IV. General Requirements for SOAR Instruments

1. Introduction

Previous chapters of this document have outlined the current science goals of the SOAR partners, the general science requirements that are already being incorporated into the conceptual design of the SOAR telescope, and general performance scalings that define how best to exploit the telescope. This chapter develops the priorities and guidelines for instruments. It is not the intent of the SAC to ``design instruments by committee''. But, to ensure that partner science goals and Observatory operational requirements are met, instruments must follow the specifications and address the priorities outlined herein. The SAC has selected a ``science over the isokinetic field'' paradigm for the SOAR telescope. Instruments are therefore specified to span 4-6' -diameter fields where feasible. However, the partners also have a strong interest in stellar spectroscopy, with high throughput to compete with larger telescopes. When these observations are sky limited or are of crowded fields, they are best executed on SOAR rather than the Blanco 4m. Reducing the instrument field to a few arcsec\( ^{2} \) from \( ~ \)50 '\( ^{2} \) will simplify optics hence, in stellar spectrometers, should maximize throughput down to the atmospheric UV cutoff. Finally, there is interest in using Gemini instruments when they are not mounted on the nearby Gemini-S telescope.

2. Provisions to Accommodate Gemini-class Instruments

A goal for the SOAR telescope is to accomodate a generic Gemini instrument. The specification is the unballasted mass of the Gemini Near-Infrared Spectrometer (GNIRS, which weighs as much as, but is smaller in two of its dimensions than, the Gemini limit) because this is the only Gemini instrument that has been identified as a candidate for SOAR. A bore-sight instrument, its image slicer spans up to a \( 6\times 8 \) arcsec\( ^{2} \) field. Because this instrument is sensitive to the thermal-IR, it is advisable to minimize the number of warm fore-optics. The approach with the least optics is to adopt its f/16 as the f/ratio of the SOAR telescope, and this is the prefered route because SOAR will attain best images in the thermally sensitive H and K-bands.\footnote{ Adopting f/16 complicates the design of some visible-band instruments because the optimal match to CCD pixels (\( ~eq \)15 \( \mu \)m) leads to a telescope of f/9. The optical imager requires a refractive focal reducer/corrector. } Modifications to a Gemini IR instrument are minor:
  1. The radius of field curvature at the f/16 focus will differ between SOAR and Gemini. Both telescopes are close to their respective Ritchey-Chretien (RC) designs, which differ because of different back focal distances and physical plate scales. In practice the difference is negligible (see SOAR Optics Definition {[}SOD{]} document for details) over the field of the GNIRS. It becomes significant over the full 7\farcm 5-diameter Gemini field at SOAR (see the SOD.)\footnotemark{}
  2. A clone of the GNIRS will need a slightly different pupil cold-stop when it resides on SOAR because the spider morphology differs and because SOAR's pupil is at M1 not M2. If the cold stop cannot be changed easily then SOAR will have a TBD% increase in emissivity from the unocculted spiders and different pupil location.
  3. The pupil is closer to the focal plane in SOAR than Gemini. On Gemini it is at M2, 16m away. On an f/16 SOAR it is at M1, 2m closer, which is a minor difference (see the SOD) but requires that the cold-pupil in a SOAR IR instrument be undersized to avoid seeing the warm telescope. The ability to change pupil stops is not a specification of current Gemini IR instruments, but is easily designed into new ones. Gemini IR instruments do not have rotating pupil masks because the M2 spiders in Gemini contribute only \( ~eq \)10% of the 2.5% emissivity of that telescope (most comes from the mirror coatings) and an even smaller fraction on SOAR. (SOAR's emissivity specification is discussed in \S\ref{emissivity.sec}.)
\footnotetext{ Flattening the f/16 field of SOAR produces 0\farcs06 of defocus at 3' \ radius, so the optical implications of adopting even a flat field are not large. }%

3. Derived Instrument Requirements

To maximize the range of partner science that will execute optimally on the SOAR side of the SOAR/ Blanco system we need five facility-class instruments on ``cold standby'' at the Nasmyth and bent-Cassegrain foci. The considerable mass of the Gemini-class instrument on one side will be somewhat counter-balanced by two instruments on the other Nasmyth port, see Fig. \ref{instlayout.fig}. Up to two additional ``user'' instruments may be present at the other bent Cass. ports.\begin{figure}[hbt] {\centering \resizebox*{0.9\textwidth}{!}{\rotatebox{-90}{\includegraphics{allrot.PS}}} \par} \caption{{\small \label{instlayout.fig}SOAR telescope foci with proposed instruments. An annular field to monitor field derotation is always accessible around the folding M4. Facility Calibration Units are located inline of the IR imager and IFU spectrometer. They feed these instruments directly, or use a double-sided M4 to direct light to the high-throughput optical spectrometer and to the GNIRS. Instruments at the bent-Cass. ports and on the Nasmyth benches rely on their own calibrators. Tip/tilt sensors near the image focus of all instruments must be used to attain the delivered image quality spec. {[}REPLACE W/ CURRENT{]}}\small } \end{figure}

3a. General Instrument Requirements

These requirements help to ensure that the high image-quality delivered by the telescope is preserved at the detector, and that high operational efficiency is achieved. Most are elaborated upon in later sections.

