCG specifications, but one is on standard ‗Starphire‘ glass
    substrate/superstrate, while the other is on glass with flatness /2 per inch.
 1516 lines/mm
 Design wavelength 650 nm
 Maximum bandwidth consistent with good efficiency
 Size 170 mm  220 mm (grating lines run in short dimension), suitable for
    operation in a 150 mm collimated beam.


8
 K. Glazebrook, ―LDSS++ - new u ltra-h igh performance red spectroscopy for the AAT‖, AAO
Newsletter 84, 1998.


                                            64                           Optical Design
                                                                                AAOmega CoD


The purpose of this order was to gain experience with the process of specifying
gratings, to evaluate the efficiency of the gratings, and to examine the imaging quality
for gratings of different flatness specification. Since the size is that recommended for
AAOmega, the grating(s) from this order could be used in the final instrument if their
performance is satisfactory. The grating on standard ‗Starphire‘ glass was received in
May 2000, and its efficiency was tested in June 2000. Due to some d ifficulties in
manufacture, two copies of the second grating (on flat substrates) were produced.
Tests on the flatness of the gratings were made in November 2000.
Efficiency tests were made using a setup 9 with a quartz halogen light source, double
monochromator, grating on a rotatable mount, and ‗articulated camera‘ consisting of a
converging lens and photoelectric detector mounted together on an arm which can be
rotated about the grating as required. With this setup it is possible to measure the
efficiency at different wavelengths with a fixed collimator–grating tilt angle
(simulating an exposure in a spectrograph, where the detector covers a substantial
wavelength range) or to follow the ‗superblaze‘ by setting both the grating tilt and the
camera articulation to the optimum for each wavelength.

Figure 7.6 shows the results. There are a number of significant aspects of these test
results:
 The peak efficiency is about 82%. With suitable antireflection coatings on both
    external surfaces this could be expected to rise to ~89%, a very satisfactory
    demonstration of the substantial efficiency advantage of VPH gratings.
 The bandwidth is also very satisfactory, and would give an efficiency drop at the
    edges of the AAOmega detector of only 15%. Simulation of this grating shows
    that an effective thickness of approximately 3.0 m and refractive index
    modulation of n = 0.09 have been achieved, explaining the excellent wide-band
    performance.
 The peak of the superblaze, i.e. the wavelength at which the best performance is
    obtained, is at 510 nm rather than the specified 650 nm. This is a significant
    difference. This error in the design wavelength may be partly due to
    misunderstanding of the priorities of our various requirements on centre
    wavelength and bandwidth. (We asked for the grating to be as thin as possible in
    order to maximise the bandwidth, but if the processing then hit the limit of
    maximum n, the result would be a superblaze peak wavelength less than
    desired.)




9
    I.K. Baldry, ―VPH Gratings in ATLAS‖, AAO internal report, 27 April 2000.


                                                  65                            Optical Design
                                                                                  AAOmega CoD



                          1
                         0.9
                         0.8
                         0.7
            Efficiency   0.6
                         0.5
                         0.4
                         0.3
                         0.2
                         0.1
                          0
                           400   500     600         700         800        900        1000
                                               Wavelength (nm)


        Figure 7.6: Results of efficiency tests of Ralcon 1516 l/ mm grating (Starphire
        substrate). The continuous curves show the efficiency vs wavelength for four different
        values of grating tilt angle (thus tuning the grating to optimise at different
        wavelengths). The heavy curve of this set shows the best results for this grating. The
        series of plotted points show the superblaze. The ‗scale bar‘ shows the wavelength
        coverage that would be obtained for this grating with the AAOmega f/1.82 camera and
        a detector of 4K  15 m p ixels. The test sample had no antireflection coating.

In November 2000, tests of the imaging quality of all three of the Ralcon sample
gratings were carried out 10 . An f/10 collimator was used to form a beam of 95 mm
diameter, with a 10 m pinhole as the light source. After passing through the VPH
grating and being diffracted into first order, a similar f/10 optical system imaged the
spot on to a CCD detector. Allowance was made for the non-zero spot size due to the
pinhole. The results showed good imaging quality from all three gratings, even though
only the later two were specified as having flat substrates and superstrates. Converting
the spot sizes to allow for AAOmega‘s faster camera (f/1.82), the rms radii on the
detector would be from 3 to 4 m. This would cause negligible broadening of the
images. We conclude that the optical flatness requirements can be met with available
gratings.
7.2.4.3 VPH gratings in AAOmega
Here we consider several aspects concerning the practical use of VPH gratings in
AAOmega.

7.2.4.3.1 Grating size
Gratings must be larger than the nominal beam diameter (e.g. 150 mm in                           the
transmissive system) because of the following factors:
1. The gratings are tilted with respect to the beam and so must be longer in                     the
    dispersion direction.
2. Due to the tilt of the grating, edge rays encounter it somewhat before or after               the
    precise pupil plane. Hence rays from off-axis objects require an increase in                 the
    grating size to avoid vignetting11 .

10
   D.Lee and J.G. Robertson ‗VPH grat ing image quality tests‘ AAO Internal report, 27 November
2000.
11
   I.K. Baldry, ―VPH Gratings in ATLAS‖, AAO internal report, 27 April 2000.


                                                  66                                Optical Design
                                                                                                    AAOmega CoD


The maximum clear aperture is required by high dispersion gratings, which are used
at large tilt angles. At the nominal maximum beam deviation of 90 (i.e. maximum tilt
of 45) the clear aperture requirement is 166  212 mm. Gratings can be smaller for
lower dispersions, but for mounting it is convenient to have them all the same size.
The actual grating size must be larger than the clear aperture to take account of the
mounting and any rim at the edge beyond the extent of the processed DCG layer. A
dimension of 170 mm (along grating lines)  220 mm (across the set of lines) was
used for ordering the prototype grating described above. Minor changes may be made
to this size for the final AAOmega gratings.

7.2.4.3.2 Camera articulation angles
Figure 7.7 shows the values of articulation angle required as a function of centre
wavelength and grating line density.
The cut-off in the curves for the two higher dispersion gratings is imposed at a
deviation angle within the DCG medium of 60; this corresponds to an efficiency loss
of somewhat less than 25%. Gratings could be pushed a little towards higher deviation
(hence higher resolution) but efficiency drops rapidly, and this is a convenient end
point. The corresponding deviation angle in air, when immersion prisms are not used,
is close to 90. Hence the specification for articulation angles between zero ( in
practice -1) and 90.


A significant mechanical issue is how fine the steps in articulation angle need to be.
When expressed as a fraction of the total wavelength range on the detector, the shift in
centre wavelength for a fixed increment of the articulation angle is constant. It is a
shift of 7.6% of the wavelength range for a 1 change in articulation. Observers would
want to control the articulation considerably more finely than this. Steps of 0.1
would be more appropriate.

                                     100

                                      90

                                      80                                      1556
                                                          2500
          Articulation angle (deg)




                                      70

                                      60
                                                                                           1200
                                      50

                                      40

                                      30                                                      600
                                      20

                                      10

                                       0
                                        300   400   500          600    700          800   900       1000
                                                          Centre wavelength /nm

       Figure 7.7: Collimator – grating – camera articulat ion angles for gratings with four
       different line densities (l/ mm). The values apply for each wavelength when set up as
       the centre wavelength. The curves for 1556 and 2500 l/ mm are cut off at the
       wavelengths which produce a deviation of 60 within the DCG.


                                                                   67                               Optical Design
                                                                                    AAOmega CoD


7.2.5 Throughput efficiency
7.2.5.1 Antireflection coatings
Multi-element transmission optical systems are considerably more attractive now than
in the past, because of the development of highly efficient broad-band antireflection
coatings. In particular, the use of silica spheres (sol- gel) has provided a material with
a refractive index close to the geometric mean of air and glass, as required to
minimise reflections. Coatings of sol-gel over a layer of the conventional magnesium
fluoride (MgF2 ) antireflection coating have proven even more effective. Use of such
coatings in astronomical instruments has been pioneered by Jim Stilburn 12 at the DAO
in Canada, and applied in the Gemini GMOS spectrograph. For the AAOmega
transmission elements we propose the use of MgF 2 overcoated with sol- gel.
Figure 7.8 shows the excellent wide-band performance of MgF2 +sol-gel. For
AAOmega we would seek to tune down the response a little, to improve the blue
response at the expense of slight additional loss in the red.
Experiments with sol-gel coating of optics have been started at the AAT, by Steven
Lee, in order to develop expertise in this technology within the AAO. This work is
intended to provide an alternative to commercial coating for the sol-gel. However,
commercial manufacturers are expected to be able to produce MgF 2 +sol-gel routinely
by the time this is needed for AAOmega.




            Figure 7.8: Throughput of antireflective coatings applied to a single air-g lass
            interface. The sol-gel + MgF2 coating can be tuned down so that high transmission
            (reflectiv ity ~ 1% per surface) can be achieved at all AAOmega wavelengths
            (370n m-1000n m). Note that this is for normal incidence. (Figure by Jim Stilburn,
            DAO)




12
   J.R. St ilburn, ―High-efficiency sol-gel antireflection coatings for astronomical optics‖, Proc. SPIE
4008, 1361, 2000.


                                                   68                                Optical Design
                                                                                     AAOmega CoD


7.2.5.2 AAOmega optical throughput
The throughput as a function of wavelength has been modelled for the following
components of AAOmega:
1. 2dF prime focus corrector/ADC + 27 m of fibre
2. Collimator (both the off-axis reflective/transmissive and the fully transmissive
    designs).
3. VPH gratings (examples)
4. Camera
5. CCD (either MITLL, EEV or Fairchild)
Altogether, this omits only the losses due to the atmosphere and the telescope primary
mirror. Since the IFU option will use SPIRAL at the Cassegrain focus, these
calculations will apply only to the MOS fibre application of AAOmega.

                        1
                       0.9
                       0.8
                       0.7
                                                                                     Corr+fibre
          Throughput




                       0.6
                                                                                     Coll_refl
                       0.5                                                           Camera
                       0.4                                                           Coll_trans
                       0.3                                                           Optics_rc
                                                                                     Optics_tc
                       0.2
                       0.1
                        0
                             300   400   500     600   700      800   900   1000
                                               Wavelength /nm

          Figure 7.9: Transmission of AAOmega optical co mponents:
          Corr+fibre: 2dF corrector, at mospheric dispersion compensator + 30 m fibre
          Coll_refl: Off-axis reflect ive/transmissive collimator H3FC301
          Coll_trans: Fully transmissive collimator AAOM EGA -COL-204
          Camera: Fully transmissive camera AAOM EGA -CAM-411
          Optics_rc: Product of the above, with reflective collimator
          Optics_tc: Product of the above, with transmissive collimator

Figure 7.9 shows the throughput curves for the corrector+fibre, collimators, and
camera. The air-glass surfaces have been modelled as having MgF 2 +sol-gel coatings,
with performance as in Figure 7.8 13 . Note the following features:
 There is a substantial drop in efficiency of the corrector+fibres towards the blue
    end. This will be unavoidable. The AAOUC have asked that it should be this that
    limits the blue performance rather than the spectrograph, i.e. the system should go
    as far to the blue as possible.



13
   No model of MgF2 +sol-gel was available for Zemax, so the calculation was done by using Zemax to
find the transmission with all air-glass surfaces having MgF2 coatings, and then again with ‗ideal‘ zero-
reflection at the same surfaces. A scaling factor derived fro m the MgF 2 and MgF2 +sol-gel curves in
Figure 7.8 was then used to find the MgF2 +sol-gel result.


                                                         69                           Optical Design
                                                                                  AAOmega CoD


   The two collimators have similar efficiency curves, although the transmissive
    design is consistently a few percent better.
   The camera has greater losses in the blue than the collimators, due to the larger
    number and thickness of optical elements.
   Collimator and camera losses in the blue might be reduced somewhat by tuning
    down the optimum wavelength of the coatings. Further study is needed of this
    possibility.
   The two lowest curves are the products corrector+fibre  collimator  camera for
    the two collimators. They show 70% transmission in the red, dropping to 20% at
    <370 nm. These curves could be compared with the corresponding ones for the
    DBSS design in Figure 7.24, with the latter having separate curves for the blue
    and red arms.


7.2.5.3 VPH Grating Efficiencies

                       1
                      0.9
                      0.8
                      0.7
         Throughput




                      0.6
                      0.5                                                         VPH_1020
                                                                                  VPH_720
                      0.4
                                                                                  VPH_722
                      0.3
                      0.2
                      0.1
                       0
                            300   400   500     600   700      800   900   1000
                                              Wavelength /nm


      Figure 7.10: Efficiency curves for examp le VPH gratings. They have been computed
      using Gsolver, a rigorous coupled-wave calculat ion which includes many orders of
      diffraction. The grating specifications are:
      VPH_1020: 1020 l/ mm, DCG thickness d= 6.5m, mean refractive index 1.4, sinusoidal
      refract ive index modulat ion to extremes of n= 0.057, incident angle in air 21.3.
      VPH_720: 720 l/ mm, d=5m, n =0.072, incident angle = 14.6.
      VPH_722: 722 l/ mm, d=8m, n =0.027, incident angle = 9.2.
      Losses at the two air-glass surfaces are not included; with MgF2 +sol-gel coatings these
      should not total more than about 1%.

Efficiency curves for example VPH gratings are given in Figure 7.10. The wavelength
range covered at good efficiency is less for the grating with the higher line density, but
the range on the detector is correspondingly reduced because of the higher spectral
dispersion. As will be shown below, both the 1020 and 720 l/mm gratings would give
good results in AAOmega. The 722 l/mm grating is shown as an example o f the
efficiency obtainable in the blue, but because it has been computed using a low
modulation strength and large DCG thickness, it has a rather narrow bandwidth in
comparison with the wavelength range on the detector. Actual gratings would be
obtained with higher modulation and hence wider bandwidth.


                                                        70                          Optical Design
                                                                               AAOmega CoD



7.2.5.4 CCD Efficiencies


                1
               0.9
               0.8
               0.7
               0.6
         RQE




               0.5                                                              MITTL3
               0.4                                                              EEV2
                                                                                Fairchild
               0.3
               0.2
               0.1
                0
                 300   400     500     600      700     800     900     1000
                                     Wavelength /nm

       Figure 7.11: Responsive quantum efficiency of the three types of CCD discussed for
       the AAOmega detector.

Two types of CCD detector have been considered for AAOmega for some time – the
MIT Lincoln Labs and EEV detectors. Their quantum efficiency curves are shown in
Figure 7.11. There is a clear superiority of the EEV device in the blue and the MITLL
in the red. Both types are available with 15 m pixels. Since both detectors are 2K 
4K, a two detector mosaic is required, and one possibility would be to use an EEV
detector at the blue end and a MITLL detector at the red end. This is only applicable to
low dispersion spectra – for higher dispersion one of the two CCDs would be of the
less suitable type. Alternatively, it would be feasible to change to a blue-sensitive or
red-sensitive detector package for a specific observing program. We present here
efficiency curves computed with both types.

However, we have recently become aware of a 4K  4K CCD from Fairchild, also
with 15 m pixels. Having a single device rather than a mosaic of two eliminates the
gap in spectra which would occur where the two detectors are butted. (The gap has to
be in the spectral direction rather than the spatial direction to allow Nod & Shuffle.)
The Fairchild CCD also has a very good RQE curve, as Figure 7.11 shows. It is now
the more likely option for the transmissive (single beam) AAOmega system.




                                               71                              Optical Design
                                                                                  AAOmega CoD



7.2.5.5 System Efficiency

                       1
                                                                            720
                      0.9
                                                                         1020
                      0.8

                      0.7
         Throughput




                      0.6
                                    722
                      0.5

                      0.4                                                   720

                      0.3
                                                                  1020
                      0.2
                      0.1

                       0
                            300   400     500    600     700      800       900      1000
                                                Wavelength /nm

       Figure 7.12a: Examp les of system throughput, with EEV and MITLL CCDs. The
       curves labelled ‗720‘ and ‗1020‘ are made up of the product of throughputs of the
       corrector+fibre  fully transmissive collimator  VPH grating of 720 or 1020 l/ mm as
       in Figure 7.10  Transmissive camera  MITLL CCD. For the curve labelled ‗722‘ the
       EEV CCD is assumed, and the 722 l/ mm g rating fro m Figure 7.10. These curves are
       examples of the complete system efficiency, except for the atmospheric and primary
       mirror losses. The ‗scale bars‘ show the wavelength range on the detector for two
       gratings.

Putting all these efficiency factors together, Figure 7.12a shows the results for the
example gratings. For the two that are optimised towards the red end, the MITLL CCD
has been assumed. This is a complete system efficiency except for the atmospheric
extinction (10% loss at these wavelengths) and the primary mirror loss (10 – 15%).
This is an excellent total throughput, well in excess of the 2dF performance. There is
some drop-off in efficiency towards the extreme ends of the wavelength range on the
detector, but by an acceptable amount (~30%). The peak 2dF throughput with the
270R/316R gratings is typically 9%. The maximum throughput in the blue is about
5%.

The 722 l/mm grating is included in Figure 7.12 to illustrate the effects of the dropping
efficiency towards the blue end of the spectrum. It has a narrower bandwidth than
desirable, and in practice a thinner grating, with larger n, would be specified. This
would extend the coverage further towards the red, but could produce only a slight
improvement in the drop-off at the blue end. Unless one choses to use a mosaic
composed of one EEV detector and one MITLL detector, the sensitivity at either the
blue or the red end would be worse than suggested by Figure 7.12a.

However, Figure 7.12b shows the corresponding throughput curves when using the
Fairchild CCD. It appears to be an excellent choice for a single detector. The system
throughput for the two red gratings shown is actually a little better than with the
MITLL device. As Figure 7.11 shows, the Fairchild CCD is better than either of the



                                                   72                             Optical Design
                                                                                  AAOmega CoD


EEV or MITLL at mid-range wavelengths, and is only slightly less sensitive than the
better of the EEV or MITLL at the ends of the spectrum.

The throughput efficiency of AAOmega when using the SPIRAL IFU has not yet been
studied in detail. The throughput of the SPIRAL fore-optics + microlenses + fibres has
been measured at ~ 85%, which compares well with the 2dF corrector + fibres.
However, SPIRAL is used at the Cassegrain focus, and therefore includes the
reflection losses of the secondary mirror, and a third mirror if it is used at the auxiliary
Cassegrain focus (for convenience). In total, the throughput for the SPIRAL IFU
would be somewhat lower than for the MOS fibres. (Note that in the case of the dual-
beam Schmidt design, IFU losses would be much more significant, due to the central
obstruction.)


                       1
                                                                            720
                      0.9
                                                                        1020
                      0.8

                      0.7
         Throughput




                      0.6

                      0.5
                                  722
                      0.4                                                   720
                                                               1020
                      0.3

                      0.2
                      0.1

                       0
                            300   400   500    600     700        800       900      1000
                                              Wavelength /nm

       Figure 7.12b: Examp les of system throughput, with Fairchild CCD. The curves are
       made up of the throughput products: corrector+fibre  fully transmissive collimator 
       VPH grating of 720, 722 or 1020 l/ mm as in Figure 7.10  Transmissive camera 
       Fairch ild CCD. These curves are examp les of the comp lete system efficiency, except
       for the atmospheric and primary mirror losses. The ‗scale bars‘ show the wavelength
       range on the detector for two gratings.


7.2.6 Spectral resolution
7.2.6.1 Introduction
Obtaining higher spectral resolution than the maximum available from 2dF (R~4000
at 900nm) is one of the main aims for the AAOmega instrument. It is achieved in two
ways:
 The higher line density of VPH gratings, and the ability of the articulated camera
    to work at the large deviation angles which follow from high dispersion gratings.
 The inclusion of an IFU facility (via SPIRAL) with 85 m fibres instead of the
    140 m MOS fibres used by both 2dF and AAOmega. The narrow fibres allow the
    IFU to act as an image slicer and so increase the spectral resolution without losing
    light from the edges of the seeing disc.




                                                 73                               Optical Design
                                                                          AAOmega CoD


The spectral resolution for both MOS and IFU fibres has been calculated. The
definition of when two spectral lines are ‗just resolved‘ is not completely obvious. For
slit spectrographs the usual definition is that the image on the detector of one side of
the slit at  is coincident with the image of the other side of the slit at  + ; in other
words  is the wavelength equivalent of the slit width. In practice aberrations will
blur the sharp sides of the theoretical slit image, giving something close to separation
by the FWHM for lines that are declared as just resolved by this criterion.
For a fibre feed we have to allow for the fact that the circular input profile is more
centrally concentrated than a (uniformly illuminated) slit. It would give a pessimistic
view of the resolution in a fibre spectrograph to simply use the fibre diameter as if it
were a slit width. Numerical experiments have shown that a separation of the profiles
equal to their FWHM corresponds to them being barely resolved. For circular images
projected to 1 dimension (as for two closely spaced lines in a spectrum, summed in
the spatial direction) the FWHM is 0.87  the fibre diameter. The resolution values in
Figures 7.13 and 7.14 below therefore represent R = / where  is the wavelength
equivalent of 0.87  the geometrically imaged fibre diameter (i.e. ignoring
aberrations). The effects of aberrations are considered below.


Note that for VPH gratings, it is neither necessary nor desirable to use second order
diffraction to achieve high resolution. Use of second order arises with ruled gratings
because the maximum line density is 1200 l/mm; with VPH gratings one would prefer
to specify a higher line density and continue to use first order, to give higher
efficiency. It is the beam deviation which controls the resolution, and using second
order is simply an alternative to a higher line density in first order, to achieve the
same angles. All calculations given here assume first order diffraction.




                                             74                             Optical Design
                                                                                 AAOmega CoD




7.2.6.2 Spectral resolution results for AAOmega

                   10000

                             2500                        1556

                                                                                1200
      Resolution




                                                                         600
                   1000




                    100
                       300     400   500      600        700       800         900       1000
                                             Wavelength /nm

    Figure 7.13: Spectral resolution R = / of AAOmega with 140 m M OS fibres and four
    different example grat ings, each labelled with the number of lines/mm. The curves show the
    resolution obtained with each wavelength regarded as the centre wavelength. The heavy lines
    show examp les of the wavelength range that can be obtained in one exposure.