Required Compatibility with Guiding Strategy

  1. \textbf{Project image-degradation specifications can only be met by using a tip/tilt guider.} All instruments must accept the quad-cell guiding sensor provided by the SOAR project (or, like Gemini instruments, provide their own.) This sensor ranges across the tip/tilt isokinetic field (5' -diameter), and sits close to the telescope focal plane to measure critically the instrument flexure. These sensors are discussed in \S\ref{oiwfs.sec}.
  2. \textbf{All instrument ports must accomodate at least one probe in the peripheral field to monitor field derotation.} The IR imager and spectrometer designs must provide a guiding strategy for work on dark clouds where visible-band guidestars will be very faint. (Gemini IR instruments will use IR sensors.)

Instrument Efficiency Requirements

  1. An instrument must fully configure itself for science within 2 minutes of its selection. During this interval, M3 and M4 will turn to the relevant port, and the peripheral guide probes and tip/tilt sensor of the newly selected instrument will acquire guide- and tip/tilt-reference stars.
  2. Instruments should not increase the light scatter of the telescope optics by more than TBD % where it makes sense scientifically.

Required Compatibility with Telescope Utilities

  1. Closed-cycle coolers will be used for the normal operation of all IR detectors and instruments. LN\( _{2} \) and heaters can be used to increase \( \Delta T \). LN\( _{2} \) in at least 30-hour holdtime dewars will be used for optical CCD's.
  2. Instruments will must be able to operate on 50-60 Hz power. Only 50 Hz will be provided at C. Pachon , with 120, 220, or 380 V available. All instruments must be compatible with the TBD grounding strategy to be adopted by the Observatory. Linear rather than switching amplifiers are prefered to minimize pickup. Instrument electronics must fit into a cooled Gemini rack enclosure (similar to what the Project will adopt) with dimensions 1.3m tall, \( ~ 1 \)m deep, and ( ~ 0.7 \)m wide. They weigh 94 kg, handle up to 1.5 kW of power, and leak <50 W into the environment.
  3. All instruments must be compatible with the focal-plane utilities. The Project will adopt those enumerated in the Gemini ICD (Interface Control Document) 1.9/3.6 Science and Facility Instruments to System Services. This document also describes the required connectors. The full suite of utilities at SOAR will include dry N\( _{2} \), 80-100 psi air flowing at 120 l/min, ethylene-glycol/H\( _{2} \)O at 0\arcdeg\ C and 15 psi flowing at 6 l/min, a He supply at 300 psi flowing at 3200 l/min, and a vacuum system to control instrument actuators.
  4. All instruments must be compatible with the Project-provided array controllers. These are TBD, see \S\ref{ccd.sec}.

Calibration Requirements

The required level of final calibration is itemized in \S\ref{calibration.sec}.
  1. All instruments must be stable enough to be calibrated to TBD levels using database entries. This permits a quicklook data-quality capability to enhance operational efficiency.
  2. Nasmyth instruments must be fed by the Facility Calibration units (unless the instrument is fully self-calibrating to the final level itemized in \S\ref{calibration.sec}.)
  3. Instruments at the bent-Cass. ports must provide their own calibrators.

Instrument Mechanical Design Requirements

  1. Spectrometers built without integral-field units (IFU's) or image slicers must attain full spectral-resolution with at least a 0\farcs5-wide slit.
  2. Uncompensated instrument flexure must change image centroids by <1/4 of the shift introduced by field derotation during a typical science exposure (up to 1 hr for the visible imager, and both visible and near-IR spectrometers; 15 mins for the near-IR imager.) This requirement applies for zenith angles ( \leq \)60\arcdeg.
  3. Instrument optical designs must include end-to-end error assessments that include atmosphere, diffraction, opto-mechanical tolerances, the alignment strategy, baffling, and optical-inhomogeneity effects. Atmospheric structure relevant to seeing will be parameterized from data being collected by Gemini at Cerro Pachon. The Project will provide seeing models in the form used by the Skylight package, BRDF's of the telescope at certain instrument entrance apertures, and the complete prescription of the telescope (including its polarization characteristics at its various foci.) These can be used by ZEMAX or Code V.

Thermal IR Requirements

The SOAR telescope is not being optimized for thermal-IR performance.
  1. The thermal emissivity of IR instruments must be <3.5% (i.e. <1/2 that of the telescope) in the low-emissivity atmospheric windows longward of \( \lambda \)2.5 \( \mu \)m if the IR array used is sensitive to this wavelength regime. Telescope emissivity is discussed in \S\ref{emissivity.sec}.
  2. For \( <\lambda \)2.2 \( \mu \)m, all near-IR instruments must have an internal instrument background of <0.5 e/pixel/s. This ensures sky-background limited performance.
  3. The near-IR instruments will need foci that are telecentric to the telescope pupil for optimal thermal control.

3b .Spatial Sampling Requirements

Imagers

Our simulations (Diaz, SOAR Internal Report) show that to minimize sampling error in stellar fluxess obtained from PSF fitting, we must sample the FWHM with at least 3-pixels. For top-quartile, tip/tilt stabilized images, we therefore need 0\farcs10 pixels in the optical and 0\farcs08 pixels in the near-IR (both set at their smallest image points in the curves plotted in Fig. 2 of Part III SOAR's Place in Southern Hemisphere Astronomy.) Binning \( 2\times 2 \) gives 0\farcs20 (0\farcs16) pixels in the optical (near-IR), i.e. 2.1 pixel sampling in bottom-quartile conditions. The combination properly over-samples 75% of the time, and allows us to back off a bit if we choose to bin in the worse 25%. In the optical a \( 4K^{2} \) \( ~eq \)15 \( \mu \)m pixel CCD or mosaic covers the isokinetic field.

Spectrometers

IFU/image slicers restrict sky coverage if the focal plane is over-sampled with 3 pixels. This is because the spectral resolution uses up most of the available detector pixels. Being more limited by sky coverage than by sky flux for most spectral resolutions, we can back off to 2 pixels in best conditions and accept spatial binning in median or worse. We therefore require 0\farcs15 pixels for optical and 0\farcs12 pixels for the near-IR.

3c. Spectral Requirements

Fig. \ref{variance.fig} shows, for SOAR's first-light detectors and good operating conditions, where in the spectral-resolution vs. source-flux plane SOAR beats the Blanco 4m. Gemini wins when these telescopes are photon-starved.\begin{figure}[hbt] {\centering \resizebox*{1.01\textwidth}{!}{\includegraphics{snr.eps}} \par} \caption{{\footnotesize \label{variance.fig}}Fractions of the total variance for SOAR that arise from (left) sky, (middle) shot, and (right) readout noise at \protect\( \lambda \protect \)0.65 \protect\( \mu \protect \)m, various spectral resolutions, and in 0\farcs45 seeing. Sky brightness is for CTIO dark time, mean extinctions at an airmass of 1.2, total telescope+instrument efficiency of 23%, slit matched to seeing FWHM, detector scale 0\farcs1/pixel, readout noise of 1.5 \protect\( ^{-}\protect \)rms, dark noise \protect\( 1.2\times 10^{-4}\protect \)e\protect\( ^{-}\protect \)/pixel/s, no binning, 1-hour exposure, and 100k e\protect\( ^{-}\protect \) full well. } \end{figure} This figure implies that the optical spectrometers on the SOAR telescope should have resolving power \( R\leq 30,000. \) Larger R at SOAR is less important because SOAR will then be no better than the Blanco, and for the case of high-resolution spectroscopy of faint objects will be \( 16\times \) slower than Gemini.

3d. Instrument Flexure Requirements

Spectrometers need <0.1 pixel flexure in 1-hr (\( \leq 15 \)\arcdeg\ change in zenith angle) for accurate sky subtraction. A TBD fraction of the image degradation with zenith angle z of (\( \sec z)^{0.6} \) will be allowed for flexure in the system error budget.

3e. Instrument Compensation for Atmospheric Dispersion

\begin{figure}[hbt] {\centering \resizebox*{0.45\textwidth}{!}{\rotatebox{-90}{\includegraphics{dispersi.ps} }} \par} \caption{\label{dispersion.fig}{\small Corrections in arcseconds that are required for STP-conditions at Cerro Pachon, at different wavelengths and zenith angles.}\small } \end{figure}It is desirable to correct this effect (see Fig. \ref{dispersion.fig}) with an Atmospheric Dispersion Corrector (ADC) in the optical imager when using broad-band filters. Because the ADC is the only element in the current imager design that attenuates light below 360 nm, it should be removable. Dispersion can be ignored for the near-IR until SOAR implements AO. If the visible spectrometer incorporates an IFU with full spatial-sampling rather than a long-slit or slitlets, then an ADC is not required for exposures near the zenith. (The varying airmass away from the zenith smears the spectrum over more pixels spatially, reducing signal-to-noise especially in the blue.)

3f. \label{calibration.sec}Instrument Calibration Requirements

Our philosophy is that, for efficiency, it will be highly desirable to use as many standard calibrations as possible that are taken within several days of the science data. However, it must be possible to calibrate instruments on a night-to-night basis to optimize results under the most critical circumstances.

Requirements Common to a Number of Instruments

Nasmyth instruments will be served by Facility Calibration Units (FCU). These units are designed to ensure that the distribution and direction of rays in the flat field illuminator match those of the telescope pupil, and to provide a high surface-brightness comparison or flat-field illumination source. Our specifications are adapted from the Gemini document Facility Calibration Unit Functional Requirements by D. Simons (v1.1 3/95), which is attached in Appendix X for reference. Briefly, the unit will provide both continuum and arc spectral-sources for optical and IR instruments. It will produce a surface-brightness gradient of \( ~ 10% \) across the field of view. This will need to be calibrated occasionally against the night sky, to determine the low-order corrections needed to reach our required accuracy of <1% variation. A separate calibration strategy must be developed for each instrument at the bent and folded Cass. ports. This will include the use of a dome flat-field screen, to be supplied as a facility capability.