Figure 7.13 gives the results for the 140 m MOS fibres. The curves for the two
higher dispersion gratings are cut off where the deviation angle within the grating is
60, which is an arbitrary but reasonable limit. Grating efficiency drops quite quickly
beyond this angle. The value 1556 l/mm was chosen as reaching this limit at 900 nm.




                                                75                                   Optical Design
                                                                              AAOmega CoD



                   100000


      Resolution




                              2500                        1556
                   10000
                                                                                  1200


                                                                                 600



                    1000
                        300   400    500      600       700       800      900         1000
                                            Wavelength /nm



    Figure 7.14: Resolution for 85 m IFU fib res in AAOmega. See Figure 7.13 for additional
    explanation.


Figure 7.14 shows the corresponding resolutions vs wavelength for the smaller IFU
fibres (note that the vertical axis now extends to 100,000). Resolutions are greater by
a factor 140/85 = 1.65.
The sampling of the fibre images is 4.68 pixels per FWHM for the 140 m fibres, and
2.84 pixels for the IFU fibres. These values correspond to the fibre diameter reduced
by the demagnification factor of 1.733, scaled by the factor FWHM/diameter = 0.87
and converted to a number of 15 m pixels. Since the spectrograph operates at
Littrow configuration, there is no anamorphic beam expansion or contraction.
Figure 7.15 shows the wavelength range on the detector, as a function of the
resolution in wavelength units, i.e. .




                                               76                                Optical Design
                                                                                                                     AAOmega CoD


                                         400

                                         350
                                                                                                               600
                                         300




                  Wavalnegth range /nm
                                         250

                                         200

                                         150
                                                                          1200
                                         100
                                                                   1556
                                         50
                                                          2500
                                          0
                                               0   0.05     0.1    0.15    0.2    0.25    0.3     0.35   0.4     0.45
                                                                  Wavelength resolution   /nm



Figure 7.15: Wavelength range on the detector, as a function of the resolution  in wavelength units
(which is more nearly constant with varying centre wavelength than R = /). The annotations give
the number of lines/mm of the grating producing that point.

7.2.6.3 The effect of aberrations on spectral resolution
Aberrations in the spectrograph optics will blur the images on the detector and so
reduce the spectral resolution. This is a significant issue because the all-transmissive
designs for the collimator and camera are challenging, and have significant
aberrations.
The effects have been assessed in two ways:
1. With some approximations, it is possible to produce a graph giving the reduction
    of resolution as a function of the net rms radius (i.e. including collimator and
    camera). With the information from sections 7.2.2.1, 7.2.2.2, and 7.2.2.4 it is then
    possible to estimate the amount of resolution degradation.
2. Some numerical experiments have been carried out in which a fibre was imaged
    through the camera. This gives examples of actual profiles.

7.2.6.3.1 Resolution as a function of rms radius
The broadening effects of aberrations can be estimated by combining in quadrature
the FWHMs of the ideal fibre profile and the aberration spot pattern from a point
source:
 The ideal fibre image FWHM is taken as fibre diameter  demagnification factor
    (1/1.733)  FWHM/diameter of a circle when projected to 1 dimension (0.87).
 The FWHM of the (projected) aberration spot pattern is estimated as 1.667  the
    rms radius (which is an azimuthal average, not a cross-cut). This scaling is exact
    for a 2-dimensional Gaussian distribution of rays. Aberration spots are not
    Gaussian, but are centrally concentrated with wings, so this provides a reasonable
    approximation. (For a 2-dimensional Gaussian the FWHM of a cross-cut and a
    projection to 1 dimension are equal.)

Strictly, it is only 2nd moments rather than FWHMs that should be combined in
quadrature. But since both the ideal and the aberration profiles are centrally
concentrated functions, it is a reasonable approximation.




                                                                            77                                          Optical Design
                                                                                                             AAOmega CoD


The smooth curves in Figure 7.16 give the results of this calculation, expressed as:
Resolution degradation factor = resolution with aberrations /
                                                         resolution without aberrations.
When expressed in this way it is independent of the particular grating being used. It is
also independent of the pixel size. The reduction is naturally worse for the smaller
IFU fibres, since the ideal width on the detector is smaller. These two curves give the
factor by which the resolutions shown in Figures 7.13 and 7.14 should be reduced for
any given net rms radius. Figure 7.16 shows that the nominal specification of 9 m
rms radius (including both collimator and camera) would result in only 2% reduction
in resolution for the MOS fibres, and 6% for the IFU fibres. Therefore it is reasonable
to still consider designs which may somewhat exceed the 9 m rms specification.


                                             1
                                                                                                140 m MOS fibre
            Resolution degradation factor




                                            0.9




                                            0.8
                                                                                           85 m IFU fibre



                                            0.7




                                            0.6
                                                  0   5                 10                 15                  20
                                                          RMS radius of aberrations / m


         Figure 7.16: Resolution degradation factor (see text) as a function of net rms
         radius of the image aberrations. The two curves are for 140 m and 85 m fibres.
         The individual points are discussed in the next section.

In order to make use of this plot, we need figures for the aberrations produced by the
optics. These vary markedly with wavelength and field position, as shown in the
Tables of rms radii in sections 7.2.2.1, 7.2.2.2, and 7.2.2.4. But if we extract a single
‗typical‘ figure for the two collimators it would be about 4 m for the reflective
system and 9 m for the transmissive system (after dividing by the demagnification
factor 1.733 in both cases). Combining in quadrature with the typical camera rms
radius of about 10 m, we have a final typical value of about 11 m when using the
reflective collimator and 14 m with the fully transmissive system. Bearing in mind
the considerable variation about this value, Figure 7.16 shows that in general there
would be little loss of resolution with the MOS fibres, but there will be areas of the
detector for which the IFU loses substantial resolution, especially at the blue end.

7.2.6.3.2 Imaging performance – computed profiles
In order to assess the effects of aberrations on actual fibre images rather than from
point sources, and to check the validity of the theoretical curves in Figure 7.16, some
numerical experiments have been carried out. These used a facility in Zemax in which
images from a circular object can be computed. The tests have been carried out only


                                                                     78                                      Optical Design
                                                                                                                                                                            AAOmega CoD


                       for the camera (AAOMEGA-CAM-411). The fibre objects were specified as a
                       circular bundle of rays emerging from the pupil, with diameters of 0.01698 for the
                       140 m fibres and 0.01031 for the 85 m fibres. The images were computed for all
                       five wavelengths and three field positions given in Table 7.3. Some representative
                       results are shown in Figures 7.17, 7.18 and 7.19.


                                             Image of 140  m fibre, on axis, 370 nm                                                           Image of 140  m fibre, on axis, 370 nm
                                                           File: Image_6                                                                           File: Image_6. Separation 72 m
                              7.00E-02                                                                                          7.00E-02
  Intensity summed across Y




                                                                                                    Intensity summed across Y
                              6.00E-02                                                                                          6.00E-02

                              5.00E-02                                                                                          5.00E-02

                              4.00E-02                                                                                          4.00E-02

                              3.00E-02                                                                                          3.00E-02

                              2.00E-02                                                                                          2.00E-02

                              1.00E-02                                                                                          1.00E-02

                              0.00E+00                                                                                          0.00E+00
                                         0           50        100         150         200   250                                           0           50         100        150         200   250
                                                                X coord /m                                                                                        X coord /m



Figure 7.17: Profile of a circu lar object, representing a 140 m M OS fibre, imaged by the transmissive camera
at  370 n m, on axis. (A perfect collimator is effect ively assumed.) The image has been summed along Y and
shown as a function of X, in the same way a spectrum would be summed spatially and displayed as a function
of wavelength. Left panel: The dotted line gives the ideal profile, and the solid line gives the computed profile
with aberrations. The imaging quality here is good – the rms spot radius is only 3.7 m. Right panel: The same
profile is shown, with a displaced duplicate and their sum. The displacement is chosen to represent the case in
which these profiles would be just barely resolved, if they represented spectral lines.



                                    Image of 85  m fibre, 9.057 deg off axis, 700 nm                                                 Image of 85  m fibre, 9.057 deg off axis, 700 nm
                                                          File: Image_34                                                                           File: Image34. Separation 44 m
                              1.20E-01                                                                                          1.20E-01
  Intensity summed across Y




                                                                                                    Intensity summed across Y




                              1.00E-01                                                                                          1.00E-01

                              8.00E-02                                                                                          8.00E-02

                              6.00E-02                                                                                          6.00E-02

                              4.00E-02                                                                                          4.00E-02

                              2.00E-02                                                                                          2.00E-02

                              0.00E+00                                                                                          0.00E+00
                                         0           50        100         150         200   250                                           0           50         100        150         200   250
                                                                X coord /m                                                                                        X coord /m



Figure 7.18: As for Figure 7.17, except that fibre diameter is 85 m, the image is in the corner of the CCD
(9.057 off axis) and the wavelength is 700 n m. The rms rad ius for the spot diagram here is 13.9 m.




                                                                                                   79                                                                         Optical Design
                                                                                                                                                                     AAOmega CoD


                                  Image of 85  m fibre, 9.057 deg off axis (X), 700 nm                                          Image of 85  m fibre, 9.057 deg off axis (X), 700 nm
                                                      File: Image_36                                                                        File: Image_36. Separation 52 m
                              1.20E-01                                                                                       1.00E-01
                                                                                                                             9.00E-02
  Intensity summed across Y




                                                                                                 Intensity summed across Y
                              1.00E-01
                                                                                                                             8.00E-02
                                                                                                                             7.00E-02
                              8.00E-02
                                                                                                                             6.00E-02
                              6.00E-02                                                                                       5.00E-02
                                                                                                                             4.00E-02
                              4.00E-02
                                                                                                                             3.00E-02
                                                                                                                             2.00E-02
                              2.00E-02
                                                                                                                             1.00E-02
                              0.00E+00                                                                                       0.00E+00
                                         0       50         100        150       200      250                                           0       50         100         150      200      250
                                                             X coord /m                                                                                     X coord /m



Figure 7.19: As for Figure 7.18, but the image has been rotated by 90 before project ion on to one axis. This
image is very asymmetric, hence the difference with respect to Figure 7.18.
                       These results show the typical ways in which aberrations affect a circular fibre image.
                       Figures 7.18 and 7.19 both refer to an image with rms radius 13.9 m, but due to its
                       severe asymmetry, the detrimental effects are much worse on one axis than the other.
                       Clearly the single rms radius value cannot fully describe such situations. (This is the
                       most significant example of asymmetry seen, and the left panel of Fig. 7.19 shows the
                       worst profile seen, despite others having higher rms radii.)
                       From the right hand panels of Figures 7.17 – 7.19 and other similar plots, the
                       minimum separation for the two profiles to be regarded as resolved was found (where
                       the dip between the peaks is about 70% of the peak) and from this the experimental
                       resolution degradation factor was calculated. These values give the individual points
                       plotted on Figure 7.16, where squares are for 140 m fibres and triangles for 85 m
                       fibres.
                       The plotted points do roughly agree with the appropriate theoretical line, but there is
                       considerable scatter, due largely to the fact that rms radius is not a complete
                       description of the profiles. The largest discrepancies occur for the points representing
                       the asymmetric profile of Figures 7.18 and 7.19 – in one case the empirical resolution
                       degradation is less than the model line, and in the perpendicular case it is greater. For
                       the more typical (less asymmetric) images the agreement with theory is better.
                       7.2.7 Spatial profiles
                       We now consider the separation of the individual fibre spectra, i.e. in the ‗spatial‘
                       direction of the spectrograph. In the case of the MOS fibres it is important that they be
                       clearly separated, with minimal contamination of a spectrum by its neighbours (cross-
                       talk). The specification is that the separation of MOS fibres be 10 pixels, so fitting the
                       400 2dF fibres on to the 4K detector. In the case of the SPIRAL IFU, 512 fibres will
                       be used, but must be fitted into 2048 pixels to allow for Nod & Shuffle, giving a
                       spacing of only 4 pixels.
                       Figures 7.20 and 7.21 show the illumination intensity profile in the spatial direction
                       for both the MOS and IFU cases, but without aberrations. It is clear that the IFU fibre
                       spacing is very close, with no unilluminated pixel between the two spectra.
                       Significant overlap is to be expected.




                                                                                                80                                                                     Optical Design
                                                                                                                            AAOmega CoD



                              1.2

                               1

                              0.8




                Intensity
                              0.6

                              0.4

                              0.2

                               0
                                    0     15    30    45     60     75    90    105 120 135 150 165 180 195
                                                                   Detector position /um

                                            140 um fibres, 1.74x demag, 15 um pixels, 10 pixel separation, no aberrations

           Figure 7.20: Spatial profile showing adjacent fibres for the MOS case. The
           intensity has been integrated across the fibre in wavelength, to give the 1 -
           dimensional spatial profile. The grid line divisions on the horizontal axis
           are at intervals equal to the pixel size, so they may be used to indicate the
           pixelisation of the image. The effects of aberrations are not included in this
           Figure. It is assumed that the objects in the two fibres are of equal
           brightness.




                              1.2

                                1

                              0.8
                  Intensity




                              0.6

                              0.4

                              0.2

                                0
                                    0            15           30           45            60            75           90
                                                                   Detector position /um
                                        85 um fibres, 1.74x demag, 15 um pixels, 4 pixel separation, no aberrations

           Figure 7.21: As for Figure 7.20, but for the 85 m IFU fibres, spaced 4
           pixels apart.

Figure 7.22 shows just two examples to indicate that there will indeed be significant
cross-talk in the spectra from adjacent IFU fibres. This is not nearly as harmful as is
cross-talk for MOS fibres, because in the IFU case adjacent spectra do come from
neighbouring positions on the sky. Some form of spectral extraction similar to that for
spatially extended objects observed with a long slit may be appropriate.




                                                                           81                                               Optical Design
                                                                                                                                                                        AAOmega CoD



                                             Image of 85 m fibre, on axis, 500 nm                                                Image of 85 m fibre, 9.057 deg off axis (X), 700 nm
                                                 File: Image_23. Separation 60 m                                                            File: Image_36. Separation 60 m
                              1.20E-01                                                                                        1.00E-01
  Intensity summed across Y




                                                                                                  Intensity summed across Y
                                                                                                                              9.00E-02
                              1.00E-01                                                                                        8.00E-02
                              8.00E-02                                                                                        7.00E-02
                                                                                                                              6.00E-02
                              6.00E-02                                                                                        5.00E-02
                                                                                                                              4.00E-02
                              4.00E-02                                                                                        3.00E-02
                              2.00E-02                                                                                        2.00E-02
                                                                                                                              1.00E-02
                              0.00E+00                                                                                        0.00E+00
                                         0          50         100         150       200   250                                           0       50         100        150      200      250
                                                                X coord /m                                                                                  X coord /m


Figure 7.22: Spatial profiles of adjacent IFU fibres, with examp le aberrations. The sum of the two
profiles is also shown. For the left-hand panel the rms radius of the spot pattern is 10.2 m, while the
right-hand panel shows the profile with the asymmetric image discussed above, along its less
favourable axis. It has an rms radius of 13.9 m (azimuthal average).




                                                                                                 82                                                                       Optical Design
                                                                       AAOmega CoD




7.3 Dual beam system with Schmidt optics

7.3.1 Origin of the concept
The dual beam system has been advanced by Will Saunders and investigated by Peter
Gillingham, based on the following points:
1. Schmidt optics could be used for both the collimator and camera. Such systems
    give good image quality and are well- understood.
2. There is some beam blockage by the detector and fibre slit at internal foci, but this
    could be minimised by using a large collimated beam diameter and having
    separate blue and red cameras (with the beam split by a dichroic), so each detector
    is only 2K  4K in size.
3. Costs are expected to be lower than for the transmissive system with its many
    large lenses. The possibility of using commercially available Schmidt systems was
    investigated, but it now appears that specially constructed optical elements would
    be required. Even so, they are not particularly expensive. The cost and maximum
    size of the dichroic has emerged as a major driver for the design. The risk
    involved in fabrication of the Schmidt optics would be substantially less than for
    the transmissive system.
4. Splitting the optics into separate blue and red arms allows the coatings and
    detectors to be optimised in each case. This helps the throughput, although the
    additional reflections (compared with the transmissive system) drops it.
7.3.2 Description
The layout of the dual-beam Schmidt system is shown in Figure 7.23. The present
concept for this design has a fibre slit of length 120 mm, containing 400 MOS fibres.
(As for the transmissive system, it is exchangable with the slit from the other field
plate, and the one with 512 fibres from the SPIRAL IFU.) The collimator has a focal
ratio of f/3.15, and produces a collimated beam diameter of 190 mm. The required
dichroic has a diameter of 250 mm, the largest size expected to be readily available.
The camera focal ratio is f/1.57, giving the desired 2 demagnification from fibre slit
to CCD.




                                           83                           Optical Design
                                                                       AAOmega CoD




                                                     Fibre slit (end on)




  Red camera
                                                                   Collimator



                  VPH gratings


                                                           Blue camera




Figure 7.23: Layout of double beam Schmidt spectrograph with 190 mm pupil diameter




                                           84                              Optical Design
                                                                                  AAOmega CoD



7.3.3 Image quality


The imaging quality has been investigated using Zemax spot diagrams. RMS radii of
images, including both the collimator and camera aberrations, are given in Table 7.4.
In the red arm, transmission through the tilted dichroic plate introduces an offset of
the optical axis, which would have to be allowed for in the alignment of the following
components. It does not, however, introduce noticeable astigmatism, because the
beam is almost collimated. The Table shows that the DBSS gives excellent image
quality, and meets the desired specification of 9 m rms radius or less, which will
result in minimal effects on spectral or spatial resolution.
   Wavelength            Blue Arm
      (nm)          Centre of slit         30 mm from centre 60 mm from centre
     365                  7.6                     7.4               8.3
       485                  3.1                       4.8                      4.4
       605                  8.0                       9.3                      9.0
                           Red Arm

       540                  5.7                       6.2                      3.1
       695                  5.1                       3.4                      2.8
       850                  4.2                       5.4                      6.4
 Table 7.4: RMS radii (m) of spot diagrams for the b lue and red arms of the DBSS. The
 optical tracing includes the effects of collimator, d ichroic and camera. The rad ii have been
 computed using the dithered spot pattern, density 7, referred to the centroid.


7.3.4 Throughput efficiency


Figure 7.24 shows the throughput of the optical system for the DBSS. It may be
compared with Figure 7.9 for the transmissive system. The obscuration loss due to
both the fibre slit assembly in the collimator and the CCD+mount in the camera has
been examined by Peter Gillingham. For the present design with a collimated beam
diameter of 190 mm, the loss is 9% for fibres near the centre of the slit, rising to 11%
at the end of the slit. The CCD and field flattener actually occupy a larger fraction of
the beam than this, but the severity of the light loss is reduced by the partial overlap of
the Schmidt obstruction with the telescope‘s central obstruction (which does partially
survive FRD in the fibre). The losses would be more severe in the case of the IFU
fibres, because they do not benefit in the same way from overlapping of the Schmidt
and telescope obstructions, and the output beam from SPIRAL is slower than that
from the MOS fibres (so a larger fraction will be blocked by the DBSS obstructions).




                                                 85                                  Optical Design
                                                                                AAOmega CoD


                        1
                       0.9
                       0.8
                       0.7
          Throughput
                       0.6
                       0.5
                       0.4
                       0.3
                       0.2
                       0.1
                        0
                             300   400   500    600     700     800       900      1000
                                               Wavelength /nm

       Figure 7.24: Throughput of the optics for the double beam Sch midt design. These
       curves give the throughput product of: corrector+fibre  collimator  dichroic
       (reflectiv ity in blue, transmission in red)  obscuration at slit centre in both
       collimator and camera  camera optics. The lens surfaces are assumed coated with
       MgF2 +sol-gel, and the camera mirror for the red arm is assumed to be silver, while
       the blue camera and the collimator mirrors are assumed to be aluminiu m. Note that
       these curves apply to MOS fibres but not for IFU fibres. The throughput at the blue
       end might be improved somewhat by tuning down the MgF2 +sol-gel coatings, but
       the major loss there is due to the corrector + fibre.

For the purpose of illustration and to facilitate comparison with the transmissive
design, we will use the same example curves for efficiency of VPH gratings. In
practice, slightly different gratings would be used, because of the larger beam
diameter. Figure 7.25 shows the system throughput, and can be compared with Figures
7.12a and 7.12 b for the transmissive system. It can be seen that over most of the range
the throughput of the Schmidt system is about 90% of that of the transmissive system
(e.g. 0.5 at peak as opposed to 0.55). Thus there is a small sensitivity penalty in the
Schmidt system, but overall the systems can be regarded as comparable in throughput.
The Schmidt system is slightly superior in the blue when compared with the
transmissive system using the Fairchild CCD.

For reference, Figure 7.26 shows the corresponding curves when using the Fairchild
CCD, and may be compared with Figure 7.12b for the transmissive system. However,
the Fairchild CCD would not be preferred for the DBSS because separate detectors
must be used in each of the blue and red arms. A size of 2K  4K is appropriate for
each one, and overall throughput is improved by having separately optimised blue and
red detectors, i.e. the EEV and MITLL detectors.




                                                   86                             Optical Design
                                                                                               AAOmega CoD



                           1
                          0.9
                          0.8
                          0.7
           Transmission   0.6                                                                 722_axis
                          0.5                                                                 722_edge
                          0.4                                                                 720_a
                                                                                              720_b
                          0.3
                                                                                              1020_a
                          0.2
                          0.1
                           0
                                300   400   500      600    700     800     900     1000
                                                   Wavelength /nm

         Figure 7.25: Examp les of transmission of the DBSS system for MOS fibres,
         including all co mponents except the atmospheric and telescope primary mirror
         losses. The two curves for the blue arm differ only in that the slightly lower one is
         for a fibre at the end of the slit. They assume the EEV CCD and the 722 l/ mm
         VPH grating (Fig. 7.10). The curves for the red end assume the MITLL CCD and
         720 or 1020 l/ mm grat ings of Figure 7.10. The two curves 720_a and 720_b
         assume different coatings for the red camera mirror; the one with silver g ives
         slightly better throughput than the one with alu miniu m.

                           1
                          0.9
                          0.8
                          0.7
           Transmission




                          0.6
                          0.5                                                                    720_c
                          0.4                                                                    722_f
                          0.3
                          0.2
                          0.1
                           0
                                300   400    500      600    700      800     900      1000
                                                    Wavelength /nm

         Figure 7.26: Examp les of transmission of the DBSS system for MOS fibres,
         including all co mponents except the atmospheric and telescope primary mirror
         losses. In this case both blue and red arms are assumed to use the Fairchild CCD.
         The obstruction losses are for on-axis fibres, and a silver mirror is assumed for the
         red camera. Gratings are 722 l/ mm in the blue and 720 l/ mm in the red (Figure
         7.10).

7.3.5 Pros and cons of the Schmidt system

We note here some of the significant differences between the double Schmidt and
transmissive systems.
1. The external focus of the transmissive system allows detectors to be readily
    changed and upgraded, whereas the detectors are built into the Schmidt system and
    would in reality be unchanged for many years.
2. The Schmidt system splits the light into separate blue and red arms. For low
    dispersion spectra these can be assembled to give the same total of 4K pixels in the


                                                             87                                 Optical Design
                                                                            AAOmega CoD


     spectral direction. But for higher dispersion work, where the desired wavelength
     range is entirely within either the blue or the red range, only 2K pixels will be
     available. Some wavelengths of interest will also fall in the dichroic transition
     region and will therefore suffer in efficiency and convenience. Separate arms do
     allow the optics to be specifically optimised for the appropriate range, although
     there is no great difference in the throughputs.
     One way to circumvent the restriction to 2K contiguous pixels at higher dispersion
     would be to have a single-beam Schmidt system, with a 4K  4K detector.
     However, obstruction losses or large beam diameters become a significant issue.
3.   The throughputs of the two systems for MOS fibres are comparable, with a small
     advantage for the transmissive system. For IFU fibres, the central obstruction of
     the Schmidt system will have a more severe effect, and a significant throughput
     disadvantage is expected. This remains to be investigated.
4.   The Schmidt system offers low aberrations. The good image quality will produce
     no significant loss of spectral resolution. In contrast, the transmissive system has
     quite significant aberrations, especially when using the transmissive collimator as
     well as transmissive camera. While not affecting MOS operation unduly, they
     would reduce spectral resolution in IFU mode, and the PSF would vary with both
     wavelength and fibre location.
5.   The spectra of adjacent fibres will be better separated in the DBSS case because it
     can achieve a demagnification factor of 2 from fibre slit to CCD, as opposed to
     1.733 for the transmissive system. In combination with the better imaging quality
     of the DBSS, the spatial separation of adjacent fibres would be markedly
     improved, especially for the IFU.
6.   The larger demagnification factor of the DBSS will result in a 15% larger
     wavelength range on the detector. This advantage is diminished to some extent by
     the drop-off in VPH throughput towards the edges of any wide wavelength range.
7.   The dual-beam system split by a dichroic does not require any order-sorting filters,
     since the dichroic essentially carries out this function. The transmissive system
     would require such a filter when working at higher dispersion in the red. With no
     filters, the DBSS system could dispense with the mechanism for focussing the
     collimator by axial motion of the slit, replacing it with an adjustable preset system.
8.   The fabrication of the Schmidt optics is expected to be easier and cheaper than for
     the transmissive optics. For the latter, the off-axis reflective/refractive collimator is
     preferred with respect to its image quality, but has tolerances which may be
     impractical in manufacture and alignment. The transmissive camera presents a
     challenge in fabrication and mounting, including the interfaces between glasses
     with different thermal expansion coefficients. In the Schmidt system the most
     difficult part is probably the large corrector which would also have to serve as the
     vacuum window for the CCD dewar.




                                              88                             Optical Design
                                                                       AAOmega CoD




8 Mechanical Design

8.1 Introduction
The AAOmega is envisaged as a bench mounted spectrograph fed by fibres from the
2dF field plates, and by the SPIRAL integral field unit, IFU, located at the Cassegrain
focus of the AAT.

8.1.1 Two contending designs

Two fundamentally different designs are considered here: a double beam Schmidt
spectrograph, DBSS, with Schmidt collimator and two Schmidt cameras fed from a
dichroic beam splitter; and an all-refractive (dioptric) design with transmissive optics
for both the collimator and the single camera.

Both designs are expected to use VPH gratings as the dispersive elements, in the case
of the DBSS design, a VPH in each of the red and blue beams. However, as the design
evolves it is possible that conventional transmission gratings or grisms may be used,
especially in low resolution configurations. Where VPH gratings are specifically
mentioned, the designs are also applicable to conventional gratings.
8.1.2 Location and Environment

By opting for a bench mounted spectrograph the effects of structural flexure are
minimised, as the gravity vector remains fixed relative to the spectrograph geometry.
Locating the spectrograph on the upgraded South Catwalk of the AAT allows fibre
feeds from both the prime (2dF) and Cassegrain foci. Enclosing the spectrograph in a
cool-room like environment will thermally stabilise the instrument.

8.2 Double Beam Schmidt Spectrograph DBSS

8.2.1 General description
The optical design of the double beam spectrograph has been described in Section 7.
The collimator, and each of the cameras are constructed as separate modules which
are mounted on a commercial optical table. The dichroic beam splitter is mounted
within the collimator module, the red and blue beams exiting the module via
individual Schmidt corrector plates.

With the selection of VPH gratings as the dispersers, the collimator to grating angle
must be tuned for optimum efficiency, and the collimator to camera angle
(articulation angle) must be selected to tune wavelength. We use stacked commercial
rotary stages to control the two angles.

The cameras are evacuated Schmidt designs, each containing a 2k  4k detector. At
the request of the telescope operations staff, we opt for liquid nitrogen cooling.


                                           89                       Mechanical Design
                                                                                AAOmega CoD



Figures 8.1 and 8.2 show the overall layout of the spectrograph.




Figure 8.1 General v iew o f Double Beam Sch midt Spectrograph. The large green canisters are the
(sectioned) Schmidt camera vacuum vessels. The orange and blue rectangular objects with handles are
the VPH gratings in their holders. Beneath the VPHs can be seen the rotary stages (orange and red).
The red stage determines the grating angle, and the orange stage determines the articulation angle.



8.2.2 Slit area
8.2.2.1 Slits
Four slits are required for the spectrograph: two slits, each fed by 400 fibres from a
2dF field plate; one slit fed by 512 fibres from the SPIRAL IFU located at a
Cassegrain focus and a long slit uniformly illuminated by an incandescent source.
The 400 fibres from a 2dF plate are spaced approximately 10 pixels apart at the
detector, and the IFU fibres, approximately 4 pixels apart.

To facilitate fibre handling and maintenance, the fibre slits will be assembled from a
number of slitlets, as has proven successful in 2dF.

8.2.2.2 Location within beam
One of the drawbacks of the Schmidt collimator topology is the location of the slit in
the beam. The obscuration induced by the slit and its support hardware reduces the
throughput of the collimator. The amount of obscuration introduced is discussed in
Section 7.3



                                                 90                         Mechanical Design
                                                                                     AAOmega CoD




Figure 8.2          A general view of the DBSS fro m another direction. Within the sectioned camera
shells we can see the focus ring, spider and detector package (red,) including the suppo rt hardware and
the field flattener lens



8.2.2.3 Back-illumination
As with 2dF, illumination of the fibre slits will be accomplished using linear arrays of
high intensity LEDs. The light from these will be focused on the fibre slit using a
cylindrical lens – possibly cut from a simple glass stirring rod - as the optical
requirements are not onerous. Light proofing the back- illuminated slit area will
follow 2dF practice with a retracting shield enclosing the end of the slit to prevent
leakage of light into the rest of the spectrograph.

With the proposed slit interchange mechanism (see 8.2.2.5) it is feasible that a single
back illumination station will service both plates. This decision will be made at a
later stage in the design.
8.2.2.4 Slit clamping
To ensure accurate registration of the observing slit at the collimator focus, it is
clamped to a set of kinematic seats. This implies a wobbly attachment of slits to slit
interchange mechanism so that kinematic location in the observing slit position is not
impeded. Clamping will be carried out by a short stroke pneumatic cylinder.

8.2.2.5 Slit interchange mechanism
The proposed scheme may be likened to a four barrel Gatling gun for photons. Four
slits are mounted on the periphery of a rather skeletal cylinder which rotates about an


                                                   91                            Mechanical Design
                                                                                      AAOmega CoD


axis parallel to the collimator optical axis. The arrangement is compact, easy to drive,
and produces minimal obstruction of the beam.

In the 2df positioner the bundles of fibres are twisted along the rotation axis to
accommodate the rotation of the tumbler cross. Generally, as long as the individual
fibres have some freedom to move relative to one another this is quite gentle on the
fibres and has caused no damage. For the slit interchange mechanism of AAOmega a
similar scheme is proposed, with the fibres routed along the rotation axis of the
mechanism, and allowed to twist (over some length) to accommodate the motion.
Clearly, mechanism rotates through less than 360.




Figure 8.3          A view of the collimator showing the slit interchange mechanism. The slits are the
yellow ob jects. The green star-like wheel structure is the slit interchange mechanis m. The observing
slit is the one furthest from the viewpoint, largely obscured by the wheel. The mirro r is seen at the left
rear, and the dichroic beam-splitter is at right front, approximately parallel to the paper. One of the
Schmidt correctors may be seen at furthest right front.



8.2.2.6 Shutter
The shutter function will be deployed immediately in front of the slit by a pneumatic
cylinder or rotator. Further light attenuation will be provided by the simultaneous
deployment of both Hartmann shutters. See section 8.2.5.
8.2.3 Filters
The presence of the dichroic in the optical paths of the DBSS removes the need for
order sorting filters, and the associated filter wheel, drive mechanism, detent and
encoding.
8.2.4 Collimator
The present design for the collimator is an on-axis Schmidt design with a spherical
mirror, followed by a dichroic beam splitter and a pair of correc tor plates, one for


                                                    92                             Mechanical Design
                                                                                     AAOmega CoD


each of the red and blue beams. The observing slit is located within the collimator at
the centre of the beam.
8.2.4.1 Focusing
With no interchangeable order sorting filters, the function being effectively carried
out by the dichroic beam splitter, there should be no need to focus the collimator, if
slit positioning is repeatable. Nevertheless, the collimator may be focused by axially
translating the mirror. The moving assembly is mounted on flexures and moved
under the control of an encoded, motorised micrometer. As there is no need for
operational focus adjustment, software control of this feature need not be available
from the operational interface.




Figure 8.4         Another view of the collimator module. Note the Hartmann shutter flags, the (red)
motor for slit interchange, and the slit back illu mination system (blue bo x). Note how a single station
may serve both 2dF slits. The observing slit is behind the aperture in the brown plate, but is largely
obscured by one of the shutters.


The range of adjustment required is no more than 1mm, corresponding to 250um
change in focus at the detector, though the mechanism should be capable of fine
adjustment, as the collimator is fairly fast (f/3.14). Resolution of ~5 micrometre is
readily obtained with a motorised micrometer.

It is not mechanically practical to individually focus the red and blue collimated
beams.




                                                   93                            Mechanical Design
                                                                        AAOmega CoD


8.2.4.2 Construction
Unlike the cameras, the collimator is not evacuated so it will be constructed on an
aluminium plate base. The general construction approach will follow that of the
SPIRAL spectrograph in that the mirror, the dichroic beam splitter, and the Schmidt
plates will be mounted on ‗billboard‘ style mounts, that is vertical plates braced by
gussets from the base plate. These have proven rigid and accurate in service in
SPIRAL.
8.2.5 Hartmann Shutters
Two Hartmann shutters will be provided in the near-collimated space within the
collimator to assist in focussing the spectrograph. These will be inserted into the
beam by pneumatic cylinders and their operation sensed by limit switches.
8.2.6 Dichroic
A dichroic filter splits the collimated beam into red and a blue beam, which are
directed to their respective dispersers and cameras. The dichroic, as a reflective
element (in the blue) must be carefully specified for flatness.

Dichroic beam-splitters with different characteristic wavelengths may be
interchanged. The changeover will be accomplished manually. To minimise
disruptions to the thermal stability of the environment, this will not be carried out
during the observing night.
8.2.7 Dispersers
The dispersers of choice for the AAOmega spectrograph are volume phase
holographic gratings, VPHs, which exhibit greater efficiencies than conventional
ruled transmission gratings and grisms at high resolution. They are also available, to
order, in a wide range of line frequencies. A range of these is envisaged for
AAOmega.

Traditional ruled gratings will be used for low resolution configurations.
8.2.7.1 Interchanging gratings
VPH gratings will be interchanged manually. (As the procedure will involve opening
the environmental enclosure, it should not normally be carried out during the
observing night – indeed it is best carried out at the end of the previous night‘s
observing.to allow maximum time for the spectrograph to regain thermal
equilibrium.) The mounted gratings will be located on kinematic seats to ensure
repeatability of location.
8.2.7.2 Grating rotation
For VPH gratings, to first order, the central wavelength is a function of the collimator-
to-camera angle (camera articulation angle), rather than a function of grating- to-
camera angle. (The grating-to-camera angle must be controlled, however, to obtain
optimum grating efficiency.) The accuracy required of the grating rotation angle is
thus, comparatively low and the Newport RV120CC rotation stages will easily
accomplish this. These stages come with integrated encoders reading to 0.01 degree.
This may be compared with the values determined in section

Grating rotation angles are set independently to optimise the throughput of the red and
blue beams.


                                            94                       Mechanical Design
                                                                         AAOmega CoD


8.2.7.3 Identification
Mounted VPHs can carry simple identification plates which allow identification of the
installed gratings via the control screens in the control room. A set of four proximity
switches will sense the presence or otherwise of holes in the plate. This simple binary
encoding scheme will identify up to 15 VPHs, and the absence of a grating. The
presence or absence of the holes may be sensed by Hall effect switches.
8.2.8 Cameras

8.2.8.1 Red and Blue cameras
Two cameras are required in the DBSS, nominally a red and a blue camera, which
receive the red and blue beams, respectively, from the dichroic. As the Schmidt
design is essentially achromatic, the cameras differ only in minor details.

8.2.8.2 Schmidt Corrector as Pressure Window
The Schmidt cameras must be evacuated to house the cooled detectors. As a
consequence, the corrector plates must be designed to resist a pressure differential of
100 kPa. As the allowable tensile stresses for exotic materials (e.g. CaF2) are not
readily available, fused silica is the clearly preferred material, from a mechanical
point of view, for the vacuum window.

For a relatively flat plate of 200mm diameter in fused silica, the plate thickness of the
current design, 12mm, safely withstands the atmospheric pressure loading, producing
a tensile stress of 8.3MPa, giving a factor of safety of ~5.9.

The vacuum loading on the window/corrector changes the shape of the corrector.
Although the plate is relatively flat and the local thickness is unchanged, the optical
performance of the corrector should be analysed after FEA determines its deflected
shape.
8.2.8.3 Obscuration
As for the Schmidt collimator, the Schmidt camera design suffers from obscuration, in
this case caused by the detector and its support hardware. The EEV detectors are
constructed for connection via the rear to minimise the blockage. Other detectors
must be assessed.
8.2.8.4 Camera articulation
The articulation angles for the red and blue cameras will be determined and set
independently. The angles will be determined by Newport RV160CC rotation stages
which are capable of accuracy of 0.023, 0.003 repeatability and 0.001 resolution.
(These values may be compared with those determined in section 7.2.4.3.2.) The
stages incorporate encoders. Two rotation stages are required, one for each camera‘s
articulation.
8.2.9 Detectors
8.2.9.1 Mounting
The detectors will be mounted with 2k pixels in the spectral direction. A field
flattener will be mounted immediately in front of the detector and will move with it
during camera focusing. The detector assemblies will be carried by spiders which are
themselves mounted on a large focus ring situated outside the beam.


                                            95                       Mechanical Design
                                                                     AAOmega CoD


8.2.9.2 Focus adjustment
Each camera will be focused by moving its detector, using three encoded motorised
micrometers provided for the purpose. Each micrometer set will drive through the
vacuum shell of its camera to reorient the focus ring, which carries the outer ends of
the spider which supports the detector package. This follows the technique used on
the 2dF spectrographs. Piston (axial focus) and two orthogonal tilts are thereby
obtained.

The Physik Instrumente M-230.25 DC-Mikes considered for this application have
resolution of ~50nm and repeatability of 2um.
8.2.9.3 Temperature regulation
The CCDs must be cooled to quite low temperatures, (typically 100K) which will be
determined when they are characterised after delivery. Following well-proven AAO
practice, the individual mounting plates for the detectors will be cooled by liquid
nitrogen, LN2, via resistive cold straps, and temperatures trimmed using heaters fed
by regulator circuits which take their feedback from diode sensors on the mounting
blocks.
8.2.10 Liquid nitrogen dewars
8.2.10.1    Thermal load
The thermal load on the cooling system is estimated to be approximately 16W from
around each detector, and a further 15 W within each LN2 dewar. The nitrogen
consumption is estimated to be 35 litres per day for the whole spectrograph.
8.2.10.2     Capacity and hold time
Two dewars of 300mm diameter, 400mm tall will each hold 13 litres of LN2 and last
16 hours between fills.
8.2.10.3     Articulation with cameras
The LN2 dewars must follow the cameras as they articulate. The dewars will be
carried on their own bearings to unload the camera structure and aid in maintaining
accuracy. Note that these dewars are not shown in Figure 8.1 and Figure 8.2.
8.2.11 Base structure and Anti-Vibration Provisions
8.2.11.1      Base
It is proposed to mount the whole spectrograph on a Newport RS2000 series optical
table. These tables are of deep honeycomb sandwich construction, which produces
very large beam stiffness for a given table mass. The honeycomb ‗meat–in- the-
sandwich‘ tends to damp vibration. The table drawn in Figure 8.1 and Figure 8.2,
1.2m  2.4m is clearly not wide enough. Two articulating cameras take up a lot of
space.
8.2.11.2      Antivibration mounts
Antivibration base leg mounts to suit the optical tables are a lso available from
Newport. We propose to use these to further reduce the susceptibility to vibration.
These legs produce a very low resonant frequency, and have built in damping based
on air driven through a porous plug.




                                          96                      Mechanical Design
                                                                        AAOmega CoD


8.3 Transmissive Camera Spectrograph Design

8.3.1 General description
The transmissive spectrograph design features an all-refractive collimator and a single
all refractive camera. There is no dichroic beam splitter, so that order sorting filters
are provided. A single grating is used, and the focal plane of the instrument is fitted
with a single detector assembly, either a single 4k  4k CCD or two butted 2k  4K
CCDs.
8.3.2 Slit area
As for the DBSS, four slits are required for the spectrograph: two slits, fed by the 2dF
field plates; one slit fed by an IFU located at the Cassegrain focus and a uniformly
illuminated long slit.
8.3.2.1 Location
Unlike in Schmidt designs, the observing slit in the all- transmissive spectrograph is
axially in line with the collimator, and the slits and interchange mechanism are
located outside the beam, causing no obstruction.
8.3.2.2 Back-illumination
Back- illumination of slits will be accomplished in the same way as for the DBSS. See
section 8.2.2.
8.3.2.3 Interchange mechanism
Although there are lesser space constraints in the transmissive design‘s slit, a similar
interchange arrangement is proposed for both designs. The drum arrangement can be
seen in Figure 8.5 and the DBSS arrangement can be seen in Figure 8.3. See section
8.2.2.5.
8.3.2.4 Slit clamping
As with the DBSS design, to ensure accurate registration of the observing slit, it is
clamped to a set of kinematic seats. See section 8.2.2.4
8.3.2.5 Shutter
The shutter will be as described for the DBSS design in section 8.2.2.6.
8.3.3 Filters

8.3.3.1 Filter Wheel
Because the spectrum covers a wide range on the detector, a number of order sorting
filters are required. The four position filter wheel will be driven directly by a DC
motor and gearbox via a lost- motion coupling. The final positioning is accomplished
by engagement of a pneumatically operated detent.
8.3.4 Hartmann Shutters
As in the DBSS, two Hartmann shutters will be provided in the collimated space to
assist in focussing the spectrograph. These will be inserted into the beam by
pneumatic cylinders.




                                            97                       Mechanical Design
                                                                                  AAOmega CoD




Figure 8.5         General view of the transmissive spectrograph design. The large red LN2 dewar can
be plainly seen, linked via bellows to the smaller detector dewar. Note the curved bearing track which
supports the outboard end of the camera, and the LN2 dewar. The green slit interchange drum can be
seen at the front left. The camera and collimator barrels are sectioned to show the optical elements..
The tabletop, as drawn is 1.2  2.4 metres. The red and gold coloured rotary stages are located below
the VPH.



8.3.5 Dispersers
As in the DBSS design VPH gratings are proposed for the transmissive spectrographs.
The transmissive design, however, only requires one disperser - otherwise, all
comments in section 8.2.7 and its subsections are applicable.
8.3.6 Camera and Collimator

In this design, the camera and collimator are all- refractive (transmissive). Although
dimensions differ, the construction techniques required for both are similar. Two
benefits follow directly from a straight through configuration: the camera does not
need to be evacuated or cooled; and there is no obscuration within the camera or
collimator, as there would be with Schmidt reflective designs. The camera and
collimator are tubes fitted with lenses, and may be designed and manufactured in
quite standard fashion. Some vendors may wish to quote for complete assemblies.




                                                  98                           Mechanical Design
                                                                         AAOmega CoD


8.3.7 Camera articulation
The same encoded rotation stage is proposed for this spectrograph as that proposed
for the DBSS – see section 8.2.8.4. As there is only one camera, a single rotator is
used.
8.3.8 Detectors
Two possible detector configurations are possible, utilising either a single 4k  4k
CCD or two butted 2k  4k CCDs. A single controller can read out and control both
of these configurations.
8.3.8.1 Butted detectors
A possible detector configuration uses two 2K  4K EEV CCD 44-82 detectors butted
together on a common temperature regulated plate. A gap, expected to be in the order
of 50 pixels, will appear in the spectrum. This orientation of the CCDs is dictated by
charge shuffle direction on the CCDs. Serendipitously, the use of butted detectors
also allows us to use a red optimised CCD at the red end of the spectrum at the cost of
some potential loss of efficiency at higher resolutions.
8.3.8.2 A single detector
4k  4k CCDs are available which are suitable for the transmissive spectrograph
design. The single detector would be mounted much as the butted 2k  4k detectors,
but no gap in spectral coverage would be seen.
8.3.8.3 Detector dewar
The detectors will be mounted in a small detector dewar which is connected, via
bellows and flexible thermal strap to the LN2 storage dewar. The flexibility of this
connection will allow the detector dewar to be moved to focus the detector, thus
removing the need to move the mass of the storage dewar and its LN2 fill during
detector focus. This area should be prototyped to verify the concept.
8.3.8.4 Field flattener as pressure window
The proximity of the field flattener to the detectors requires that it is a window for the
evacuated dewar. (Placing a window ahead of the field flattener would place it in a
rapidly converging beam, where it would influence image quality.)