Calibration requirements of specific instruments

Here we outline additions to the Gemini document that we will require for SOAR. \paragraph{IR imagers } achieve <5% photometric accuracy from dithered sky flats. Gemini feels that a projection flat-field system will be required to do \( ~ 1% \) photometry. SOAR will provide this from the FCUs, over the 3-4' \ fov of the IR imager. Flats will be acquired as the difference between two frames, one with the calibration lamp on and one off (see PASP \textbf{104}, 441.) \paragraph{Spectrometers } must scatter <10\( ^{-4} \) of night-sky line flux. Dense calibration lines will be provided by many orders from a small, fixed-gap etalon in the FCU. \paragraph{CCD's} Optimal UV transmission is provided by a thinned backside illuminated chip, but existing UV-optimized CCDs show fringing of 0.5--15% (depending on manufacturer) Peak-to-Valley with fine structure. In principal fringing is directly calibrated in spectrometers if the telescope pupil is correctly mimicked by the calibration system (i.e. all rays reach the CCD from the right directions), and if the spectrometer does not flex. The last is the big problem, and can be dealt with by minimizing flexure in the design and by calibrating the residuals at all telescope and instrument positions. In the case of intermediate- and broad-band direct imaging, fringing is extremely difficult to calibrate because the flat-field source does not have the same spectrum as the night sky. Deep-depletion, high resistivity chips provide higher QE in the R/I bands with minimal fringing while preserving the high DQE in the visible of blue-optimized chips. It therefore makes sense for one of the two CCD's in SOAR's optical mosaic imager to be optimized for the UV and the other to be a deep depletion chip with performance tuned for the red/I.

3g. Instrument Throughput Requirements

\label{emissivity.sec}IR Thermal Emissivity Requirements

\begin{figure}[hbt] {\centering \resizebox*{0.85\textwidth}{!}{\includegraphics{sky3.PS}} \par} \caption{{\small \label{thermal.fig}The total background radiation entering an IR instrument on SOAR. This assumes an atmospheric temperature of 253K, airmass 1, 1.2mm of water vapor, a warm optical train of M1-3 at 275K for a total emissivity of 7%. (Insert): sky transmission at Cerro Pachon, with 1mm of precipitable water vapor at an airmass of 1.}\small } \end{figure}These are summarized in Fig. \ref{thermal.fig}. SOAR aims to obtain 7% emissivity, <1/2 that of the Blanco 4m. This will require regular CO\( _{2} \) cleaning and perhaps also monthly mirror washes, so the telescope is being designed to facilitate these activities. In addition, both M4's should be parked in a downlooking position when not in use, covered if possible by a large ``dark slide'' immediately in front of the mirror surface. A proof slide should be mounted immediately adjacent to it, to test for dust accumulation.

Reducing Reflective and Transmissive Losses on Optics

Table \ref{coatings.tab} summarizes requirements and goals, assuming a Sol-Gel-on-MgF\( _{2} \) coat on each lens surface.\begin{table}[hbt] {\centering \begin{tabular}{|c|c|c|c|} \hline {\small Broad-band}& {\small \( \lambda \)\( \lambda \)0.33-0.4}& {\small \( \lambda \)\( \lambda \)0.4-0.7}& {\small \( \lambda \)\( \lambda \)0.7-1.1}\\ \hline {\small Reqd.}& {\small <1%}& {\small <1%}& {\small <2%}\\ \hline {\small Goal}& {\small <0.5%}& {\small <0.5%}& {\small <1%}\\ \hline {\small Narrow-band}& {\small }& {\small }& {\small }\\ \hline {\small Reqd.}& {\small <0.5%}& {\small <0.5%}& {\small <1%}\\ \hline {\small Goal}& {\small <0.2%}& {\small <0.2%}& {\small <0.2%}\\ \hline \multicolumn{3}{|l|}{{\small Reflectivity (fresh coatings)}}& {\small }\\ \hline {\small Reqd.}& {\small >0.88}& {\small >0.88}& {\small >0.84}\\ \hline {\small Goal}& {\small >0.92}& {\small >0.98}& {\small >0.98}\\ \hline \end{tabular}\small \par} \caption{\label{coatings.tab}{\small Limits on coating and reflectance losses at each optical surface.}\small } \end{table}

3h. Instrument Control Requirements

Instruments must interface with the TBD Data Handling System. This will be based on Gemini concepts, but not EPICS messaging.

\label{ccd.sec}CCD Array Controller Capabilities

Instruments must be compatible with the SOAR project array controllers. The SOAR project has identified science that requires high time-resolution over the entire detector array. This requires a multi-channel controller that reads out at a rate of order 20 Mpixels/s and can control either TEC or LN\( _{2} \) coolers. There are two possibilities which also satisfy a commonality requirement to ease maintenance and support. Gemini has adopted the Leach Gen2 controller from San-Diego State University and are awaiting delivery at the end of 1998. Another possibility is an ARCON from CTIO. Both would run the optical and near-IR arrays. The SOAR project will decide between these CCD controllers toward the end of 1998.