The field flattener is 5mm thick LAL7, which is rather thin for its 98mm working
diameter. Stresses of the order of 12 MPa can be expected. To verify the integrity of
the lens, and reduce the risk to the CCDs, the window should be tested as a pressure
window before fitting the detector(s).

The distortion of the field flattener under pressure loading may be the source of
unacceptable reduction in image quality and the optical of the deflected lens ( the
shape determined by FEA) should be analysed.
8.3.8.5 Focus adjustment

Focus adjustment will be accomplished using three encoded, motor driven
micrometers, and will allow for piston (axial focus), and two orthogonal tilts of the
detector assembly. The micrometers will adjust the position of the detector dewar
shell to which the detectors will be attached.




                                            99                       Mechanical Design
                                                                    AAOmega CoD


The Physik Instrumente M-230.25 DC-Mikes considered for this application have
resolution of ~50nm and repeatability of 2um.
8.3.8.6 Cooling and temperature regulation
The detectors are mounted on a common cooled mounting plate and thus are operated
at a common temperature. This temperature will be regulated in the same manner as
that proposed for the DBSS. See section 8.2.9.3.
8.3.9 LN2 Dewar

8.3.9.1 Thermal load
The thermal load on the cooling system is estimated to be approximately 27W from
the detector, and a further 18 W within the LN2 dewar itself. The nitrogen
consumption is thus estimated at 25 litres per day.
8.3.9.2 Capacity and hold time

A dewars of 3500mm diameter, 400mm tall will each hold 17Litres of LN2 and last
16 hours between fills.
8.3.9.3 Articulation with cameras
The LN2 dewar must follow the camera and detector as they articulate. The dewar
will be carried on its own bearings to unload the camera structure and aid in
maintaining accuracy. The bearing track provided for the purpose may be seen
clearly in Figure 8.1
8.3.10 Base structure and Anti-Vibration Provisions
8.3.10.1     Base
As can be seen from Figure 8.5 the transmissive spectrograph sits comfortably on a
2.4  1.2 metre optical table.
8.3.10.2      Antivibration Mounts
The same antivibration mounts are proposed for this spectrograph as for the DBSS.
See section 8.2.11.2.

8.4 Comparison of Spectrograph Designs
Mechanically, the transmissive design spectrograph is less complex, with fewer
detectors, cameras, and dewars to make. However, the optical components are much
simpler in the DBSS design. There are no insurmountable mechanical problems in
either design.

The DBSS clearly requires more space and a larger optical table as the dual
articulating cameras swing through a large volume. This must be considered in
allocating space on the south catwalk.

8.5 Infrastructure

8.5.1 South Catwalk
The AAOmega spectrograph is bench mounted to eliminate deflections. The bench
will be located on a rebuilt South Catwalk. To eliminate vibration effects, the new


                                         100                     Mechanical Design
                                                                        AAOmega CoD


catwalk will be supported by the telescope mount structure, not the dome main floor,
as the telescope structure is isolated from the dome vibrations.
8.5.2 Optical fibre routing
Fibres from the Cassegrain IFU can remain permanently in place, except that the IFU
unit may be disengaged from a particular front end to change configurations.
8.5.2.1 Fibre handling for top end changes
Since the 2dF top end must be removed from the AAT to allow use of Cassegrain
foci, it is essential to remove the 2dF slits from the spectrograph, and coil the fibres
up, attaching them to the top end before lifting it from the telescope. Since the 2dF
fibres are ‗woven‘ into the tumbler, and attached to their respective retractors, they
must remain with the 2dF top end. Some type of interlock system will need to be
provided to ensure the fibres are safe before a top end lift.

To facilitate this fibre handling and disengagement from telescope truss and the
spectrograph, the fibre runs should be surface mounted as far as possible.
8.5.3 Temperature Stable Enclosure
Provision of a well- insulated enclosure for the spectrograph will contribute to a stable
thermal environment, with long time constant. To minimise the disturbance to this
passive system, the opening times for the enclosure (grating change, maintenance, etc)
should be minimised.
8.5.4 Compressed Air
The extensive use of pneumatic devices provides simple, economic and reliable
actuation of many mechanisms on the AAOmega spectrograph. Furthermore, the
solenoid valves used to control them are easy to interface to the control system.

The system is designed to run on dry, oil- free compressed air regulated at 500 kPa,
sourced from the telescope‘s existing supply. A local pressure switch will indicate to
the instrument controller that air is available. The instrument will not operate witho ut
air. As air is only consumed when valves are operated, the consumption will be low.
8.5.5 Compressed N2
Nitrogen gas will be used to flush windows to prevent condensation. The flow will be
low and can easily be supplied by the existing AAT nitrogen system.          A local
pressure regulator and flow valve will control the flow and a pressure switch will be
provided. Nitrogen has been chosen here, above, say, compressed air, for its purity
and the lowered likelihood that it may contaminate the optical surfaces.
8.5.6 Liquid Nitrogen
The consumption of liquid nitrogen is addressed in sections 8.2.10 and 8.3.9.
8.5.6.1 Asphyxiation Hazards
Nitrogen is an odourless asphyxiating gas and the enclosure surrounding the
spectrograph is a confined space with limited ventilation. As nitrogen boils off in the
normal dewar operation it is easily possible for the concentration to exceed safe
limits. At a minimum, a warning notice should be affixed to the enclosure door, and a
safe working routine established. Consideration will have to be given to the effect of
this routine on the thermal stability of the enclosure. A fan should also be provided to
force ventilate the space prior to entry of personnel. Consideration will have to be
given to the effect of this routine on the thermal stability of the enclosure.


                                            101                      Mechanical Design
                                                                      AAOmega CoD


8.5.7 Mains electricity
Ordinary 240VAC mains electricity should be available from GPOs located on the
South catwalk.
8.5.8 Location of Electronics racks
To aid in stabilising the spectrograph‘s environment the electronics racks should be
located outside the enclosure.




                                           102                     Mechanical Design
                                                                       AAOmega CoD




9 Detectors

For both the transmissive and DBSS designs, we require 4k x 4k of detector coverage.
Because of concerns of spectral separation on the detector and optical design
considerations, a pixel size of 15 m was preferred. We are investigating various
detector options, but our choice will not have significant impact on the Concept
Design, as we are primarily considering the intrinsic spectrograph optical design
concept properties. A final decision on the detectors would be made later in the
project, on the basis of performance and cost of the latest CCD's available. Future
devices are expected to improve still further in QE vs wavelength over the current
devices, so it is worth delaying the choice to as late as is feasible. We now separately
discuss the two spectrograph optical concept in term their detector requirements and
currently available CCD devices.

9.1 Transmissive Design
For the transmissive design with single camera, our preference is for a single 4kx4k
CCD. There are few 4kx4k 15 m CCDs on the market (for instance, there is a SITe
4kx4k CCD, but with 12 m pixels; EEV may also produce a 4kx4k CCD, but likely
with 13.5 m pixels). However, we have identified a promising 4kx4k 15 m CCD,
namely the CCD486 produced by Fairchild. This device has excellent QE from 370-
1000 nm (see Figure 9.1), reasonable cost and availability, but the read noise is higher
than we would like (3.5-4 electrons). The QE curve for this device shown in Figure
9.1 is for a broadband coating; different coatings could be applied to optimise for
either the blue or the red. We are currently pursuing further information about this
device.

Another option would be to butt together two 2kx4k CCDs. This would allow some
optimisation for the blue and red ends of the spectrum, but would have the
disadvantage of a gap in the spectral direction (this gap would be ~1mm,
corresponding to 10-15 Angstrom at high dispersion [2500 l/mm grating], and ~60
Angstrom at low dispersion [600 l/mm grating]; the gap would have to be in the
spectral direction, to allow charge-shuffling, which can only be done along columns).
In this case, the two CCDs could either both be blue-optimised or both red-optimised,
or we could have a blue-optimised and a red-optimised device. There are advantages
and disadvantages of both choices. In high dispersion mode, both CCDs would be
needed to give full spectral coverage. In this mode, it would be an advantage to have
both CCDs similarly optimised, and a disadvantage to have them different. In low
dispersion mode, not all 4k spectral pixels are required, and it would be an advantage
to have two different CCDs, with a blue-optimised CCD covering the blue end, and a
red-optimised CCD covering the red end. It would be a disadvantage to have two
similar CCDs, as the throughput would be compromised.

9.2 DBSS Design
For the DBSS design, we would require two 2kx4k CCDs, one blue-optimised for the
blue arm, and one red-optimised for the red arm. In the blue, a 15 m device


                                         103                                  Detectors
                                                                       AAOmega CoD


(CCD44-82) provides the best performance of the available devices. This CCD offers
excellent blue sensitivity, as can be seen from Figure 9.1. Other characteristics (read-
noise, CTE, flatness, etc) are also excellent, as demonstrated by the recently
commissioned EEV2 device on the AAT. In the red, there is the choice of either an
EEV 15 m red-optimised CCD, or a MITLL device (cf. MITLL3, currently in use on
the AAT). Both have excellent sensitivity in the red (see Figure 9.1), but a MITLL
device is preferred at present (lower read noise, fringing, and cost).

9.3 CCD Controllers
For the DBSS design, one controller would be needed for each CCD, since each CCD
would be mounted in a separate dewar. For the Transmissive design, only one
controller would be needed, regardless of whether we have one or two CCDs (and
even if the two CCDs are from different vendors). If we have two CCDS, both could
be read-out simultaneously. Readout times for either 2k x 4k, or 4k x 4k CCDs are
approximately 5 minutes with a single amplifier, in ‗slow‘ readout mode. The new
AAO-II controllers will allow charge-shuffling with one or two CCDs (unlike the case
for 2dF, where charge-shuffling can only be done with one CCD).

Tradeoffs:
 The DBSS design allows both arms to be separately optimised for the blue and the
   red. The transmissive design, with a single 4kx4k CCD would not be as well-
   optimised. However, the Fairchild device offers excellent broad-band QE, giving
   the very good performance below 400 nm, and is within ~10% of a blue or red
   optimised 2kx4k CCD across the whole optical spectrum. Different coatings could
   be used to optimise the Fairchild device for either the blue or the red. A
   transmissive system, with a blue-optimised and red-optimised CCD butted
   together, could offer better performance at low spectral resolution, but with a
   potential cost at higher spectral resolution.
 The DBSS design only allows 2k pixels of continuos spectral coverage at high
   dispersion, but does allow the possibility of simultaneous coverage of blue (2k
   pixels) and red (2k pixels) features (e.g. H/H, Mgb/Ca triplet). The
   Transmissive design allows 4k pixels of continuos spectral coverage at all
   dispersions.
 In the Transmissive design, a single 4kx4k CCD gives uninterrupted spectral
   coverage, while having two 2kx4k CCDs would entail a spectral gap. As
   discussed above, having two different 2kx4k CCDs for the transmissive system
   would allow better optimisation for the blue and the red, but this would be a
   disadvantage at high spectral resolution.
 The read noise for the Fairchild 4kx4k CCD is 3.5-4 electrons, slightly higher than
   for either EEV (3-3.5 electrons) or MITLL (1.5-2 electrons) 2k x 4k devices.
   Calculations show that observations with MITLL or EEV detectors would never
   be read noise dominated for any spectral resolution, for exposures of one hour or
   longer. However, spectra taken with the Fairchild CCD would have read noise
   comparable to sky noise at high spectral resolution (2500l/mm) in the blue,
   especially for the smaller IFU fibres. For bright objects with exposure times of 30
   minutes or less, even the EEV detector would be read noise dominated at high
   spectral resolution in the blue, while the Fairchild detector will have read noise
   comparable to sky even at low spectral resolution; the MITLL detector would still
   be sky- limited. However, given the configuration time of ~1 hour, the overheads


                                         104                                  Detectors
                                                                     AAOmega CoD


    associated with such observations would be high, and they would not be
    commonly done.
   The readout time with a single amplifier (~5 min) for either the 2kx4k or 4kx4k
    CCDs is a significant overhead, especially for beam-switched observations, where
    readouts would be done on ~15 min timescales. However, these devices have two
    or more readout amplifiers; thus, each CCD could be read out more quickly using
    all the amplifiers, with the cost of having different read-noise and gain
    characteristics for different regions of the detector. Such multiple readout could
    still be done for two CCDs with one controller.
   The DBSS design has an internal focus, and changing detectors would be a
    difficult and expensive procedure. The transmissive design, with an external
    focus, would make it much easier to change/upgrade detectors. An internal focus
    also means that the CCD package size has to be as small as possible, to minimise
    obstruction losses.
   The CCD costs are roughly comparable for both the DBSS and transmissive
    designs, being virtually independent of the choice of CCD configuration discussed
    above.




Figure 9.1 Comparison of suitable CCD devices currently available that have suitable
performance and 15m pixels.




                                         105                                Detectors
                                                                       AAOmega CoD




10 Electronics

10.1 Introduction
The electronics systems required for the AAOmega Spectrograph (both options)
include the Instrument Control Electronics for the Spectrograph, and the Detector
Control Electronics for the detectors.


10.2 Instrument Control Electronics
The Instrument Control Electronics includes the following:

   Instrument Control Computer System – this includes a single board computer
    packaged in a chassis with a backplane and power supply. The single board
    computer will include memory, serial communication ports, a local area network
    interface and an input/output bus interface. Additional interface boards in the
    instrument control computer will include one or more motion controllers, one or
    more digital input/output interfaces and an analog input interface.
   Instrument Interface Electronics Chassis – this contains a number of circuit boards
    including interfaces to the instrument control computer, digital input/output
    conditioning electronics, solenoid drive electronics and sensor inp ut electronics.
   Servo Motor Drive Chassis – this contains servo amplifiers for the DC servo
    motors in the instrument, and associated power supplies.
   Power Supply Chassis – this chassis contains the power supplies for the
    Instrument Interface Electronics Chassis and the Servo Motor Drive Chassis.
   Cables, Connectors, Termination Boards and Enclosures – all the components
    necessary to connection the actuators and sensors to the electronics circuit boards.

A block diagram of the AAOmega Instrument Control Elec tronics is shown in Figure
10.2.

10.2.1 Spectrograph Mechanisms and Sensors
The AAOmega Spectrograph (regardless of the design), is expected to include the
following electro- mechanical components, which need to be remotely controlled or
sensed.

10.2.1.1    Servo Motor Driven Mechanism
A servo motor driven mechanism is the most complex type of mechanism used in the
AAOmega spectrograph.

A typical servo motor mechanism is shown in Figure 10.1.




                                          107                               Electronics
                                                                         AAOmega CoD


The motion controller includes an intelligent processor implementing a servo
algorithm, such as a Proportional Integral Derivative loop. On receipt of an
appropriate motion command from the Instrument Control Computer, the motion
controller generates an analog command signal which is amplified by the servo
amplifier to drive the motor. The command signal is generated as a function of the
requested position or velocity from the Instrument Control Computer and the current
position and velocity of the mechanism. The motor is mechanically linked to an
incremental encoder, the output of which is fed back to the motion controller. The
motion controller can derive the relative position and velocity (including direction) of
the mechanism using the feedback from the encoder. The encoder may also provide
an index or reference position output which can be used by the motion controller to
establish a zero position.

The mechanism will also include limit switches which are fed back to the motion
controller to establish the electrical end-of-travel boundaries of the mechanism. If one
of these limit switches is activated, the motion controller will perform braking on the
mechanism to stop its motion as quickly as possible.

Additional limit switches may also be used to indicate the mechanical end-of-travel
boundaries for the mechanism. These limits are fed back to the amplifier and if
activated, cause the amplifier to be disabled. These limit switches would normally
only be activated if the axis runs past the electrical end of travel limit switch for so me
reason.

There may also be a mechanism home switch which is fed back to the motion
controller. This switch may be used to establish a home or park position for the
mechanism.

The motion controller has an enable output which is used to enable the servo
amplifier. If not enabled, the servo amplifier will not respond to analog command
input from the motion controller. The servo amplifier generates a fault output which is
fed back to the motion controller. This could be generated as the result of an over
temperature condition, or a mechanical end-of-travel limit switch activation.

10.2.1.2      DC Motor Driven Mechanism
A DC motor driven mechanism is a simple open loop system in which the motor
drives a load at constant speed to a well defined position, usually set by a position
switch. This is controlled using digital input and output bits converted to suitable
voltage levels by appropriate interface electronics.

10.2.1.3     Pneumatic Valve Solenoid
A pneumatic valve solenoid mechanism is a solenoid driven from a digital output bit
converted to a suitable voltage level by appropriate interface electronics.



10.2.1.4    Sensors
The following types of sensors will be used in the AAOmega spectrograph.



                                           108                                 Electronics
                                                                      AAOmega CoD


-   Microswitches – simple mechanically operated microswitches with Normally
    Closed and Normally Open contacts. These would typically be used for limit
    switches and enclosure status switches.
-   Proximity Sensors – Inductive proximity switches which operate when a metallic
    target close to the sensor completes a magnetic circuit. These may be Normally
    Open or Normally Closed. These would typically be used for limit switches and
    position sensors.
-   Air Pressure Switch – operated when the desired air pressure is reached. This is
    required to indicate that there is sufficient air pressure available to operate the
    pneumatic components.

In general, limit switches will be of the Normally Closed type and position switches
will be of the Normally Open type. Position sensor switches will be connected to
digital inputs by appropriate interface electronics.

10.2.2 Instrument Control Computer System
The AAOmega Spectrograph requires a control computer mounted near the
spectrograph assembly to provide intelligent remote control for the spectrograph
mechanisms.

A block diagram of the AAOmega electronics control system is shown in Figure 10.2.

The main requirements of the control computer system are that it must:
 Support an open bus architecture for which various types of interface boards can
   be readily purchased (or designed if necessary).
 Have adequate interface expansion capacity for the installation of all the interface
   boards needed for controlling the AAOmega Spectrograph.
 Be capable of being mounted in a standard 19- inch rack.
 Be able to run a real-time operating system, preferably without a locally
   connected disk drive.
 Be powerful enough to carry out the desired functions of the AAOmega
   Spectrograph without compromising the desired system performance.
 Be able to be connected to a Local Area Network and be able to be remotely
   operated and maintained via the Local Area Network.
 Have RS232 serial ports for connection to stand alone devices and a console
   terminal.

A block diagram of the AAOmega Control Computer System is shown in Figure 10.3.

There are three possibilities for the type of control computer system which can be
used to fulfil these requirements:

1. An Intel x86 system running a diskless version of an embedded or real-time
   Linux. This would not be a standard desk-top PC, but instead, would be housed in
   an industrial PC chassis, with a large number (~10) of PCI bus slots and possibly
   some ISA bus slots. The Linux operating system would be loaded and run out of
   a Flash disk and/or using network file system disks (the disks being located on an
   AAT Sun workstation). It would have sufficient memory to support the use of
   RAM disks.


                                         109                               Electronics
                                                                    AAOmega CoD



    Advantages:
     There is a wide availability of interface boards for the PCI bus.
     The cost of PCI bus interface boards is relatively low, and while the cost of
       industrial PC systems is higher than desktop systems, they are less expensive
       than VMEbus or CompactPCI bus systems.
     Low cost for operating system and development tools.
     Development can be done using low cost desk-top PCs, and can be more
       easily shared by multiple developers.
     Linux device drivers exist for many interface devices.

    Disadvantages:
     There could be some effort required to get diskless real-time Linux
       operational.
     Device drivers may need to be developed for some interface boards.
     Adapted from desktop use to industrial use.
     Long term reliability may be an issue.
     Hardware may become obsolete quickly, making replacement or repair
       difficult.

2. VMEbus or CompactPCI system running the VxWorks real time operating system
   on a single board computer (which may be PowerPC or UltraSPARC based). This
   system would boot VxWorks over the Local Area Network from a Sun host
   computer.

    Advantages:
     Proven real-time architecture and cross development environment.
     Cross development is carried out on the Sun host computer system.
     Proven operating system reliability and performance.
     Proven diskless/network architecture.
     Probably more reliable in long term.
     Purpose designed and engineered for industrial use.
     Adequate expansion capability for all required interfaces.
     May be able to procure device drivers for some devices.

    Disadvantages:
   Higher cost for hardware and software.
   May require more space (6U rack + 1U Fan Tray).
   May need to develop device drivers for some devices.

3. A VMEbus or CompactPCI system running diskless Linux. This system
   combines the high reliability of engineered industrial hardware with the low
   expense of Linux.

It is felt that at this stage of the project it is not necessary to make a final
recommendation on the type of computer system to be used. A firm recommendation
will be made after further development work and testing by AAO to configure a PC
system running diskless real-time Linux during the Preliminary Design Phase. Such a
decision would also need to take into account compatibility issues with the AAT


                                        110                               Electronics
                                                                      AAOmega CoD


instrument spare parts inventory, and the skills base of the Site Technical Support
staff.

The AAOmega Control Computer System will include the following interfaces:
 One or more digital input/output interface(s) to control Spectrograph pneumatic
   functions, Spectrograph DC motor driven (open loop) mechanisms and to read the
   status of various Spectrograph mechanisms. The number of interface boards
   required depends on the number of input and output bits needed to satisfy the
   Spectrograph requirements and the number of bits which are available on the
   available boards. However, many makes of boards support 32-bits of input/output
   capability and it is likely that this would be insufficient for both Spectrograph
   designs. Therefore either two 32-bit digital input/output interfaces or a single 64-
   bit interface are the likely possibilities for AAOmega.
 A multi-channel analog to digital interface for telemetering various variables,
   including Spectrograph and enclosure temperature, power supply voltages etc.
 Two or three 4-channel servo motion controllers for the Spectrograph servo
   mechanisms. Motion controllers are generally available as 1, 2, 4 or 8 channel
   units. It is felt that given that the Transmissive design requires 6 servo channels,
   and the DBSS design requires 11 servo channels, and that the AAOmega spares
   inventory should be minimised, the optimum number of channels per motion
   controller should be 4. Based on this, the Transmissive design requires two 4-
   channel motion controllers (plus one spare), while the DBSS design requires three
   4-channel motion controllers (plus one spare).

The choice of motion controller will be made following the selection of the type of
control computer. In addition to the requirements of the Spectrograph servo systems,
the choice of motion controller will also depend on the availability of a device driver
for the selected control computer operating system or the availability of low level
hardware information to allow device driver development. It must be noted that if a
motion controller that has not been previously used at the AAO is selected for the
AAOmega application, it will be necessary to prototype a single servo axis during the
Preliminary Design phase. As a minimum, the motion controller should meet the
following specifications:

-   available as a four channel device compatible with the Control Computer system
    bus
-   device driver available for Control Computer operating system, or device level
    information available to permit driver development
-   Capable of stand-alone operation to assist with development and debugging
-   Intelligent built- in servo loop algorithm (such as PID)
-   10V analog command output for each channel
-   Differential TTL compatible quadrature encoder inputs for each channel,
    including a reference position signal
-   end of travel limit switch inputs for each channel
-   home or reference switch input for each channel
-   amplifier enable output and amplifier fault input for each channel
-   optically isolated inputs and outputs




                                         111                               Electronics
                                                                         AAOmega CoD


10.2.3 Instrument Interface Electronics Chassis
The Instrument Interface Electronics Chassis contains the signal conditioning
electronics necessary to interface the Instrument Control Computer to the various
Spectrograph electro- mechanical controls and sensors.

It is envisaged that the Instrument Interface Electronics Chassis would be packaged as
a 3U high 19- inch wide Eurocard rack, with a number of circuit boards plugging in to
a common backplane. Cable connections to the rack would be made using connectors
mounted either on the front panels of the boards or on the rear panel of the rack.
Connector and cable types will be decided in the Preliminary Design phase.

The following types of interface boards will be located It is envisaged that this chassis
will contain a number of different types of one or more interface boards:

   Eight sensor input interface board – supports up to 8 sensor inputs such as micro-
    switches or proximity switches to be read by the Instrument Control Computer.
    All sensor inputs will be galvanically isolated between the sensor input and the
    digital logic signal.
   Eight solenoid driver output board – allows up to 8 solenoid devices to be
    controlled by the Instrument Control Computer. All solenoid outputs will be
    galvanically isolated between the solenoid output and the digital logic signal.
   Two LED Array driver output board – allows 2 back illumination LED arrays
    devices to be controlled by the Instrument Control Computer. All LED control
    outputs will be galvanically isolated from the digital logic signals.
   Single DC motor control board – allows a single multi-position open loop DC
    motor to be controlled by the Instrument Control Computer. All motor control
    outputs and sensor inputs will be galvanically isolated from the digital logic
    signals.
   Digital input/output interface board – this board interfaces the digital logic signals
    in the Instrument Interface Electronics Chassis to the Instrument Control
    Computer digital I/O interface.
   Analog input conditioning board – this board conditions analog inputs from the
    Spectrograph.
   Analog input interface board - this board interfaces the analog signals in the
    Instrument Interface Electronics Chassis to the Instrument Control Computer
    analog input interface.
   Instrument Interface Electronics Chassis Backplane – this is a monolithic
    backplane into which all boards in the chassis are plugged. It provides
    interconnects between boards and power to boards. If the Instrument Connectors
    are mounted on the chassis backplane, all sensor inputs and actuator outputs will
    be provided on the backplane.

The Solenoid driver board will include an external trigger input, to allow an external
device (e.g. a CCD Controller) to trigger a solenoid (e.g. the shutter). For test
purposes, the Instrument Control Computer will also be able to operate the shutter.

The LED driver board will include an external trigger input to allow an external
device (e.g. the 2dF Robot Positioner) to turn on the LED driver. For test purposes,




                                          112                                 Electronics
                                                                     AAOmega CoD


the Instrument Control Computer will also be able to operate the LED driver. The
LED driver circuit will be implemented as a current source.

The numbers of boards required depends on the Spectrograph design:

Transmissive Design
Spectrograph Function       Electro- mechanical       Interface Board
                            Control
Fibre Slit Interchange      DC Motor and Limit DC motor control board
                            Switches
Observing Slit Clamp Pneumatic Valve Solenoid Solenoid driver board
Actuator
Back Illumination Slit A Pneumatic Valve Solenoid Solenoid driver board
Actuator
Back Illumination Slit A Proximity                 or Sensor input board
Position                    Microswitches (2)
Back Illumination Slit A LED Array?                   LED driver board (with
LEDs                                                  input from 2dF Robot)
Back Illumination Slit B Pneumatic Valve Solenoid Solenoid driver board
Actuator
Back Illumination Slit B Proximity                 or Sensor input board
Position                    Microswitches (2)
Back Illumination Slit B LED Array                    LED driver board (with
LEDs                                                  input from 2dF Robot)
Shutter Actuator            Pneumatic Valve Solenoid Solenoid driver board
                                                      (with input from Detector
                                                      Controller)
Shutter Position            Proximity              or Sensor input board
                            Microswitches (2)
Filter Wheel                DC Motor and Position DC motor control board
                            Switches
Filter Wheel Detent         Pneumatic Valve Solenoid Solenoid driver board
Hartman        Shutter   A Pneumatic Valve Solenoid Solenoid driver board
Actuator
Hartman        Shutter   A Proximity               or Sensor input board
Position                    Microswitches (2)
Hartman        Shutter   B Pneumatic Valve Solenoid Solenoid driver board
Actuator
Hartman        Shutter   B Proximity               or Sensor input board
Position                    Microswitches (2)
VPHG Identification         Hall Effect or Proximity Sensor input board
                            Switches (4)
Camera Brake Actuator       Pneumatic Valve Solenoid Solenoid driver board
Camera Brake Position       Proximity              or Sensor input board
                            Microswitches (2)
Pneumatic System Air Air Pressure Switch              Sensor input board
Pressure
Spectrograph      Enclosure Microswitch (2?)          Sensor input board
Open


                                       113                              Electronics
                                                                 AAOmega CoD



DBSS Design
Spectrograph Function        Electro- mechanical       Interface Board
                             Control
Fibre Slit Interface         DC Motor and Limit DC motor control board
                             Switches
Observing Slit Clamp Pneumatic Valve Solenoid Solenoid driver board
Actuator
Back Illumination Slit A Pneumatic Valve Solenoid Solenoid driver board
Actuator
Back Illumination Slit A Proximity                  or Sensor input board
Position                     Microswitches (2)
Back Illumination Slit A LED Array                     LED driver board (with
LEDs                                                   input from 2dF Robot)
Back Illumination Slit B Pneumatic Valve Solenoid Solenoid driver board
Actuator
Back Illumination Slit B Proximity                  or Sensor input board
Position                     Microswitches (2)
Back Illumination Slit B LED Array?                    LED driver board (with
LEDs                                                   input from 2dF Robot)
Shutter Actuator             Pneumatic Valve Solenoid Solenoid driver board
                                                       (with input from Detector
                                                       Controller)
Shutter Position             Proximity              or Sensor input board
                             Microswitches (2)
Hartman        Shutter     A Pneumatic Valve Solenoid Solenoid driver board
Actuator
Hartman        Shutter     A Proximity              or Sensor input board
Position                     Microswitches (2)
Hartman        Shutter     B Pneumatic Valve Solenoid Solenoid driver board
Actuator
Hartman        Shutter     B Proximity              or Sensor input board
Position                     Microswitches (2)
Blue VPHG Identification Hall Effect or Proximity Sensor input board
                             Switches (4)
Blue      Camera       Brake Pneumatic Valve Solenoid Solenoid driver board
Actuator
Blue      Camera       Brake Proximity              or Sensor input board
Position                     Microswitches (2)
Red VPHG Identification      Hall Effect or Proximity Sensor input board
                             Switches (4)
Red      Camera        Brake Pneumatic Valve Solenoid Solenoid driver board
Actuator
Red      Camera        Brake Proximity              or Sensor input board
Position                     Microswitches (2)
Pneumatic System Air Air Pressure Switch               Sensor input board
Pressure
Spectrograph       Enclosure Microswitch (2?)          Sensor input board
Open


                                      114                            Electronics
                                                                       AAOmega CoD



The following board numbers and types would be required for the Transmissive
Design:

Board Type                                       Number       Available   Used
                                                 of Boards    Inputs or   Inputs or
                                                              Outputs     Outputs
8-input sensor board                             3            24          19
8-output solenoid driver board (1 external       1            8           8
trigger input)
DC motor control board                           2            2           2
LED Driver board (2 external trigger inputs)     1            2           2

The following board types and quantities would be required for the DBSS Design:

Board Type                                       Number       Available   Used
                                                 of Boards    Inputs or   Inputs or
                                                              Outputs     Outputs
8-input sensor board                             4            24          25
8-output solenoid driver board (1 external       1            8           8
trigger input)
DC motor control board                           1            1           1
LED Driver board (2 external trigger inputs)     1            2           2

At least one spare of each board will also be required.

A block diagram of the Instrument Interface Electronics Chassis for both the
Transmissive and DBSS designs is shown in Figure 10.4. The DBSS design requires
one additional 8-sensor input board.

10.2.4 Servo Motor Drive Chassis
The Servo Motor Drive Chassis contains the amplifiers for each of the servo motors
used in the spectrograph for camera and grating rotation, etc. These amplifiers may be
purchased commercially, or may be designed and manufactured in-house if no
commercial alternative can be identified. The amplifiers will be mounted in a 19- inch
chassis. The format of the amplifiers depends on the source. If they are designed in-
house, they will be a single height Eurocard format plugged in to a common
backplane. Power for the amplifiers will come from an external power supply.

The amplifiers are likely to be a Pulse Width Modulated type design. Ideally, they
will accept a 10V analog input command signal from the associated channel in the
motion controller. Each amplifier will have an enable input (from the motion
controller channel), a fault output (to the motion controller channel), and limit switch
inputs to indicate mechanism end of travel.

The Servo Motor Drive chassis will contain machine interface boards for the encoder
and limit switch inputs to simplify cabling. These are likely to be 2 or 4-channel
boards, which also plug in to the common backplane.



                                          115                                 Electronics
                                                                      AAOmega CoD


The Servo Motor Drive chassis will contain motion controller interface boards, which
interface the motion controller to the amplifiers.

The number of amplifiers depends on the Spectrograph design:

Spectrograph Design                         Number of Servo Amplifiers
Transmissive                                6
DBSS                                        11

A block diagram of the Servo Motor Drive Chassis showing 4 servo amplifiers and
their interface boards is shown in Figure 10.5.

10.2.5 Power Supply Chassis
It is envisaged that a 3U x 19- inch chassis containing power supplies for the
Instrument Interface Electronics and the Servo Motor Drive chassis will be required.
Most likely, the power supplies themselves will be switch- mode type supplies
purchased off-the-shelf.

10.2.6 Spectrograph Electro-Mechanical Connections
A method of interconnecting the cables from the Spectrograph Control Electronics
and the individual sensors, motors, switches and encoders will be employed on
AAOmega. Most likely, this will be done using small printed circuit boards located at
various places on the Spectrograph assembly. It may also be feasible to use standard
DIN rail terminal blocks mounted in commercially sourced enclosures. The types of
cable and connectors to be used will be made during the preliminary design.

10.2.7 Electronics Mounting
There are likely to be several chassis units which need to be accommodated for the
Spectrograph Electronics Control. These include:
 Control Computer System
 Instrument Interface Electronics
 Servo Motor Amplifiers
 Power Supplies

It is likely that these chassis‘ would be 19- inch rack mountable units which would be
mounted in a suitably configured rack near the Spectrograph.

10.2.8 Telescope Electrical and Data Connections
10.2.8.1     Power Connections
A single phase power connection to the telescope Un- interruptible (Nobreak) Power
Supply will be required. If this is not easy to achieve, a local Un-interruptible Power
Supply unit for the Spectrograph electronics can be used.

10.2.8.2    Network Connections
A compatible Local Area Network connection (probably using fibre optic cable) will
be made between the AAOmega Spectrograph Control Computer System and the


                                         116                               Electronics
                                                                     AAOmega CoD


AAT Local Area Network, accessible either in the Cassegrain cage or via a new fibre
optic termination box mounting near the South Catwalk.

10.2.8.3    Cable Routing
Provision will need to be made for installation of the cable loom between the
Spectrograph electronics rack and the Spectrograph. Ideally, this would be
implemented using a covered cable duct.

10.2.9 Spares
Ideally, there should be a spare of each of those components which cannot be repaired
quickly, to keep the Mean Time To Repair as short as possible, and/or a spare of any
component which could have a low Mean Time Between Failure due to the nature of
its use or operation.

It is recommended that the following spare parts inventory be maintained:

Component                                        Number of Spares
Single Board Computer                            1
Control Computer Power Supply                    1
Motion Controller                                1
Digital I/O Board                                1
Analog Input Board                               1
Sensor Input Board                               2
Solenoid Driver Board                            2
DC Motor Control Board                           1
Digital I/O Interface Board                      1
Analog I/O Interface Board                       1
Instrument Interface Electronics Backplane       1
Servo Amplifier Board                            2
Servo Interface Board                            2
Servo Amplifier Chassis Backplane                1
Instrument Interface Electronics Power Supply    1
Servo Amplifier Power Supplies                   1




                                        117                                 Electronics
                                                                                           AAOmega CoD



Figure 10.1


                                                                                   Mechanical End of Travel Limit Switch

                                                   Mechanical Linkages             Electrical End of Travel Limit Switch
             Enable

             Command
                       Amplifier     Motor               Load            Encoder   Home Switch
 Motion
             Fault
Controller
Channel                                                                            Electrical End of Travel Limit Switch

                                                                                   Mechanical End of Travel Limit Switch

                                   Encoder (A,B,Z)
                                   Mechanical Limits
                                   Electrical Limits
                                   Home Position




                                                        118                                            Electronics
                                                                                    AAOmega CoD


Figure 10.2




                                       Digital Input/Output (Solenoids, Sensors)
                Instrument Control     Analog Input (Temperature, Voltage)
     Ethernet   Computer System
                                       Servo Command, Encoder, and Limit Switches




                                       Pneumatic Solenoids
                                       Position Switch Sensors
                Instrument Interface   Temperature Sensors
                Electronics Chassis
                                       DC Motors and Switches




                                                                                       AAOmega
                                                                                      Spectrograph




                                       Motor Drive Outputs
                  Servo Amplifier      Encoder Inputs
                     Chassis
                                       Limit Switch Inputs




                   Power Supply
                     Chassis




                                            119                                        Electronics
                                                                                                           AAOmega CoD




Figure 10.3




                                                                                     Ethernet Interface
                              Single Board Computer


    Console Terminal




                                                                                    Instrument Interface
                                                                Digital Input/        Electronics Rack
                                                                   Output
                                                                 Interfaces
                              Computer Input/Output Bus




                                                                   Multi-
                                                                  Channel           Temperature Sensors
                                                                Analog Input
                                                                 Interface




                                                                                  Spectrograph Mechanism
                                                                                 Servo Amplifiers, Encoders,
                                                                 4-Channel        Home and Limit Switches
                                                                Servo Motion
                                                                 Controller




                                                                                  Spectrograph Mechanism
                                                                                 Servo Amplifiers, Encoders,
                                                                4-Channel         Home and Limit Switches
                                                                  Motion
                                                                Controller1,2




                                                                                  Spectrograph Mechanism
                                                                                 Servo Amplifiers, Encoders,
                                                                 4-Channel        Home and Limit Switches
                                                                   Motion
                                                                 Controller3




                       Notes:
                       1 - In the Transmissive design, 2 channels of the second 4-channel
                       motion controller are used (and 2 are spare).
                       2 - In the DBSS design, all 4 channels of both the first and second 4-
                       channel motion controllers are used.
                       3 - In the DBSS design, a third 4-channel motion controller is
                       required with 3 channels used (and 1 spare).




                                                          120                                                  Electronics
                                                                                                  AAOmega CoD


Figure 10.4
                                Backplane                                           Control Computer Analog Input
                                                     8-Channel Analog Input                   Interface
                                                            Board


      Power Supply Inputs                                                                Temperature Sensors
                                                    8-Channel Temperature
                                                   Sensor Conditioning Board


     LED Trigger (2dF Robot)                                                        Back Illumination LED Drive A, B
                                                        LED Driver Board


                                                                                         Slit Interchange Motor
                                                                                               and Sensors
                                                     DC Motor Control Board



                                                                                    Filter Wheel Motor and Sensors
                                                    DC Motor Control Board2



                                                    8-channel Solenoid Driver         Pneumatic Valve Solenoids
                                                             Board

         Shutter Trigger
       (Detector Controller)                                                            Position/Status Sensors
                                                      8-input Sensor Board1



                                                                                        Position/Status Sensors
                                                      8-input Sensor Board



                                                                                        Position/Status Sensors
                                                      8-input Sensor Board



                                                                                        Position/Status Sensors
                                                      8-input Sensor Board


                                                                                      Control Computer Digital I/O
                                                                                               Interface
                                                    Digital I/O Interface Board



                               Note 1 - The fourth 8-input sensor board is required for the
                               DBSS design, but not the Transmissive design.
                               Note 2 - The DC motor control board for the filter wheel is
                               required for the Transmissive design, but not the DBSS design.




                                                        121                                              Electronics
                                                                                               AAOmega CoD




Figure 10.5

Backplane
              Cmd 1                                Motor 1




                            Amp 1
              ML 1



            Enc 1                                   Encoder 1

            ELH 1                           Electrical Limits and Home 1

            ML 1                                Mechanical Limits 1

            Enc 2     Amplifier Interface           Encoder 2              Mechanisms 1 and 2

            ELH 2                           Electrical Limits and Home 2

            ML 2                                Mechanical Limits 2




              Cmd 2                                Motor 2
                            Amp 2




              ML 2




              Cmd 3                                Motor 3
                            Amp 3




              ML 3



            Enc 3                                   Encoder 3

            ELH 3                           Electrical Limits and Home 3

            ML 3                                Mechanical Limits 3

            Enc 4     Amplifier Interface           Encoder 4              Mechanisms 3 and 4

            ELH 4                           Electrical Limits and Home 4

            ML 4                                Mechanical Limits 4




              Cmd 3                                Motor 3
                            Amp 3




              ML 3




            Enc 1                                   Encoder 1

            ELH 1                           Electrical Limits and Home 1

            Cmd 1                                   Command 1

            Enc 2                                   Encoder 2

            ELH 2                           Electrical Limits and Home 2

            Cmd 2        Motion                     Command 2

            Enc 3
                        Controller                  Encoder 3              Motion Controller
                        Interface
            ELH 3         Board             Electrical Limits and Home 3

            Cmd 3                                   Command 3

            Enc 4                                   Encoder 4

            ELH 4                           Electrical Limits and Home 4

            Cmd 4                                   Command 4




                                                                   122                            Electronics
                                                                        AAOmega CoD




10.3 Detector Controller Electronics
It is intended to use the AAO2 CCD Controller to operate the detectors in the
AAOmega Spectrograph. The number of controllers and their configuration depends
on the Spectrograph design, the number of detectors used and the number of output
amplifiers used on each detector. If a detector has two output amplifiers, its readout
time can be halved by reading out each half of the device through the two output
amplifiers. Each output amplifier used on a detector requires a separate video board in
the controller. As previously noted, when separate video boards are used to read out
two halves of a single detector, the read noise and gain characteristics of each half
will be different.

In the Transmissive design, two detectors will be mounted in the same dewar. A
single AAO2 controller is needed for these detectors. The controller would be
configured with one or two video boards per detector (depending on the number of
output amplifiers used), and one clock board per detector. If multiple dewars (for
different detector options) are made available, there should be one controller per
dewar.

The DBSS design has two detectors, each mounted in separate cryostats (some
distance apart). Each detector will require its own AAO2 controller. Each controller
will require one or two video boards (depending on the number of output amplifiers
used) and one clock board.

Prototype development of the AAO2 controller (which is a rework of the IRIS2
version of the controller) is a separate project and is not covered here. However, there
will be detector development work required for AAOmega to design the internal
dewar detector electronics and the controller analog backplane and flex circuits
(which are detector specific), and to test and characterise the AAOmega detectors.

Design            Detectors    Outputs per Controller       Video           Clock
                               Detector    Units            Boards          Boards
Transmissive      21           1           1                2               2
Transmissive      21           2           1                4               2
DBSS              22           1           2                23              23
DBSS              22           2           2                43              23

Notes
1. Two detectors in the same cryostat.
2. Two detectors in separate cryostats.
3. This is the total number of video or clock boards for two controllers.

Detector Control and Data Acquisition can occur using a single Sun/VMEbus system,
in both the Transmissive and DBSS designs. When two CCD controllers are used, two
VMEbus fibre optic link interfaces are required.

In both designs, the cryostat containing the detector will need to be electrically
isolated from the main body of the Spectrograph assembly, in order to avoid the


                                          123                                Electronics
                                                                      AAOmega CoD


possibility of earth loops generating noise which could affect the detector during
readout. In order to minimise the possibility of noise generated by the Spectrograph
Instrument electronics affecting the CCD Controller, operation of the Spectrograph
mechanisms during detector readout should be avoided.




                                         124                               Electronics
                                                                      AAOmega CoD




11 Software

The software for AAOmega consists of the following components:

AAO Configure           The software which allocates 2dF fibres to objects.
Instrument Control      The low- level software which runs the instrument hardware.
Observation Control     The high level software which combines interaction with the
                        instrument and control of the 2dF positioner and Detector
                        System.
Data Reduction          Basic and generalised data reduction software.


11.1 AAO Configure
Previously known as 2dF Configure, this program has been under development for
some time in the context of the 6dF and OzPoz projects. Whilst the current list of
bugs and required minor enhancements is quite large, it is expected that most will
have been attended to by the end of the OzPoz project. Limited changes will be
required to adapt AAO Configure to AAOmega.

The scientific requirements are outlined in Section 3. For AAOmega, Configure will
adapt the 2dF instrument definition to specify only one spectrograph rather then two.
Support will be added for obtaining the same configuration on both plates. Support
for beam switching and Nod & Shuffle are required.

It is recommended that the project scientist and interested parities gives further
thought to the desirability of adapting Configure (or providing additionally tools) to
ease the generation of target fields for large survey work. The current configure
approach is just to allocate targets from a given list with no thought give n to sky
coverage issues. Such effort has not been fully considered in this project and requires
proper analysed in view of scientific requirements to determine time scales and
costing.