\label{shuffle.sec}Non-standard CCD readout modes

\paragraph{HIGH TIME-RESOLUTION WITH SUB-ARRAY READOUTS } Many SOAR science programs require time resolution at rates that exceed the (0.5,0.1) Hz time for a full-frame (IR,optical) readout. Several of these programs would benefit from complete readout control, whereby charge in each CCD row is stepped across the chip at a rate precisely synchronized with that of the astrophysical process under study. Drift scanning is one example, but transfer rates up to 100 Hz/row are also required. This is actually a special case of \paragraph{CHARGE SHUFFLING} This mode is relevant to optical CCD arrays. It operates by clocking charge back and forth across an array, periodically (every 1-10 minutes) moving part of the charge under an opaque mask that obscures part of the chip. If other characteristics of the instrument are ``beam-switched'' in synchronization, it becomes possible to cancel out ALL time-dependent systematics without incurring the time or noise overhead of multiple image readouts. An application of this is tunable-filter imaging. Here a Fabry-Perot etalon is varied in its wavelength transmission (by altering the plate separation hence optical gap) between on- and off-bands that are selected entirely for their astrophysical content rather than what filters one happens to have. Altering the gap alters in no way the optics or illumination path. Consequently, \textbf{all} time-dependent effects that normally limit the reliability of image subtraction are avoided, and very deep narrow-band images result that can go far below sky. This technique works equally well for long- or multi-slit spectroscopy, where the telescope is now ``nodded'' onto blank sky every few minutes, as charge is shuffled under the mask, to completely remove time-variable night-sky emission. These operating modes may drive specifications for the telescope-instrument interaction just as much as does mid-IR beam-switching. Instruments which might not otherwise need interchangable focal-plane maks will require them to support various charge-shuffling modes. These masks can be introduced at the telescope focus (which is curved), best at intermediate foci in reimaging cameras (which can in principal be designed flat), or at the detector itself.

Control Modes for IR Arrays

IR arrays are read out using a direct capacitative connection to each pixel, so there is no analog to charge shuffling while integrating. There is no capability for on-chip binning and so no reduction in read-noise. However, because the signal charge is not destroyed on readout, read noise can be reduced by multiple reads of the same datum. One method is to make n reads at both the start and the end of an integration, achieving a \( \sqrt{n} \) reduction. Another method is to read at intervals as the charge accumulates during an integration, and fit a slope to the result. Minimum integration time will be the readout time unless a shutter is used. The total readout time for an array depends on the per pixel clocking speed (adjustable in the controller, but faster is noisier) and the number of pixels per output (fixed by the multiplexer {[}MUX{]} design). MUXes and controllers designed for InSb arrays usually support high read-rates to allow operation in the high-background thermal IR. MUX flexibility -- for example the ability to define arbitrary subrasters -- may be sacrificed to achieve this goal. HgCdTe systems need not provide such high read-rates, and may choose not to, to reduce the complexity and cost of the associated electronics. MUX flexibility may also be sacrificed on the grounds of simplicity and cost. The bottom line: one must understand the performance specifics of a given device to assess what is permitted operationally. Examples: the \( 1024^{2} \) HgCdTe arrays produced by Rockwell for U. Hawaii read out a frame in four independent quadrants no faster than 500 millisec. The ALADDIN InSb \( 1024^{2} \) devices also have four quadrants but use 8 output lines per quadrant to read out up to \( 10\times \) faster (50 millisec.) Electronics for InSb are correspondingly more complex and costly compared to HgCdTe.

4. The Operational Environment for SOAR Instruments

This section describes the environment provided by the SOAR telescope in which the instruments will operate. Specifications are found in Chapter II SOAR Telescope Requirements .

4a. \label{oiwfs.sec}On-Instrument Tip/tilt Sensors

\textbf{The specified Project image-quality can be attained only if an on-instrument tip/tilt sensor is used.} This will be fed from a prism or 45\arcdeg\ mirror to sample a bright enough star to close the tip/tilt control loop. The field of view of the probe will be very small (TBD arcsec\( ^{2} \)), to minimize obscuration of the science field. The density of stars suitable for tip/tilt control --- 800 stars/deg\( ^{2} \) with mag-R <17.5 at the South Galactic Pole (see Table \ref{tiptilt.tab}) --- means that the sensor must span ( ~ \)14 '\( ^{2} \) to ensure 3 candidates on average. This is within the expected isokinetic field. Cases where there is no tip/tilt reference star will be relatively few. The probablities of getting such a star are similar for SOAR and Gemini, i.e. >90% anywhere in the sky. (The true numbers are somewhat uncertain. Optimistically the probabilities could be as high as 99% even at high latitude.) Sampling at 400 Hz will allow the tip/tilt unit to fully correct telescope motions up to 40 Hz, and would get the tip/tilt correction above at least the first harmonic of the lowest telescope resonances.\begin{table}[hbt] {\centering \begin{tabular}{|c|c|c|c|c|} \hline {\small R-mag}& {\small Photons/4m}& \multicolumn{2}{|c|}{{\small \# stars in (60}{\small '}{\small )\( ^{2} \)}}& {\small quad-cell S/N}\\ {\small }& {\small 0.3 \( \mu \)m/msec.}& {\small Lat. 30}{\small \arcdeg} {\small }& {\small Pole}& {\small 400 Hz, RN=5e\( ^{-} \)}\\ \hline {\small \( \leq 14 \)}& {\small 990}& {\small 5}& {\small 2}& {\small 60}\\ \hline {\small \( \leq 15 \)}& {\small 400}& {\small 9}& {\small 4}& {\small 24}\\ \hline {\small \( \leq 16 \)}& {\small 160}& {\small 17}& {\small 5}& {\small 10}\\ \hline {\small \( \leq 17 \)}& {\small 60}& {\small large}& {\small 9}& {\small 4}\\ \hline \end{tabular}\small \par} \caption{{\small \label{tiptilt.tab}Photon rates and R-band guide-star visibility. RN=5 assumes a fast-read CCD rather than a 4-element APD. }\small } \end{table}