11.2 Instrument Control
This software is responsible for moving the mechanisms, sensing the status etc. of the
AAOmega Spectrograph. It will run on a rack mounted Linux or VxWorks System.

It is presumed that for all hardware to be controlled appropriate drivers exist or
sufficient hardware information is available to write drivers.

The Instrument software will be written using the AAO DRAMA Software
Environment in C or C++. It will provide a DRAMA task command and parameter
interface that closely resembles the existing 2dF Spectrograph task (for various
reasons the existing 2dF Spectrograph task is not adaptable to AAOmega.).
Additionally, it will provide a FITS header interface appropriate for the new AAO -II
CCD system software.


                                         125                                 Software
                                                                        AAOmega CoD



There are no particular software engineering concerns in current mechanical and
electronics concept designs for AAOmega. The Transmissive and DBSS designs are
very similar from a software viewpoint – with duplication of mechanisms in the
DBSS design adding little effort over single equivalent mechanisms in the
Transmissive design.

An Engineering level GUI will be provided, written in JAVA or Tcl/Tk. This
interface will provide easy access to the instrument without the need to start the whole
2dF software system. Additionally, it will provide testing and debugging facilities
appropriate for both hardware and software maintenance.


11.3 Observation Control
This refers to the integration of the instrument control with the detector and 2dF robot
systems. It is assumed that the AAO-II Controller system will be available for
AAOmega. It is likely that some of the development will have taken place on 2dF
system using AAO-II CCD Controllers by the time it is required for the AAOmega
project.

The base effort required is to adapt the 2dF control software (tdfct) for the new one
spectrograph AAOmega system. This will consist of removing reference to the second
spectrograph and adapting the tdfct user interface to the AAOmega spectrograph
hardware. If the instrument software presents a similar interface to the existing 2dF
spectrograph software, there should be few changes needed to the underlying code.

If 2dF had not already been adapted to AAO-II CCD Controllers, then this change is
also required. Again, it should be possible to largely reflect the existing interfaces in
the new system, minimising the effort required.

11.4 Data Reduction
2dfdr is the 2dF data reduction software package. It has recently been adapted for the
6dF and SPIRAL instruments and has therefore proven quite adaptable to new
instruments. The scientific requirements for an upgrade 2dfdr are given in Section 3
and considerable effort is required.

There is probably more risk in 2dfdr then any other part of the software project.
Considerable investigation of options and techniques may be required. The
appropriate expertise is unlikely to be readily available, probably requiring a new
person to become familiar with the software, significantly increasing the effort
required.




                                          126                                  Software
                                                                         AAOmega CoD




12 2dF Upgrade

12.1 Requirements of 2dF upgrades program
This section details the requirements for the upgrading of the 2dF infrastructure. This
includes upgrades to enhance reliability and functionality of the 2dF top end systems
(corrector, positioner, tumbler, focal plane imager, fibre retractor units). Not included
are the spectrographs (to be replaced as part of the AAOmega project) and the fibre
probes themselves, which are to be replaced with new longer fibres of a different type,
although the probe design itself (button, prism, magnet) will remain unchanged.

12.2 Reliability

12.2.1 Gripper unit
The main sources of the gripper unit reliability problems are the jaw and Z-axis drive
systems. The jaw is susceptible to occasional glitches and loss of encoder counts. This
problem is unpredictable and completely random. Replacement of the theta axis slip
rings is a possible route to improve matters.

The Z-axis suffers from a mechanical instability in the servo system when the
telescope is at large zenith distances. This is caused by backlash between the motor
and the driven mass (via a gearbox and rack and pinion). While upgraded linear
bearings for the Z-axis reduced the problem, a pneumatic pre- load was unable to
eliminate the problem. This problem exhibits itself with reduced positioner
performance (increased configuration times) and prisms being subjected to increased
mechanical shocks due to some loss of Z-axis position.

The Z-axis encoder is also extremely vulnerable to moisture and dirt with a potential
for immediately fatal loss of counts on the Z-axis.

12.2.2 Gripper Gantry

The gripper gantry has been largely problem free since the replacement of the encoder
optical fibres and the regular replacement of the counterweight cords. It is unlikely
that better solutions to these items can be identified in a more cost-effective manner
than regular maintenance.

One potential problem identified during testing of the gripper unit is the Y-axis linear
bearing and the beam structure it is mounted on. It is known that the Y-axis beam sags
towards the field plate by up to 150m as the griper is driven along the Y-axis. Since
the tests were done on off the telescope it is not known if this sag is a permanent set
or a gravitationally varying effect. The end result of this sag is that since the Z-axis
position is defined at the centre of the field plate the effective plate height is changing
across the field plate. Therefore, if a fibre button is placed correctly at the centre of
the field plate it will be dropped from a 150m height at the edge of the field plate.


                                           127                               2dF Upgrade
                                                                         AAOmega CoD


A second major problem associated with the Y-axis is that the gripper unit rotates
about the Y-axis bearing with the magnitude of the movement dependent on the
attitude of the telescope. This causes problems of the same type as the sag of the Y-
axis, namely uncertainty in the Z-axis position and reduced positioner performance
and reliability.

12.2.3 Retractor units

The retractor unit refurbishment program has made huge improvements to the
reliability of the retractor units. Most of the problems identified have been due to poor
quality control in the manufacture of individual components (eg over/under size
pulleys). At this stage a complete redesign is not recommended, instead selected re-
manufacture of critical components with stricter quality control is preferred.

One remaining reliability issue associated with the retractor units is the heights of the
individual park plates. These park plates are glued to the retractor units and they are
neither all at the same height or even glued on parallel to the field plate. This problem
results in reduced positioner performance while parking fibres at the edge of the field
plate and increased risk of prisms falling off during the parking procedure.

12.2.4 Field plates

The current field plates consist of a sandwich of aluminium base and magnetic surface
glued together with epoxy. An array of 21 optical fibres is embedded into the field
plate to provide the two gantries with reference marks.

Two main problems exist with the field plate units. Firstly the field plate surface is not
flat with a peak to valley variation of up to 100m. Secondly the mounting flange of
the retractor units around the edge of the field plates is not of constant thickness (it
was manufactured out of specification). This second problem results in a variable
height of the retractor units (and their park plates).

No major design flaws exist in the field plates, the problems are simply manufacturing
problems and a re- manufacture of the field plates is recommended.

12.3 Functional upgrades

12.3.1 Plate rotation

The hardware for plate rotation is in place and tested (pre-1998). Substantial effort is
required to implement and test the plate rotation software and then commission on the
sky. The effort in implementing plate rotation should be played off against the gains
that would be achieved by correcting the field rotation component of atmospheric
refraction.




                                          128                               2dF Upgrade
                                                                    AAOmega CoD



12.3.2 Autoguider

The autoguider software has already been tested and found to be limited by the
characteristics of the Quantex TV system. Once the TV system is replaced, there will
be scope for re- implementing the autoguider system for 2dF. This is likely to be
linked to the Cassegrain focus autoguider infrastructure program and is an integral
component of the AAOmega 2dF upgrade.




                                        129                            2dF Upgrade
                                                                         AAOmega CoD




13 Risk Management

13.1 Purpose
The purpose of the risk management process is to minimise the risks of not achieving
the objectives of the project. Risk analysis is an important factor in selecting the
spectrograph concept to pursue. In general, risk management techniques are used to
assist the project management in setting priorities, allocating resources, and
implementing actions and processes that reduce risk.

The AAOmega team members have worked together in identifying and assessing the
risks facing the two spectrograph concepts and the other work packages that form the
AAOmega project.

The aim of the risk assessment was to:
 Identify and understand the risks to which the project is likely to be exposed;
 Assess the priority of these risks;
 Develop an initial risk register;
 Assign responsibility of AAOmega team members to analyse the risks identified;
 Consider the risks identified when making a decision on the spectrograph concept
   to be pursued.

13.2 Approach
The process followed the steps in the Australian/New Zealand Standard on Risk
Management AS/NZS 4360:1999. An overview of the steps is indicated in Figure
13.1.
Figure 13.1: The risk manage ment approach, AS/NZS 4360
                                    Consult & communicate




 Establish          Identify            Analyse             Evaluate              Treat
 the context        the risks           the risks           the risks             the risks
 Objectives                             Review controls     Evaluate risks        Identify options
                    What can
 Stakeholders       happen?             Likelihoods                               Select the best
                                                            Rank risks
 Criteria                               Consequences                              Develop plans
                    How could it                            Screen minor
 Key elements       happen?             Level of risk       risks                 Implement




                                      Monitor and review




                                        131                             Risk Management
                                                                        AAOmega CoD


13.3 Establishing the context
Establishing the context involved a review session with the project team. At this
session the objectives of the project were revisited. Based on these objectives, the
criteria for success were developed.

The project was then divided up into suitable key elements for the risk assessment.
13.3.1 Project objectives
 Performance, reliability, availability, maintainability, safety and quality consistent
   with user requirements
 Technical excellence
 Financial viability
 Supply to agreed timeframe.
13.3.2 Key elements

13.3.2.1      Spectrograph - Transmissive concept
1      Camera and collimator optics
2      Dispersive elements (VPH gratings)
3      Detectors
4      Filters
5      Fibre slits
6      Hartman shutters
7      Spectrograph enclosure
8      Lens mounts
9      Grating mount
10     Detector dewar
11     Grating rotator
12     Camera articulation mechanism
13     Spectrograph base
14     Detector focus mechanisms
15     Collimator/slit focus mechanism
16     Slit interchange mechanism
17     Back illumination
18     Calibration
19     Shutter
20     Instrument control electronics
21     Control computer
22     Pre-run software
23     Instrument control software
24     Observation control software
25     Data reduction software




                                          132                         Risk Management
                                                                    AAOmega CoD



13.3.2.2      Spectrograph - DBSS concept
1      Camera and collimator optics
2      Dispersive elements (VPH gratings)
3      Detectors
4      Dichroic
5      Fibre slits
6      Dichroic mount
7      Hartman shutters
8      Spectrograph enclosure
9      Lens mounts
10     Grating mount
11     Detector mount
12     Camera dewars
13     Grating rotator
14     Camera articulation mechanisms
15     Spectrograph base
16     Detector focus mechanisms
17     Collimator/slit focus mechanism
18     Slit interchange mechanism
19     Back illumination
20     Calibration
21     Shutter
22     Instrument control electronics
23     Control computer
24     Pre-run software
25     Instrument control software
26     Observation control software
27     Data reduction software
13.3.2.3     Key elements common to both concepts
28     Fibre bundles
29     SPIRAL upgrade
30     South catwalk upgrade
31     2dF refurbishment
32     Controller development
33     Infrastructure

13.4 Risk identification and risk analysis
A risk identification and assessment brainstorming workshop was conducted. For
each key element, the workshop participants:
       identified the main potential risks;
       assigned likelihood and consequences;
       established initial risk priorities.

The risk priority rating was established with consideration to the likelihood of the
problem to occur and the severity of the impact of the consequences using the system
outlined in Table 13.1.



                                        133                       Risk Management
                                                                     AAOmega CoD


                           Consequences (C)                          Likelihood (L)
Negligible   Minor       Moderate   Major       Catastrophic
    E           D            C        B               A
Medium       Major       Extreme   Extreme      Extreme         A Almost certain
Minor        Medium      Major     Extreme      Extreme         B Likely
Minor        Minor       Medium    Major        Extreme         C Moderate
Minor        Minor       Medium    Major        Extreme         D Unlikely
Minor        Minor       Minor     Medium       Major           E Rare

Table 13.1 Risk priority rating.

13.5 Risk assessment
The project team reviewed the identified risks and agreed the final priority ratings.
The responsibilities for addressing the risks were assigned. The output of the
workshop was a database in the form of a Risk Register.




                                        134                        Risk Management
                                                                                                                AAOmega CoD




13.5.1 Risk Register


13.5.1.1    Spectrograph - Transmissive concept

Item Ele ment          Risk             Consequences           Controls in        L   C   Risk priority   Responsibility
                                                               place
1     Camera and       Unavailability   Significant redesign   Check              C   A   Extreme         JGR
      collimator       of exotic        & delay;               availability now
      optics           glasses          Scrap transmissive
                                        design
2     Camera and       Final design     Change spec or         Design analysis/ A     B   Extreme         RH
      collimator       does not meet    start again            feedback
      optics           spec
3     Detectors        Delays in        Delays in finishing                       A   C   Extreme         GF
                       delivery time    the spectrograph
4     Data reduction   Transfer of      Delays &               Ensure             B   B   Extreme         TJF
      software         code support     performance            documentation
                                                               in place
5     Camera and       Exchange rate    Cost increases                            B   C   Major           GF
      collimator       and other
      optics           fluctuations
6     Camera and       Manufactured     Poor performance       Select             D   B   Major           JGR
      collimator       item does not                           appropriate
      optics           meet                                    manufacturers
                       specification
7     Detectors        Poor quality     Poor performance       Select reputable   C   B   Major           TJB
                                                               manufacturers
8     Fibre slits      Inadequate       Poor performance                          D   B   Major           GAS
                       positioning




                                        135                    Risk Management
                                                                                                               AAOmega CoD



Item Ele ment          Risk               Consequences         Controls in       L   C   Risk priority   Responsibility
                                                               place
9    Spectrograph      Short term         Poor performance                       D   B   Major           GAS
     enclosure         temperature        or drifting off
                       stability          central wavelength
                       inadequate
10   Lens mounts       Unable to meet     Poor performance     Tolerance         C   B   Major           GAS
                       lens positioning                        analysis on the
                       tolerances                              optical design
11   Coatings          Unable to get      Poor performance     Prototype plant   C   B   Major           RH
                       Solgel coatings                         at site
12   Detector dewar    Unable to meet     Poor performance     Design            C   B   Major           GAS
                       focusing spec
13   Camera            Movement           Poor performance                       D   B   Major           GAS
     articulation      misaligns
     mechanism         optical train
14   Spectrograph      Inadequate         Poor performance     Design            D   B   Major           GAS
     base              stability
15   Collimator/slit   Poor               Poor performance                       D   B   Major           GAS
     focus             repeatability
     mechanism
16   Back              Light leakage      Degraded data        2dF system        D   B   Major           GAS
     illumination      contaminates                            model
                       spectra
17   Instrument        Forced to          Delays & increased                     B   C   Major           LGW
     control           choose control     cost
     electronics       electronics we
                       haven‘t used
                       before




                                          136                  Risk Management
                                                                                                                AAOmega CoD



Item Ele ment           Risk              Consequences          Controls in       L   C   Risk priority   Responsibility
                                                                place
18   Camera and         Delay in          Delays in                               C   C   Medium          GF
     collimator         delivery of       completion of
     optics             optics            instrument
19   Fibre slits        Broken fibres     Decreased                               C   C   Medium          GAS
                                          observing
                                          efficiency
20   2dF                No expertise to   Delays                                  B   D   Medium          LGW
     refurbishment –    carry out
     plate rotation     required work

21   Pre-run            Can‘t be          Can‘t do Nod &        Project           D   C   Medium          TJF
     software           converted to      Shuffle               management;
                        Nod & Shuffle                           monitor
                        configuring
22   Instrument         Unavailable       Delay & increased     Select hardware   D   C   Medium          LGW
     control software   driver software   cost                  appropriately
23   Detector dewar     Unable to meet    Operational loading                     D   C   Medium          GAS
                        the spec in
                        terms of time
                        between
                        nitrogen fills
24   Calibration        Poor flat         Poor performance                        C   D   Minor           TJB
                        fielding




                                          137                   Risk Management
                                                                                                               AAOmega CoD




13.5.1.2   Spectrograph - DBSS concept


Item Ele ment          Risk               Consequences          Controls in      L   C   Risk priority   Responsibility
                                                                place
1     Camera and       Central            Reduced               Design analysis/ A   B   Extreme         PG
      collimator       obstruction        throughput            feedback
      optics           losses
2     Detectors        Delays in          Delays in                             A    C   Extreme         GF
                       delivery time      completing the
                                          spectrograph
3     Camera dewar     Large              Poor efficiency                       B    B   Extreme         GAS
                       obstruction due
                       to detector
                       package
4     Shutter          Unable to          Light leakage; data                   C    A   Extreme         GAS
                       position shutter   compromised
                       in appropriate
                       place
5     Data reduction   Transfer of        Delays &              Ensure          B    B   Extreme         TJF
      software         code support       substandard           documentation
                                          performance           in place
6     Camera dewar     Vacuum             Destruction of        Mechanical      E    A   Major           GAS
                       window             detector and          design
                       implodes           Schmidt plate
7     Camera and       Manufactured       Poor performance      Select          D    B   Major           PG
      collimator       item does not                            appropriate
      optics           meet spec                                manufacturers




                                          138                   Risk Management
                                                                                                               AAOmega CoD



Item Ele ment          Risk              Consequences         Controls in        L   C   Risk priority   Responsibility
                                                              place
8    Detectors         Poor quality      Poor performance     Select reputable   C   B   Major           TJB
                                                              manufacturers
9    Fibre slits       Inadequate        Poor performance                        D   B   Major           GAS
                       positioning
10   Dichroic          Poor optical      Poor performance     Specification      C   B   Major           PG
                       quality
11   Spectrograph      Short term        Poor performance                        D   B   Major           GAS
     enclosure         temperature       or drifting off
                       stability         central wavelength
                       inadequate
12   Lens & mirror     Unable to meet    Poor performance     Tolerance          D   B   Major           GAS
     mounts            positioning                            analysis on the
                       tolerances                             optical design
13   Camera dewar      Unable to meet    Poor performance     Design             C   B   Major           GAS
                       focusing spec
14   Camera dewar      Optical           Delays in            Prior cryogenic    C   B   Major           GAS
                       misalignment      commissioning and    experience and
                       associated with   service; poor        design
                       cryogenic         performance;
                       temperatures      breakage
15   Camera            Movement          Poor performance                        D   B   Major           GAS
     articulation      misaligns
     mechanism         optical train
16   Spectrograph      Inadequate        Poor performance     Design             D   B   Major           GAS
     base              stability
17   Collimator/slit   Poor              Poor performance                        D   B   Major           GAS
     focus             repeatability
     mechanism




                                         139                  Risk Management
                                                                                                                AAOmega CoD



Item Ele ment          Risk              Consequences           Controls in       L   C   Risk priority   Responsibility
                                                                place
18   Back              Light leakage     Degraded data          2dF system        D   B   Major           GAS
     illumination      contaminates                             model
                       spectra
19   Instrument        Forced to         Delays & increased     prototype         B   C   Major           LGW
     control           choose control    cost
     electronics       electronics we
                       haven‘t used
                       before
20   Camera and        Delay in          Delay in delivery of                     C   C   Medium          GF
     collimator        delivery of       spectrograph
     optics            optics
21   Camera and        Exchange rate     Cost increases                           B   D   Medium          GF
     collimator        and other
     optics            fluctuations
22   Fibre slits       Broken fibres     Decreased                                C   C   Medium          GAS
                                         observing
                                         efficiency
23   Camera dewar      Unable to meet    Operational loading                      D   C   Medium          GAS
                       the spec in
                       terms of time
                       between
                       nitrogen fills
24   2dF               No expertise to   Delays                                   B   D   Medium          LGW
     refurbishment –   carry out
     plate rotation    required work

25   Coatings          Unable to get     Poor performance       Prototype plant   C   C   Medium          RH
                       Solgel coatings                          at site




                                         140                    Risk Management
                                                                                                              AAOmega CoD



Item Ele ment           Risk              Consequences        Controls in       L   C   Risk priority   Responsibility
                                                              place
26   Pre-run            Can‘t be          Can‘t do Nod &      Project           D   C   Medium          TJF
     software           converted to      Shuffle             management;
                        Nod & Shuffle                         monitor
                        configuring
27   Instrument         Unavailable       Delay & increased   Select hardware   D   C   Medium          LGW
     control software   driver software   cost                appropriately
28   Calibration        Poor flat         Poor performance                      C   D   Minor           TJB
                        fielding




                                          141                 Risk Management
                                                                                                          AAOmega CoD




13.5.1.3   Common elements risks

Item Ele ment         Risk               Consequences          Controls in   L   C   Risk priority   Responsibility
                                                               place
1     Fibre bundles   High FRD           Poor efficiency –                   C   A   Extreme         RH
                                         slightly worse for
                                         DBSS
2     Fibre bundles   Fibre routing      Instrument out of                   C   A   Extreme         RH
                      causes breakage    action
3     Fibre bundles   Poor choice of     Poor performance-                   D   B   Major           RH
                      fibre core         compromised
                      diameter           spectrograph design
4     SPIRAL          Poor slit design   High FRD/poor                       C   B   Major           RH
      upgrade                            performance
5     2dF             Upgrade not        Delay                               B   C   Major           CMC
      refurbishment   well defined
6     Controller      Not finished on    Delays & cost                       D   B   Major           LGW
      development     time
7     South catwalk   Interference       Delay                               C   C   Medium          GF
      upgrade         with telescope
                      ops




                                         142                   Risk Management
                                                                      AAOmega CoD



13.6 Risk analysis
The assessment identified 24 risks for the transmissive concept, 28 risks for the DBSS
concept, and 7 risks that apply to either concept. The breakdown of the risks by
priority rating is given in Tables 13.2, 13.3 and 13.4.