4b. Object Acquisition

A guide camera will be provided at each Nasmyth side to 1) position accurately an ``invisible'' object onto narrow (0\farcs2) spectrometer slits, and 2) verify field derotation. For instruments with integral field units (IFU's) or image slicers, acquisition requirement 1) is relaxed. Our acquisition tactic is that we know where all stars bright enough to guide on are, so we only need to view \( ~ 400 \) arcsec\( ^{2} \) to verify pointing. (This is an appropriate error box to show the telescope operator in early operations.) The Gemini OCS to be adopted by SOAR will incorporate the HIPPARCHOS/TYCHO catalogs for astrometric reference. These will provide Johnson B- and R-mag catalogs down to 21-mag with 0\farcs3 rms accuracy, star/galaxy id and ellipticity information, and allow the OCS to make first-order refraction corrections based on guide-star color. The Gemini Observing Tool will provide easy interactive selection of guide stars from these catalogs. \begin{figure}[hbt] {\centering \resizebox*{0.75\textwidth}{!}{\includegraphics{scienceprograms.ps}} \par} \caption{\label{scienceprograms.tab}{\small Percentages of partner science to be executed with candidate SOAR instruments. Here S = shot-noise limited, B = background-noise limited, D = detector-noise limited, F = diffraction limited, () signifies reasonable but not ideal performance. A dash in a telescope column means that that telescope+instrument combination is not competitive.}\small } \end{figure} It should be straightforward to monitor variations in the telescope-delivered image quality and atmospheric transmission during an exposure (although not the true seeing measured on the detector within the stabilized isokinetic patch) by grabbing occasional readouts from the guider. It is a project goal to provide occasional filtration of these images, to derive first-order color terms for extinction measurements. These would be tied in with multi-star measurements from a site-wide automated photometric telescope to show conditions along the telescope line of sight during the observation. \begin{comment} Data Reduction and Archiving Quick-Look Requirements The Project intends to adapt the Gemini Observatory and Data Handling Control Systems to the SOAR telescope. These systems provide quick-look capability built around 4 tools: data display tool, movie tool, data reduction tool (which runs scripts to permit quantitative scientific assessment of data), and the observing assistance tool (guide star selection and other observer support tasks.) The specific functions of each tool and their scientific requirements are based on the Gemini document Scientific Perspectives on Gemini Data Reduction by P. Puxley (v1.0 3/24/96). A copy of this report is included as Appendix Y. Other capabilities can be added as necessary. In general the TCS and instrument controller provide World Coordinates together with geometric distortions so that coordinates are available in both pixels and e.g. sky angles. Data error or quality (e.g. a power spectrum) can also be displayed. Data Reduction Pipeline It is an eventual project goal to fully reduce data shortly after the exposure ends. This allows the observer to refine subsequent exposures in the observing queue. It will usually be necessary to consult a database of calibration exposures. There are implications for instrument stability, and also for the philosophy of calibrating queue observations (i.e. whether to calibrate in advance of all potential observing setups, or to await observations then calibrate only setups which were used.) Basic reductions often depend strongly on instrument peculiarities, while final data quality may depend on the success of the first reduction steps. The most time-consuming phase of pipeline reduction is to find the best procedure for the various operational modes of each instrument. These procedures should be established by the end of the commissioning phase. This task will be better done by the instrument builders and frequent users of the telescope. Once the pipeline is well tested, it may be preferable to reduce all data in a standard optimized way rather than distribute custom IRAF software and cookbooks. Reduction of queue observations would be more difficult for the observer, and the Observatory will have all the required (shared) calibrations. Given that there will be many new concepts in SOAR instruments, we feel that keeping control and responsibility for data reduction within the SOAR operations team or in a central group for each partner may be the best way to assure the maximum possible quality for the data obtained. It is is likely that typical observers would have limited or no experience at all in using and reducing data taken with some of the planned instruments. Data Archive Requirements SOAR should archive all raw data and calibration frames, as well as the reduced-data products delivered by the operations team. We should at a minimum provide a means of recalling such data on an occasional basis (e.g. when the first distribution is accidently lost or destroyed.) However, the goal is not to provide a full retrieval archive at the level provided by e.g. HST or Gemini. Data storage requirements of SOAR instruments can be estimated crudely from their likely detectors. The HgCdTe-based near-IR imager will comprise 4-16 Mpixels, each pixel providing 32 bits. Of order 500 frames might be obtained nightly. The optical imager and MOS are both likely to use mosaics of \( 2\times 4K \) CCD's. We assume that \( ~ \)200 full frames might be obtained in combination, each of 16 bits. The IR spectrometer would be expected to generate \( ~ \)200 frames from its \( 1024^{2} \) InSb array, each 32 bits. The nightly total from science instruments is then \( 8\times 4\times 500+16\times 2\times 200+1\times 4\times 200=23 \) Gbytes, which will fit on 1.5 WO-DVD disks. Only 5% of a DS3-link would transmit each night of data in \( ~ \)200 minutes. The \( ~ \)600 disks produced annually will occupy <2 feet of shelf space. DVD jukeboxes will surely be available, and so access to at least the last two months should be automatic. The Gemini OCS is based on a commercial database. It is the intent of the Project to adopt this OCS, so its search and retrieval tools will be available. Others can be programmed as required. \end{comment}

5. Proposed Initial Instrument Complement

The SAC asked all SOAR partners to estimate how they would distribute their observing share between the different modes that are listed in the first column of Table \ref{scienceprograms.tab}. Based on the arguments of Table 3 in Chapter III (SOAR's Place in Southern Hemisphere Astronomy ), the SAC then assigned some of these modes to the Blanco telescope because they did not need the unique capabilities of the SOAR telescope. The SAC concludes that the SOAR telescope should have the following instruments to execute the desired scientific programs. Program options are strongly desired; goals can be accomodated as upgrades. Priorities for each instrument are ordered highest to lowest, with lower priorities possibly implemented as upgrades. The percentages in Table \ref{scienceprograms.tab} produce no clear priority for deployment. Therefore, instrument schedules will be paced simply by the finances available to the builders.