Table 13.2: Summary risk profile for the spectrograph - trans missive concept
       Extre me     Major        Medium        Minor        Total
       Risks        Risks        Risks         Risks
       4            13           6             1            24


Table 13.3: Summary risk profile for the spectrograph - DBSS concept
       Extre me     Major        Medium        Minor        Total
       Risks        Risks        Risks         Risks
       5            14           8             1            28


Table 13.4: Summary risk profile for the common elements
       Extre me     Major        Medium        Minor        Total
       Risks        Risks        Risks         Risks
       2            4            1             0            7


13.6.1 Risk Action Plan
This plan looks into identifying the options for eliminating, reducing, controlling or
otherwise managing the high-priority (extreme and major) risks.
13.6.1.1     Spectrograph – Transmissive concept
1) The transmissive collimator and camera use a variety of glass types, in order to
   obtain a spread of both refractive indices and dispersions. Of necessity, some
   types are expensive and more difficult to obtain than others. However, similar
   refractive instruments have been successfully built (e.g. GMOS). The risk can be
   managed by ordering glass blanks early, and by checking availability and lead
   times for glass types as the design is finalised. If necessary, substitutions can be
   made at this stage.
2) The transmissive camera and collimator design does not curre ntly meet the
   technical specification. Unless significant refinements are made, or a decision is
   made to change the specification, the current design would be ruled out as not
   viable and the DBSS system chosen in preference. Refinement of the system
   could take considerable time, significantly delaying the project and eventually
   may not produce a design that performs to specification. A change in the
   specification would compromise AAOmega's scientific capability and reduce its
   value to the Astronomy user community. A decision needs to be made before
   September 2001 on this as the optical design is pivotal to the whole spectrograph


                                         143                        Risk Management
                                                                          AAOmega CoD


     design and delays in the decision would adversely impact all other aspects of the
     spectrograph design.
3)   The detectors are items that require long lead-time for delivery. From previous
     experience, this may be six months or longer. To ensure that we receive the
     detectors on time, we will have to finalise the optical design of the spectrograph
     during the preliminary design stage, which would enable us to specify the detector
     requirements. Placing the order early minimises the risk of delaying the
     completion of the project. However, this needs to be balanced against deferring
     the order to get the maximum possible benefit from any improvements in detector
     performance.
4)   Quality scientific output from AAOmega is reliant on good data reduction
     facilities. 2dfdr is a good foundation for this. Unfortunately, the author of 2dfdr
     has flagged the possibility he may leave the AAO to take up a scientific position
     at the end of the year. This means the software must be transferred to another
     person and software engineers who understand of the relevant issues are hard to
     come by. This creates some significant risk to the project. In order to minimise
     this risk, we must ensure that 2dfdr is well documented from a maintenance
     perspective before the author departs AAO.
5)   The currency exchange rate fluctuations are difficult to predict. These have a
     direct impact on the cost of the items purchased overseas, which are a fair
     proportion of the contract cost. The most expensive items are probably the optics,
     including lenses, mirrors, dichroic, detectors. Placing the orders for these
     components as early as possible would significantly reduce the risk of cost
     overruns due to currency exchange fluctuations. We also need to budget for
     contingencies. Alternatively, we could look into buying forward cover for
     exchange rate.
6)   If optical components do not meet the specification, imaging quality will suffer.
     Contracts for manufacture will need to include agreed testing procedures and
     pass/fail criteria. In the event of a component not meeting the specification, the
     manufacturer will replace it.
7)   We will minimise the risk of obtaining poor quality CCDs, by: selecting from
     reputable manufacturers with a proven track record; talking to other customers of
     these manufacturers who have experience with these devices; giving a detailed
     and tight specification (here the experience of people such as John Barton and
     Chris Tinney will be invaluable); maintaining good communication with the
     manufacturer and any subcontractors during all stages of the fabrication process;
     ensuring that any devices are adequately tested by the manufacturer prior to
     delivery; and thoroughly testing all delivered devices in the laboratory prior to
     installation in the instrument.
8)   Each slit will be finally located, in the observing position, on kinematic mounts.
     A pneumatically operated clamping device will be provided at each of the three
     interface points. The reference points against which the slits are clamped will be
     rigidly attached to the structure of the spectrograph. Note that the slits will have to
     be set-up on a reference jig at manufacture to ensure their conformity to the
     nominal slit geometry specification.
9)   The enclosure, which will be based on general cool-room principles of
     construction, will ensure a large thermal resistance between dome ambience and
     its interior. This, combined with the thermal inertia of the spectrograph will lead
     to a long thermal time constant. However, to maintain this advantage, entry to the
     enclosure, to change gratings, for example, will have to be minimised.


                                            144                         Risk Management
                                                                         AAOmega CoD


10) The lens mounts will be designed to facilitate adjustment at alignment and to
    ensure stability in operation. The lens alignment tolerances are compatible with
    common, achievable, instrument dimensions.
11) The use of Solgel anti-reflection coating over a single layer magnesium fluoride
    significantly improves the system throughput and reduces stray reflections and
    ghosting. The AAO has set up a Solgel capability at the AAT, however the
    AAOmega optics are relatively large and we have little experience with coatings.
    A prototyping and development program needs to be implemented as soon as
    possible to ensure that we can reliably So lgel coat large optics with the required
    central wavelength. A backup to this is commercial Solgel coating, but it is rather
    specialised and not likely to be available in Australia.
12) This is a challenging design problem, which will be solved by attention to detail.
    Some early testing of the concept should be carried out to ensure that any
    problems be resolved before they impact on important milestones.
13) Camera articulation should be resolved by attention to detail in design and careful
    alignment at instrument assembly. The rotary tables proposed for driving this
    motion are adequate for the job. The outboard bearing must be aligned, at
    assembly, with the rotation axis of the table.
14) An optical table of deep sandwich construction is specified for the main base of
    the instrument. Furthermore, it will be mounted on a set of pneumatic vibration
    isolators. The combination should provide excellent vibration isolation and high
    inherent stiffness within the table itself.
15) As the collimator is relatively fast, the focus mechanism must be capable of fine
    adjustment. However, as the slits can be repeatably aligned in the observing
    position (see 8, above) the need for a focus adjustment is only brought about by
    the apparent change in slit position caused by differing filter optical thicknesses.
    Translating the collimator lens assembly may provide this compensation. The
    range of adjustment required should be no more than 1mm. The collimator may
    thus be suspended on a flexure and positioned using a motorised micrometer.
    Resolution of ~1m can be obtained.
16) The 2dF spectrographs provide a prototype for design of the back-illumination
    system. This area of the 2dF spectrographs has been particularly satisfactory. This
    design will be copied as much as is practical.
17) Instrument control electronics, where possible, will be select to use electronics
    that is the same or similar to that with which we have had previous experience. If
    this is not possible, prototype systems involving new electronics during the
    Preliminary Design phase.
24) The main calibration risk is that we will not obtain satisfactory detector flat
    fielding, and the consequence is that the data quality will not be as high as it could
    be. We will minimise the risk associated with poor flat fielding by: specifying the
    necessary level of flat fielding; implementing a capability for detector flat fielding
    using a long-slit and suitable lamp; and testing detector flat fielding both in the
    laboratory prior to delivery and on the telescope to ensure satisfactory
    performance.




                                          145                         Risk Management
                                                                        AAOmega CoD



13.6.1.2        Spectrograph – DBSS concept
1) Central obstruction losses have been predicted with reasonable accuracy – the
    main source of uncertainty is the effect of fibre FRD.
2) As per item 3 for the transmissive concept.
3) As the obstruction problem is inherent to Schmidt camera designs, to large degree
    we must accept it. The only strategy that can be adopted here is to design with
    attention to detail to remove unnecessary obstruction. The significant gains may
    be made in the area taken by detector support hardware although choice of
    detectors can alleviate the problem. The EEV CCDs, for example, have low ratios
    of device area to sensitive area, due to rear facing connectors.
4) The appropriate location for the shutter is close to the slit. Attention to design
    detail is the best strategy we can adopt, but closure of both Hartmann shutters will
    provide an additional barrier to the passage of light from slit to detector.
5) Quality scientific output from AAOmega is reliant on good data reduction
    facilities. 2dfdr is a good foundation for this. Unfortunately, the author of 2dfdr
    has flagged the possibility he may leave the AAO to take up a scientific position
    at the end of the year. This means the software must be transferred to another
    person and software engineers who understand of the relevant issues are hard to
    come by. This creates some significant risk to the project. In order to minimise
    this risk, we must ensure that 2dfdr is well documented from a maintenance
    perspective before the author departs AAO.
6) The camera dewar windows will undergo a simple testing regime that will validate
    the design before delicate components are subject to possible damage. Initially an
    otherwise empty camera will be tested under vacuum with a dummy window, then
    the corrector, vacuum window will be fitted and tested. Only after this will the
    camera be fully assembled and tested. With fused silica as the corrector material,
    there should be little risk in designing to support atmospheric pressure with a high
    safety factor. The camera correctors, currently specified as 20 mm thick, could be
    made substantially thicker if mechanical analysis required this. Other than the cost
    of the material and weight of the cameras, there would be no negative impact.
7) The specification to the optical manufacturer should include a test arrangement
    that will provide null tests of the Schmidt correctors. Such tests should not be too
    difficult or expensive.
8) Minimising the risk of obtaining poor quality CCDs, will be achieved by:
    selecting from reputable manufacturers with a proven track record; talking to
    other customers of these manufacturers who have experience with these devices;
    giving a detailed and tight specification (here the experience of people such as
    John Barton and Chris Tinney will be invaluable); maintaining good
    communication with the manufacturer and any subcontractors during all stages of
    the fabrication process; ensuring that any devices are adequately tested by the
    manufacturer prior to delivery; and thoroughly testing all delivered devices in the
    laboratory prior to installation in the instrument.
9) The fibre slits will be treated in the same manner as in risk no. 8 for the
    transmissive design option.
10) As far as its effect on optical aberrations are concerned, the dichroic needs only to
    be coated on a plane parallel plate with modest flatness and parallelism tolerances.
11) The spectrograph enclosure will be treated in the same manner as in risk no. 9 for
    the transmissive design.



                                          146                         Risk Management
                                                                          AAOmega CoD


12) The lens and mirror mounts risks are the same as risk no. 9 for the transmissive
    design.
13) The 2dF spectrograph camera (a Schmidt design) serves as a model for the camera
    focus scheme in the DBSS. The system used on 2dF has proven satisfactory in
    service.
14) Fortunately, we have prior experience with evacuated, cryogenic Schmidt cameras
    in Fisch and the 2dF spectrograph. Furthermore, the number of optical
    components in the cryogenic environment is low and the assemblies relatively
    simple, mechanically. The mirror and corrector plates for DBSS are not at
    cryogenic temperature.
15) Camera articulation risks are similar to no.13 of the transmissive design option.
16) Spectrograph base risks are similar to no.14 of the transmissive design option.
17) As the collimator is relatively fast any focus mechanism must be capable of fine
    adjustment. However, as the slits can be repeatably aligned in the observing
    position (see no. 9, above, and comments on risk no. 8 for the transmissive design
    option) and there is no need for order sorting filters (the function is effectively
    carried out by the dichroic filter), so there should be no need for operational focus
    adjustment. However, should it be determined that the risk has not been fully
    mitigated, focus compensation may be provided by translating the collimator
    mirror assembly. The range of adjustment required should be no more than 1mm.
    The collimator mirror may thus be suspended on a flexure and positioned using a
    motorised micrometer. Resolution of ~1m can be obtained.
18) Back illumination risks are the same as no.16 of the transmissive design option.
19) Where possible, select and use electronics that is the same or similar to that with
    which we have had previous experience for the instrument control. If this is not
    possible, prototype systems involving new electronics during the Preliminary
    Design phase.
28) The main calibration risk is that we will not obtain satisfactory detector flat
    fielding, and the consequence is that the data quality will not be as high as it could
    be. We will minimise the risk associated with poor flat fielding by: specifying the
    necessary level of flat fielding; implementing a capability for detector flat fielding
    using a long-slit and suitable lamp; and testing detector flat fielding both in the lab
    prior to delivery and on the telescope to ensure satisfactory performance.
13.6.1.3      Common elements
1) The Focal Ration Degradation (FRD) of the optical fibres is a key performance
   driver for the AAOmega system. Both the overall FRD performance and fibre to
   fibre variations must be kept to a minimum. The 2df corrector output focal ratio is
   f/3.4, however FRD within a well designed system can degrade this so that
   typically less than 75% of the light would be collected within f/3.4. In order to
   improve the light coupling efficiency into the spectrograph AAOmega has been
   designed with a f/3.15 collimator. However, there is a significant risk that without
   careful design, prototyping and testing of the optical fibre, the fibre bundles, fibre
   slits and fibre routing, the amount of light within the f/3.15 collecting capacity of
   the collimator will be greatly reduced in the multi-object mode. Typically the
   most significant contribution to FRD is localised stressing of the fibre from
   pinching or similar effects. A length of 27m fibre with similar expected
   characteristics will be tested to ensure that the f/3.15 collimator design is suitable.




                                           147                         Risk Management
                                                                           AAOmega CoD


     Note: The IFU is fed nominally fed with a f/5.5 beam, which becomes f/3.9 across
     the microlens diagonals and has been characterised and is therefore a much lower
     risk.
2)   With the "bench mounted" spectrograph design, the MOS fibres from the
     telescope prime focus must be carefully routed from the telescope top end to the
     South Catwalk. The routing needs to take account of telescope movement and all
     possible telescope orientations without putting and significantly stress on the
     fibres. Also when the 2df top end is removed from the telescope the MOS fibres
     will need to be removed with the top end. Failure to have the fibre fully
     disengaged from the spectrograph, or fibre routing system could have catastrophic
     effects, damaging the spectrograph, 2df top end, the fibre bundle and present a
     significant risk to AAT staff engaged in the task. It is vital that sufficient handling
     procedures and interlocks are in place to prevent this from occurring.
     Note: The fibre routing for the Cassegrain IFU is lower risk, but still needs similar
     careful consideration and design.
3)   The fibre core size for the MOS mode of AAOmega is governed by a large
     number of parameters, including the object size (corrector performance, the
     seeing, galaxy/extended source or star/point source), sky brightness, button
     positing accuracy, integration time of the target field (the astrometric distortion
     changing with time), spectral resolution requirements, fibre sampling at the
     detector, etc. The choice of fibre core size is a complex trade off between, these
     often conflicting parameters, each of which capable of having a significant effect
     on the overall efficiency of the system. It is important that each of the appropriate
     parameters is well understood and how any trade off would compromise the
     performance before a suitable core size can be arrived at. The technical
     specification, 2df performance and other observing constraints limit most of these
     parameters. The fibre core size was carefully addressed during the 2df project,
     however it would be reviewing the choice in view of some differing requirement
     of the AAOmega system.
     Note: The IFU fibre core size is fixed at 85m, as the current SPIRAL B fibre
     bundle is to be used.
4)   The slit designs for both the MOS and IFU mode need to be carefully considered.
     There are a number of techniques for positioning fibres in a "long slit" to very
     high accuracy, however a number of these should be investigate and prototyped to
     ascertain the most appropriate technique for AAOmega.
     a) It needs to emulate the requirements of the spectrograph optical design, ie.
         approximate a curved surface, and have the fibres correctly pointing at the
         spectrograph pupil, or the system throughput will be compromised.
     b) Variations in the fibre spacing (most particularly the IFU where adjacent
         spectra are likely to overlap) and the position in the spectral direction should
         be keep to a minimum. Failure to achieve good alignment of the fibres in the
         slits could lead to significant complications in the data reduction process and
         loss in system sensitivity, due to contamination of spectra by features in
         adjacent fibres.
     c) The slit assembly needs to be designed to allow ease of maintenance/repair and
         introduce the minimal amount to localise stressing of the fibre in order to
         minimise the fibre FRD.
5)   The refurbishment of 2dF needs to be well defined and planned around the
     telescope scheduling in order for the work to be carried out by telescope staff,
     without disrupting their telescope support duties.


                                            148                         Risk Management
                                                                       AAOmega CoD


6) The Optical CCD Controller design is based on the IRIS2 IR array controller
   design. The analog design of the controller does need to be converted to run
   optical detectors - this is part of the Observatory Infrastructure Controller upgrade
   project. It is unlikely that the digital design and associated firmware and
   laboratory test infrastructure will need to change too much unless there is a
   problem with component availability. Availability of all components needs to be
   confirmed as soon as possible. Adequate resources need to be made available as
   soon as possible to start prototype design work on the analog boards. The
   AAOmega project plan must reflect realistic time scales for the development of
   the controller. This will be defined within the AAO Observatory Infrastructure
   project.




                                         149                         Risk Management
                                                                        AAOmega CoD




14 Costs

14.1 AAOmega System with Fully Transmissive Spectrograph

14.1.1 Total estimated costs

Spectrograph
Materials
1. Optics                                       $    1,005,000
2. Mechanical                                    $     137,565
3. Electronics                                   $     468,000
                           Total materials                             $    1,610,565
Labour
1. Optical design                               $       15,000
2. Mechanical                                   $      600,107
3. Electronics                                  $      630,042
4. Software                                     $      267,797
5. Management
Project manager                                 $      129,850
Project astronomer                              $       39,000
Instrument scientist                            $      122,500
                         Total labour                                  $    1,804,296
                 Total spectrograph                                    $    3,414,861
                   Catwalk upgrade                                     $       52,000
                  2dF refurbishment                                    $      200,000
                    SPIRAL upgrade                                     $       10,000
              TOTAL PROJECT COST                                       $    3,676,861
Table 14.1 Estimated costs for the AAOmega system with a fully transmissive spectrograph


14.1.2 Labour costs

It was estimated that the labour required for the transmissive AAOmega syst em
design, construction, assembly, integration, testing and commissioning would be
41,737 hours. This includes 50% contingency. In calculating the labour costs, the pay
rates were taken from the maximum step at the level required for the particular type of
work. No allowance was made for general pay increases.




                                          151                                      Costs
                                                                        AAOmega CoD




                                                  Work (hours)                   Cost

Optical design                                                           $    15,000
Mechanical                                                14,749         $ 600,107
Electronics                                               15,509         $ 630,042
Software                                                   5,329         $ 267,797
Project manager                                            2,650         $ 129,850
Project scientist                                          1,000         $    39,000
Instrument scientist                                       2,500         $ 122,500
                           TOTAL                          41,737         $ 1,804,296
Table 14.2 Labour estimates for the AAOmega system with a fully transmissive
spectrograph


14.1.3 Materials costs

The estimated costs for materials, equipment, and off-the-shelf components are shown
in more detail in Tables 14.3 and 14.4. There is a high uncertainty with regards to the
cost of glass blanks and manufacture of camera and collimator lenses, and the
contingency allowed for is probably insufficient. For the other items, the contingency
varies, from very little - when recent quotes are available, to 50% - when the
estimates are only based on previous experience.

The estimated materials costs for the mechanical section appear very small. This is
due to the fact that all manufacturing was included in labour costs. At this stage it is
unclear what percentage of the parts would be manufactured in-house.

Fully Transmissive Spectrograph
             Item            Qty           Unit cost       Total cost
OPTICS
Camera
CaF2 blanks                   3                            $ 100,000
Other glass blanks            6                            $ 100,000
Manufacturing                 9                            $ 130,000
Collimator
Blanks                                                     $ 165,000
Manufacturing                                              $ 125,000
Other
CCD                                                        $ 200,000
Filters                                                    $ 15,000
VPH gratings                  3            $ 10,000        $ 30,000
Replication transmissive      1            $ 20,000        $ 20,000
grating
Fibre                                                      $ 120,000
          Total optics costs
                                                                          $1,005,000

                                          152                                     Costs
                                                              AAOmega CoD


MECHANICAL
Slit interchange motor       1            $ 1,000     $    1,000
Collimator focus motor mike  1            $ 2,085     $    2,085
Filter motor drive           1             $   700    $      700
Rotator - VPH grating        1            $ 10,000    $   10,000
Rotator - camera             1            $ 12,525    $   12,525
Camera focus motor mikes     3            $ 2,085     $    6,255
Camera outboard bearing      1            $ 1,500     $    1,500
Optical table                1            $ 16,000    $   16,000
Vibration isolators          1            $ 8,000     $    8,000
Pneumatic valves             10            $   250    $    2,500
Cylinders                    10            $   300    $    3,000
Miscellaneous al, steel and                           $   30,000
hardware
Vacuum valves                1             $  500     $      500
Vacuum gauge set             1            $ 3,500     $    3,500
Spares                                                $   10,000
Prototyping                                           $   10,000
Thermal enclosure            1            $ 20,000    $   20,000
      Total mechanical costs                                       $ 137,565
ELECTRONICS
Instrument Controller
Computer Control System         1         $ 30,000    $   30,000
4 Channel DC Servo              2         $ 7,500     $   15,000
Controller
Digital I/O Board (32-bits)     2         $   5,000   $   10,000
Analog I/O Board (16            1         $   8,000   $    8,000
channels)
DC Motor Control Board          2         $   2,000   $   4,000
Solenoid Driver Board           1         $   2,000   $   2,000
Sensor Board                    3         $   2,000   $   6,000
LED Driver Board                1         $   2,000   $   2,000
Digital Interface Board         1         $   2,000   $   2,000
Analog Interface Board          1         $   2,000   $   2,000
Instrument Interface Chassis    1         $   3,500   $   3,500
Backplane
DC Servo Amplifier              6         $   2,000   $   12,000
Machine Interface Board         3         $   2,000   $    6,000
Servo Amplifier Backplane       1         $   3,500   $    3,500
Motion Controller Interface     2         $   3,500   $    7,000
Board
Power Supplies                  3         $   3,333   $   10,000
Racks and Enclosures                                  $   20,000
Termination Boards                                    $   30,000
Miscellaneous Components                              $   20,000
Cable and Connectors                                  $   40,000


                                    153                                 Costs
                                                                              AAOmega CoD


Prototyping                                                      $    50,000
Spares                                                           $    70,000
Detector Systems
Test Dewar and Electronics              1         $ 10,000       $    10,000
Dewar Detector Electronics              1         $ 10,000       $    10,000
Detector Controller                     1         $ 75,000       $    75,000
Enclosure
Environment Control                                              $    20,000
     Total electronics costs                                                    $ 468,000
                 Total costs
                                                                                $1,610,565
Table 14.3 Materials costs for the AAOmega fully transmissive spectrograph

Other material costs             Qty Unit cost               Cost
          Catwalk Upgrade                                                      $   52,000
2dF Refurbishment
Retractors                        800       $      25        $       20,000
Buttons                           800       $      30        $       24,000
Prisms                            800       $      30        $       24,000
Robot upgrades                                               $       40,000
 Total 2dF Refurbishment                                                       $ 108,000
Table 14.4 Materials costs for the catwalk upgrade and 2dF refurbishment




                                            154                                       Costs
                                                                     AAOmega CoD




14.2 AAOmega System with DBSS Spectrograph

14.2.1 Total estimated costs
Spectrograph
Materials
1. Optics                                     $       635,000
2. Mechanical                                 $       171,145
3. Electronics                                $       600,000
                      Total materials                                   $     1,406,145
Labour
1. Optical design                             $         5,000
2. Mechanical                                 $       647,122
3. Electronics                                $       730,517
4. Software                                   $       267,797
5. Management
Project manager                               $       129,850
Project astronomer                            $        39,000
Instrument scientist                          $       122,500
                        Total labour                                    $     1,941,785
                Total spectrograph                                      $     3,347,930
                  Catwalk upgrade                                       $        52,000
                 2dF refurbishment                                      $       200,000
                   SPIRAL upgrade                                       $        10,000
           TOTAL PROJECT COST                                           $     3,609,930
Table 14.5 Estimated costs for the AAOmega system with a DBSS spectrograph

14.2.2 Labour costs
It was estimated that the labour required for the DBSS AAOmega system design,
construction, assembly, integration, testing and commissioning would be 45,878
hours. This includes 50% contingency. In calculating the labour costs, the pay rates
were taken from the maximum step at the level required for the partic ular type of
work. No allowance was made for general pay increases.