5a. Priorities for Very High-efficiency Optical Stellar Spectrograph

A simple short-slit instrument that emphasizes:
  1. highest throughput at R = 5000 to 850 nm. Intent is to maximize efficiency for work on faint compact sources, and to reach the Ca triplet at zero-redshift.
  2. UV response to atmospheric cutoff (320 nm required, 305 nm desirable). Intent is to access the large number of spectral diagnostics in the rest UV, an important underexploited niche for a S-hemisphere telescope (UVES is the only one doing it, at VLT UT2.)
  3. spectral PSF consistency appropriate to better than 1% sky subtraction over the useful field of view. Metric \( \Delta \)D(80)/D(80) < 2%. Intent is to minimize field variations in the monochromatic PSF for accurate background subtraction.
  4. low flexure of <0.1 pix / hr. Intent is accurate sky subtraction.
Program options:
  1. R up to 20,000 at lower throughput. Intent is to enable limited stellar work while not pushing R to photon starvation. Maximum R depends on CCD read noise.
  2. Operation to \( \lambda \)1 \( \mu \)m. Intent is to reach optical spectral features at higher redshift.
  3. ADC Correction down to 360 nm, must be removable to preserve U. Intent is to maximize S/N for exposures away from the zenith. Without an ADC this light is smeared across a region larger than the monochromatic seeing disk during the exposure. Conventional ADC's use glass types that absorb all UV.
  4. 3' \ fov w/multi-slit mask. This requires the ADC for accurate spectrophotometry. Intent is to enable study of clustered targets and not compromise sky subtraction and introduce scattered light with fibers.
  5. Motorized slit mask exchange. Intent is to improve operational efficiency by not involving the telescope operator/observer in this task.
Upgrades:
  1. > 3' \ fov w/multi-slit mask. Intent is to improve multiplex on certain targets, up to the size of the isokinetic field.
  2. Image slicer/dissector for spatial resolution over whatever few arcsec\( ^{2} \) field does not compromise requirements. Intent is to improve throughput and preserve full spectral resolution in poor seeing.

5b. Priorities for High-Spatial Resolution (IFU) Optical Spectrograph

A lower throughput device with more grating options that emphasizes:
  1. 2D-coverage over field using an integrated IFU or image slicer , 2-pixel sampling of best quartile, center field, tip/tilt stabilized images (0\farcs15/pixel near \( \lambda \)1 \( \mu \)m). If the IFU is fiber-fed, foreoptics will be required to speed up the telescope beam from f/16 to f/6-8.
  2. minimum of 1500 contiguous spatial samples , <5% of the cross-talk introduced by seeing. Intent is to properly sample a roughly \( 5\times 10 \) \arcsec\ field.
  3. R up to 30,000. Intent is to enable stellar work on the cores of bright star clusters.
  4. wavelength coverage \( \lambda \)\( \lambda \)0.36-1 \( \mu \)m with one octave (factor of 2) interval on the detector at once. Intent is to enhance operational efficiency.
  5. better than 15% throughput at 350 nm (including CCD + telescope). Intent is to study stellar populations in the cores of low-redshift galaxies among other things.
  6. low flexure < 0.04 pix/hr. Intent is to bench-mount this instrument at Nasmyth to attain this stability.
  7. sky subtraction 1% residuals over 180\arcdeg\ field rotation
  8. multiple fibers in fixed sky pattern (or applicable sky suppression strategy)
  9. operation to \( \lambda \)1 \( \mu \)m. Intent is to fully overlap in wavelength with the IR spectrometer.
Program options:
  1. > 15% throughput at < 350 nm (including CCD + telescope). Intent is improved performance for stellar population work.
  2. sky subtraction <1% residuals over 180\arcdeg\ field rotation
  3. provision for slit translation. Intent is to allow different fiber feeds to coexist, e.g. one might come from a bench-mounted AO system, the other directly from the telescope focal plane. This could be handled by a manual change.
Upgrades:
  1. Multiple IFU -- 1000 spatial samples divided between several deployable heads distributed across the 5' \ isokinetic field. Intent is to complement Hydra-S by providing spatial resolution at each point and sampling closer than 23\arcsec .
  2. Operation to \( \lambda \)1.4 \( \mu \)m with necessary thermal suppression.
  3. AO feed with spatial scale 0\farcs08/pixel to ensure 2-pixel sampling of top-quartile, center field, AO corrected images.