                                                   Work (hours)                   Cost

Optical design                                                      $            5,000
Mechanical                                                 16,318   $          647,122
Electronics                                                18,081   $          730,517
Software                                                    5,329   $          267,797
Project manager                                             2,650   $          129,850
Project scientist                                           1,000   $           39,000
Instrument scientist                                        2,500   $          122,500
                          TOTAL                            45,878   $        1,941,786
Table 14.6 Labour estimates for the AAOmega system with a DBSS spectrograph



                                        155                                      Costs
                                                                          AAOmega CoD


14.2.3 Materials costs
The estimated costs for materials, equipment, and off-the-shelf components are shown
in more detail below. The contingency varies, from very little - when recent quotes
are available, to 50% - when the estimates are only based on previous experience.

The estimated materials costs for the mechanical section appear very small. This is
due to the fact that all manufacturing was included in labour costs. At this stage it is
unclear what percentage of the parts would be manufactured in-house.

Double Beam Schmidt Spectrograph
                 Item                  Qty         Unit cost Total cost
OPTICS
CCD                                     2          $100,000 $ 200,000
Dichroic                                1          $ 68,000 $ 68,000
Camera and collimator optics                                $ 147,000
VPH gratings                            6          $ 10,000 $ 60,000
Replication transmissive grating        2          $ 20,000 $ 40,000
Fibre                                                       $ 120,000
                    Total optics costs                                          $ 635,000
MECHANICAL
Slit interchange motor                  1          $    1,000 $       1,000
Collimator focus motor mike             1          $    2,085 $       2,085

Rotator - VPH grating                          2   $ 10,000 $        20,000
Rotator - camera                               2   $ 12,525 $        25,050

Camera focus motor mikes                       6   $    2,085 $      12,510
Camera outboard bearing                        2   $    1,500 $       3,000

Optical table                                  1   $ 16,000 $        16,000
Vibration isolators                            1   $ 8,000 $          8,000

Pneumatic valves                              10    $     250 $       2,500
Cylinders                                     10    $     300 $       3,000

Miscellaneous al, steel and hardware                            $    30,000
Vacuum valves                        2    $   500               $     1,000
Vacuum gauge set                     2 $ 3,500                  $     7,000
Spares                                                          $    10,000
Prototyping                                                     $    10,000
Thermal enclosure                    1 $ 20,000                 $    20,000
             Total mechanical costs                                             $ 171,145
ELECTRONICS
Instrument Controller
Computer Control System                1 $ 30,000               $    30,000
4 Channel DC Servo Controller          3 $ 7,500                $    22,500


                                           156                                       Costs
                                                                           AAOmega CoD


Digital I/O Board (32-bits)              2 $ 5,000 $ 10,000
Analog I/O Board (16 channels)           1 $ 8,000 $    8,000
DC Motor Control Board                   1 $ 2,000 $    2,000
Solenoid Driver Board                    1 $ 2,000 $    2,000
Sensor Board                             4 $ 2,000 $    8,000
LED Driver Board                         1 $ 2,000 $    2,000
Digital Interface Board                  1 $ 2,000 $    2,000
Analog Interface Board                   1 $ 2,000 $    2,000
Instrument Interface Chassis             1 $ 3,500 $    3,500
Backplane
DC Servo Amplifier                      11 $ 2,000 $ 22,000
Machine Interface Board                  6 $ 2,000 $ 12,000
Servo Amplifier Backplane                1 $ 3,500 $    3,500
Motion Controller Interface Board        3 $ 3,500 $ 10,500
Power Supplies                           3 $ 3,333 $ 10,000
Racks and Enclosures                                $ 20,000
Termination Boards                                  $ 35,000
Miscellaneous Components                            $ 25,000
Cable and Connectors                                $ 50,000
Prototyping                                         $ 50,000
Spares                                              $ 70,000
Detector Systems
Test Dewar and Electronics               1 $ 10,000 $ 10,000
Dewar Detector Electronics               2 $ 10,000 $ 20,000
Detector Controller                      2 $ 75,000 $ 150,000
Enclosure
Environment Control                                 $ 20,000
                Total electronics costs                        $ 600,000
                            Total costs                       $1,406,145
Table 14.7 Materials costs for the AAOmega DBSS spectrograph

Other material costs             Qty Unit cost              Cost
          Catwalk Upgrade                                                   $   52,000
2dF Refurbishment
Retractors                        800     $      25        $    20,000
Buttons                           800     $      30        $    24,000
Prisms                            800     $      30        $    24,000
Robot upgrades                                             $    40,000
 Total 2dF Refurbishment                                                    $ 108,000
Table 14.8 Materials costs for the catwalk upgrade and 2dF refurbishment




                                          157                                     Costs
                                                                       AAOmega CoD



14.3 Estimated project cost - by phase
The estimates shown in Table 14.9 were made before selecting the spectrograph
concept to be pursued. This was possible due to the fact that the cost estimates for the
two concepts came within very close range. Where the costs were considerably
different, an estimate close to average was used.


Preliminary design phase
Labour
Optical design                           $     10,000
Mechanical design                        $    135,000
Electronics design                       $    120,000
Software                                 $     54,000
Management                               $     70,000
                        Total labour     $    389,000
               Materials: prototyping    $     60,000
                                Total                                   $       449,000
Final design phase
Labour
Mechanical design                        $    170,000
Electronics design                       $    217,000
Software                                 $     52,000
Management                               $     70,000
                        Total labour     $    509,000
                    Materials: optics    $    800,000
                                Total                                   $    1,309,000
Manufacturing , integration, testing & commissioning
Labour
Mechanical                               $    320,000
Electronics                              $    320,000
Software                                 $    162,000
Management                               $    150,000
                        Total labour     $    952,000
Materials
Mechanical                               $    150,000
Electronics                              $    540,000
                      Total materials    $    690,000
                                Total                                   $    1,642,000
           Total spectrograph cost                                      $    3,400,000
                  Catwalk upgrade                                       $       52,000
                 2dF refurbishment                                      $      200,000
                   SPIRAL upgrade                                       $       10,000
           TOTAL PROJECT COST                                           $    3,662,000
Table 14.9 AAOmega system cost estimates - by phase




                                          158                                     Costs
                                                                     AAOmega CoD



15 Comparison and Conclusions
One of the goals of the concept design was to cost and compare the various optical
concepts and select a single optical design for further development during the next
phase of the project. The two main competing optical concepts are the fully
Transmissive Single Beam Spectrograph (TSBS) and the Dual Beam Schmidt
Spectrograph (DBSS). A list of what was felt to be the most important comparison
criteria primarily based on optical performance, risk, cost were addressed and
compared for each concept. A summary of the arguments for each criteria is presented
below. The comparison between the two systems was remarkably well balanced,
however it was felt that the very high risk associated with the optical components of
the TSBS significantly out weighted all the other pros and cons. It was on this basis
that the AAO has selected the DBSS as the preferred option.

15.1 Throughput in MOS mode
The throughput versus wavelength of the two systems are reasonably similar, but on
average the TSBS is higher particularly in beyond 550nm and the DBSS is better
below 500nm. The ratio of the DBSS/TSBS throughput with the likely CCD choices,
given currently available CCDs are tabulated below (Table 15.1) and plotted in Figure
15.1.

     Wavelength (nm)           DBSS/TSBS (VPH 772)         DBSS/TSBS (VPH 720)
           370                         1.14                       0.00
           400                         1.12                       0.00
           450                         1.06                       0.00
           500                         1.00                       0.00
           550                         0.92                       0.00
           600                        0.022                       0.75
           650                         0.02                       0.83
           700                        0.014                       0.86
           750                        0.013                       0.88
           800                        0.013                       0.95
           850                        0.014                       1.05
           900                        0.014                       1.06
           950                        0.010                       0.88
           950                        0.006                       0.69
Table 15.1 The ratio of throughput for DBSS/TSBS systems. A Fairchild CCD was
assumed for the TSBS. An EEV CCD for the "blue" arm of the DBSS and MITLL
CCD for the red arm.




                                        159             Comparison and Conclusions
                                                                                         AAOmega CoD



                                           Throughput of DBSS / Transmissive

                                  1.2
   Throughput DBSS/Transmissive


                                   1

                                  0.8

                                  0.6

                                  0.4

                                  0.2

                                   0
                                    300   400     500      600      700        800      900     1000
                                                        Wavelength /nm


Figure 15.1 Plot of the relative efficiency of the DBSS/TSBS concepts from 370-
1000nm, including VPH grating and typical detector efficiencies.


                                          Throughput of DBSS / Transmissive
                                                Optics only - no CCD or VPH
                                  1.6
   Throughput DBSS/Transmissive




                                  1.4

                                  1.2

                                   1

                                  0.8

                                  0.6

                                  0.4
                                  0.2

                                   0
                                    300   400     500      600      700        800     900     1000
                                                        Wavelength /nm


Figure 15.2 Plot of relative throughput (optics only) of DBSS/TSBS optical concepts
from 370 to 1000nm.



                                                             160              Comparison and Conclusions
                                                                       AAOmega CoD


The relative efficiencies of the systems are very dependent on the detectors chosen
and the efficiency of the optical coating. The efficiencies of CCD's have been
improving greatly over the past few years and they are expected to continue do so.
The detector for AAOmega will not be chosen at this stage and will be deferred to
take full advantage of the latest developments when the purchase becomes necessary.
Comparative throughput of the systems without CCD or grating (optics only) is given
in Figure 15.2.

There are uncertainties in the transmission values for both concepts. The TSBS
concept has very a large number of air/glass surfaces (20) and the MgF 2 + Solgel
coatings may not perform as well as predicted. Only a small change would make a
significant difference. The DBSS makes certain assumptions about preservation of the
hole in the primary mirror within output pattern of the fibres (i.e. that it is not
completely smeared out by FRD) and the obscuration introduced by the detector and
slit assemblies within the system. However, the TSBS system performance is only
likely to degrade, where as the DBSS has the small possibility that the obscuration
losses as a result of FRD (i.e. the loss of the primary mirror hole pattern) may be less
than predicted.

15.2 Throughput in IFU mode
It is expected that the inherent spectrograph transmission in IFU mode for the TSBS
will remain the same as that in MOS mode. However, due to obscuration losses the
DBSS system is expected to have significantly higher losses in IFU than MOS mode.
The increased loss has two sources. Firstly the SPIRAL IFU was designed to feed the
f/4.8 SPIRAL spectrograph and has a much slower output beam than the AAOmega
MOS fibres. This will significantly under fill the DBSS and the inherent central
obstructions will have a much greater impact on the throughput. Secondly, the IFU
microlenses re- image the telescope entrance pupil onto the fibre face and as a result
the fibre output beam cone will tend to be, if anything, centrally weighted, unlike
MOS cone in which some the telescope primary mirror central obstruction is expected
to be preserved. Both these factors combine to the detriment of the DBSS IFU
throughput. A detailed estimate for the increase obscuration losses in DBSS system
has not been made, but the system transmission, as represented in Figure 7.25 would
be expected, very roughly, to fall another an other 5-10%.

Based on overall inherent spectrograph throughput the TSBS wins out over the DBSS,
but not by a large enough margin to rule out the DBSS. The significance of the
throughput difference depends strongly on the relative importance of the MOS versus
the IFU mode (TSBS clearly wins) of operation and the requirement to obtain high
throughput down to 370nm, where the DBSS clearly wins out. However, if IFU
spectroscopy becomes a very high priority then a new IFU could be designed, for
relatively little cost (~$250K), that would better match the spectrograph and thereby
considerably reduce the impact of the obstruction losses in the DBSS.

15.3 Image Quality
Sub-sections 7.2.6 and 7.2.7 discuss, in detail, the effect of image quality on spectral
and spatial resolution. It can be seen from Tables 7.2 and 7.4 that the DBSS (3.1m to

                                         161              Comparison and Conclusions
                                                                      AAOmega CoD


9.3m RMS radius) clearly outperforms the TSBS (4.15m to 24.55m) by a very
significant margin, with the TSBS a long way from meeting the image quality
specification of 9m RMS radius. It is expected that which further development,
some improvements would be possible for both systems, but the DBSS system is
inherently much better in this regard and for IFU science the current TSBS design
significantly degrades the spectral resolution and has unacceptable overlap between
adjacent spectra in the spatial direction.

15.4 Wavelength range in a single observation
Due to the larger demagnification factor of the DBSS, and coverage of both arms
combined, the DBSS has a 15% larger wavelength coverage in a single shot.
However, the DBSS in high dispersion with, for example, 2500 l/mm grating, the
dispersion is ~0.2 Angstrom/pixel, giving ~400 Angstrom on the detector in either
arm (2048 pixels in spectral direction). With the transmissive system, the spectral
coverage is ~720 Angstrom with 4k spectral pixels. Therefore DBSS gives ~55% of
the contiguous wavelength coverage of the transmissive system. For many programs,
400 Angstrom of coverage at high resolution will be adequate, i.e. programs that are
only studying one line or a few lines that are close together. For other programs the
reduced spectral coverage would be undesirable: e.g. stellar spectroscopy in the blue,
where one typically wants to study many lines over a wider wavelength region. Many
stellar programs will want to cover the region from Ca H&K (3900 Angstrom) up to
Fe5335, with the highest spectral resolution possible. Of course, even the transmissive
system (~720 Angstrom total coverage) wouldn't be able to cover this region in a
single shot. However, the transmissive system would be able to do this in two s hots,
while it would take four with the DBSS. Even going from H&K to Hgamma (4340) in
one shot wouldn't be possible with the DBSS.

With DBSS you have the option of observing two very disjointed spectral regions at
high dispersion (e.g. Ca H&K at 3900 Angstrom, Ca triplet at 8600 Angstrom; Halpha
at 6560 Angstrom/ Hbeta at 4860 Angstrom. This is not possible with the
transmissive system. The DBSS could have several different dichroics to
accommodate a very wide range of different cuton/cutoff wavelengths. For example,
you could have dichroics with cuton/cutoff at say around 4500 Angstrom or 7500
Angstrom, which would allow studying two spectral regions both in the "blue" (or
both in the "red"). This is done in the ISIS instrument 14 that has a blue/red arm
combination with 4 dichroics that have their crossover wavelength at nominally 5400
Angstrom, 5700 Angstrom, 6100 Angstrom and 7500 Angstrom. However, one of the
advantages of the DBSS is the ability to optimise each arm over a narrower waveband
than the TSBS. The DBSS is likely to have one arm "blue" optimised and the other
"red" optimised, so with the "all blue/all red" dichroics the throughput would be
compromised in one arm.

It is not clear which concept provides the most desirable features with respect to
wavelength range covered in a single exposure, as it will be strongly dependent on the
science.


14
     http://www.ing.iac.es/Astronomy/instruments/isis/index.html

                                          162            Comparison and Conclusions
                                                                      AAOmega CoD


15.5 Spectral Resolution range
There is not much to chose between the concepts as both fulfil the functional
specification requirements. However, given identical demagnification of the systems
and identical gratings, the better image quality characteristics of the DBSS would
provide higher spectral resolution than the TSBS particularly in IFU mode.

15.6 Scattered light
Though this was thought to be an important comparison criterion, there has not been
sufficient time to carry out modelling of scattered light in either concept. It's is
therefore not at all clear which system would perform better in this regard. However,
the DBSS is a relatively simple and well-defined configuration and could be relatively
easily modelled. The TSBS system has a large number of air/glass (20) and
glass/glass (8) interfaces, so with uncertainties on surface finishes and coating, this
could be very difficult to model and has the potential to be highly significant.

15.7 Cost
The cost estimates of both concepts are effectively the same. However due to the
complexity and use of specialised glasses there is thought to be extremely high risk
associated with the TSBS optics cost which constitute a significant proportion of the
total cost (~$1M). The DBSS system is much simpler, which is reflected in the cost
estimate for the optics (~$0.64M), and much less risk associated with cost
uncertainties.

15.8 Completion date
It is estimated that both systems could be completed in approximately the same time
(2004). However, the DBSS is likely to require marginally more effort. The labour
estimates are ~$1.9M for the DBSS compared to ~$1.8M for the TSBS.

15.9 Support/Maintenance
Because of the added complexity of the internally mounted CCD's in the DBSS
cameras it is likely to be more difficult to support and maintain. Also due to the dual
beam nature of the DBSS there are more mechanism and associated electronics. In
view of the observatory's aim to significantly reduce the operations overhead at the
AAT this is perceived as reasonably important. However, the AAO does have a great
deal of experience with Schmidt camera systems.

15.10        Upgradability
As with support and maintenance the upgradability of the DBSS system is
compromised by the internally mounted detectors. Any modification could have a
significant impact on the detector packaging and obstruction losses. Replacing the
detectors is not ruled out in the DBSS, but it would require significant down time,
where as different detectors could be ready changed with the externally mounted



                                         163             Comparison and Conclusions
                                                                       AAOmega CoD


TSBS detector dewar. However, it is conceivable, given the relatively lost cost of the
DBSS camera assemblies, that the whole camera units could be exchanged.

15.11        Risk
The major risk areas identified as part of the concept study have been detailed in
Section 13. The risks associated with the spectrograph software and electronics are
virtually identical for both system, other than risk to the detector which have been
discussed in Sub-sections 15.9 and 15.10. The remaining risks can be broadly divided
into two main areas: the optical concept, and the associated mechanical design.
15.11.1         Mechanical risks
Most of the mechanical risks are very similar for both concepts. The complexity of
the articulation system design in the DBSS is greater than the TSBS, but is not
considered a high risk. The major mechanical risk difference between the TSBS and
DBSS is the camera, which for the DBSS is enclosed in a cool vacuum environment.
This does complicate the design and has a higher inherent risk as a result, but the
AAO has extensive experience with such systems and it is not considered an extreme
risk.
15.11.2         Optical concept risk
The risks associated with the performance of the two concepts have already been
discussed in Section 13 and Sub-sections 15.3, 15.6 and 15.7. There is believed to be
extremely high risk associated with the cost, availability, complexity and manufacture
of the glass components of the TSBS system. This includes deep aspheric lenses and
specialised glasses. This is in addition to concerns that even with further development,
the TSBS optical concept would fail to meet the required image quality specification.

15.12        Comparison Conclusions
The detailed comparison of the two main designs demonstrated that both concept have
their pros and cons. On balance of performance and estimated cost they came out very
close. The TSBS system won out in: overall system transmission (particularly in IFU
mode), inherent mechanical design risk, ease of maintenance/support, upgradability
and benefited from double the contiguous wavelength coverage at high resolution of
the DBSS.

The DBSS won out in terms of: delivered image quality (the poorer performance of
the transmissive system noticeably degrades spectral and spatial resolution
particularly in the IFU mode), blue performance, inhere nt spectral resolution limit,
better low resolution coverage, and ability to perform high dispersion observations of
two greatly separated spectral regions simultaneously.

The very high risks associated with the cost, availability, complexity and manufac ture
of the glass components of the transmissive system (including deep aspheric lenses
and specialised glasses) leads us to conclude that the DBSS is the preferred option for
the design of the AAOmega spectrograph. There are additional concerns that even
with further development, the transmissive optical concept would fail to meet the
required image quality specification. The estimated cost of the TSBS optics comes in
at approximately $1.0M with a considerable uncertainty associated with it. The DBSS

                                          164             Comparison and Conclusions
                                                                      AAOmega CoD


optics are much simpler, which is reflected in the cost estimate of approximately
$0.6M and the uncertainty in this is much smaller. We have reasonably reliable quotes
for the manufacture of the DBSS optics, but there have been considerable difficulties
in getting estimated costs of the glass blanks and manufacture of the lenses for the
TSBS. Even information about glass availability has been problematic.

If the IFU performance is viewed as a very high priority then it may be worth
reconsidering the TSBS, but any change in decision should be weighed off against the
relatively low risk option of making a new IFU at an additional cost of about $250K.
A new IFU may be regarded as a more expectable cost risk than the risk associated
with the TSBS optics, which could lead to an increase in project cost well beyond that
of a new IFU.




                                         165            Comparison and Conclusions
                                                              AAOmega CoD



16 List of Acronyms
AAO      Anglo-Australian Observatory
AAOSPS   Anglo Australian Old Stellar Population Survey
AAOUC    AAO User Committee
AAT      Anglo-Australian Telescope
ADC      Atmospheric Dispersion Compensator
CCD      Charge Coupled Device
DAO      Dominion Astrophysical Observatory (Canada)
DCG      Dichromated Gelatin (medium of VPH gratings)
DBSS     Dual Beam Schmidt Spectrograph
DQE      Detective Quantum Efficiency (of CCD detectors; allows for noise)
FEA      Finite Element Analysis
FoV      Field of View
FRD      Focal Ratio Degradation (of light transmitted through optical fibres)
FWHM     Full Width at Half Maximum
GMOS     Gemini Multi-Object Spectrograph
GPO      General Purpose Outlet (for 240 Vac electricity)
HWHM     Half Width at Half Maximum
IFU      Integral Field Unit
LDSS     Low Dispersion Survey Spectrograph
LDSS++   Upgraded Low Dispersion Survey Spectrograph
MOS      Multi-Object Spectrograph
MSM      Mean Sky Method of sky subtraction
PSF      Point Spread Function (image of a point object in an optical system)
R        Spectral Resolving Power, R = /
RQE      Responsive Quantum Efficiency (pure QE of CCD detectors)
SDSS     Sloan Digital Sky Survey
TSBS     Transmissive Single Beam Spectrograph
VPH      Volume Phase Holographic (diffraction grating)
Zemax    Optical ray tracing program




                                167                         List of Acronyms

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:40
posted:3/30/2011
language:English
pages:167