5c. Priorities for Near-Infrared Stellar Spectrograph Requirements

  1. bore-sight spectroscopy of stars over \( \lambda \)\( \lambda \)1-2.5 \( \mu \)m and several arcsec\( ^{2} \) for simultaneous sky determination, R=4000 will fill 2K\( ^{2} \) array w/ single atmospheric window, 2-pixel sampling of best quartile, center-field, tip/tilt stabilized images (0\farcs1/pixel around ( \lambda \)1.5 \( \mu \)m.)
  2. R <=18000 (2 pixels). Intent is to ``work between OH sky lines'' to reach darkest sky.
  3. Detector > \( 1K^{2} \), < 0.1e-/s dark, <30e- ron
  4. Slit > 20\arcsec. Intent appears to be sky subtraction?
  5. Throughput > 30% w/detector
  6. Flexure < 0.1 pixel/hour worst case. Intent is accurate sky subtraction.
Program Options
  1. 0\farcs3/pixel for median seeing
  2. R>20,000. Intent is to enable some work on stars.
  3. Cross-dispersion \( \lambda \)\( \lambda \)0.9-2.5 \( \mu \)m. Intent is observational efficiency on point sources.
  4. IFU
  5. Detector >\( 2K^{2} \), < 0.01e-/s dark
Upgrades
  1. \( \lambda \)\( \lambda \)0.9-5.5 \( \mu \)m operation. Intent is to span longer baseline for extinction measurements and to access a few spectral diagnostics in the ISM.
  2. Slit 60\arcsec. Intent is crude mapping of extended emission-line objects.
  3. \( \lambda \)\( \lambda \)0.9-2.5 \( \mu \)m echellete mode (R=2000)

5d. Priorities for Near-IR Imager

Requirements
  1. 0\farcs08 pixels. Intent is for the instrument to preserve the near-diffraction-limited performance of the telescope at \( \lambda \)1.4 \( \mu \)m.
  2. > 80\arcsec\ fov
  3. > 6 filter positions
  4. Cryo pupil stop, D(80)<1% of pupil size
  5. Throughput > 30% with detectors and filters
  6. \( \lambda \)\( \lambda \)1-2.5 \( \mu \)m
  7. \( \lambda \)2.5 \( \mu \)m dark current < 1\( e^{-} \)/s
  8. \( 1.6\times 1.6 \)\arcsec\ subarray (\( 20\times 20 \) pixels) readout at 100Hz
Program Options
  1. 20 filter positions
  2. fov> 200\arcsec
  3. \( 1.6\times 1.6 \)\arcsec\ subarray readout at 20Hz
Upgrades
  1. tunable filter. Intent is to avoid a large number of filters in the dewar.
  2. R<2000 grisms + aperture masks
  3. Coronograph
  4. >\( \lambda \)4 \( \mu \)m operation. Intent is to measure spectral energy distributions and extinction over a longer wavelength baseline.
  5. Minimum field of view of 200\arcsec\ diameter.

5e. Priorities for Optical Imager

Requirements
  1. 0\farcs08/ pixel. Intent is 3-pixel sampling of best quartile, tip/tilt stabilized images (i.e. around \( \lambda \)1 \( \mu \)m)
  2. \( 5\times 5 \)' \ fov. Intent is to cover the isokinetic patch in best quartile seeing
  3. ADC
  4. operation down to 320 nm
  5. Provision for a minimum of 6 parfocal filters
Program Options
  1. More filters. Intent is background sky supression in I using e.g. a Rugate filter (does not require a collimated beam.)
  2. Filter that gives spectral R tunable over 100-800. Intent is to enable sky-supressing modes of operation by frequency shifting, to avoid purchase of custom interference filters, and to do high-fidelity on/off-band image subraction.

6. Adaptive Optics for the SOAR Telescope

Even with this instrument complement, a SOAR telescope without adaptive optics (AO) will be beaten soundly in many important science areas by several other southern-hemisphere 4m-class telescopes. We therefore strongly urge at the earliest possible date a \textbf{low-order AO system} with priorities:
  1. Working over \( \lambda \)\( \lambda \)0.4--2 \( \mu \)m and a field of view of 1-2'
  2. Laser guide stars if technically practical.
The SAC believes that the highest-priority science application for this system is optical spectroscopy over a two-dimensional field of view. This is elaborated in \S2.3 of Chapter III SOAR's Place In Southern Hemisphere Astronomy. The most likely configuration is to bench-mount the AO system on one of the Nasmyth-focus optical benches, to feed the IFU spectrometer. \begin{comment} \begin{figure}[hbt] {\centering \resizebox*{0.5\textwidth}{!}{\includegraphics{ha.ps}} \par} \caption{\label{hadetect.fig}{\footnotesize S/N in H\protect\( \alpha \protect \) in the H-band at 2-pixel spectral resolutions R=1000, 3500, and 5000, together with the fraction of redshifts with given S/N or better. A system efficiency of 30% and exposures of 7500s are assumed. The H\protect\( \alpha \protect \) line flux is taken to be \protect\( 10^{-17}\protect \) rg/s/cm\protect\( ^{2}\protect \)arcsec\protect\( ^{2}\protect \), corresponding to a typical galaxy with an H\protect\( \alpha \protect \) luminosity of 0.1 L{*}(H\protect\( \alpha \protect \)), a star-formation rate of 3 M\protect\( _{\odot }\protect \)/yr, and a FWHM linewidth of 100 km/s.}\footnotesize } \end{figure} \end{comment} \end{document}