Goals of AST2:

The AST2 is the first large aperture telescope with a modern echelle spectrograph to be developed for remote, automated operations. The telescope will be sited in southern Arizona and remotely controlled from TSU in Nashville. Data, or data products, will likely be transmitted to astronomers at TSU and other locations via the Internet. While conceptually simple, implementation is potentially very complex. Much thought must go into the design and development of the hardware, the controls and software to ensure a relatively maintenance-free facility operable by preplanned schedules from a remote location. The data acquisition, reduction and archiving are equally complex as the facility is capable of generating large data volumes that must be recorded, reduced, archived and distributed. The intended detector is a 2kx4k SITe CCD. Given the echelle format and the intended program of relatively bright stars, data volumes of the order of several gigabytes per day are very probable.

Where the project is currently:

A high standard of excellence has been achieved in the CASS in the area of long-term, ultra-high precision photometric stability through the Automated Photometry Telescope (APT). These photometric results are recognized throughout the world as technically leading-edge, and as enabling a coherent approach to monitoring many solar-Iike stars for white light variations over year-to-year and eventually decadal time scales. This is research that clearly falls within well-established NASA interests of the Sun-Earth connection area. Dr. Greg Henry has established the careful procedures and control necessary to maintain year-to-year precisions at the 0.0002 magnitude level. The primary science results from this research can only follow after monitoring a large number of stars over greater than 10 years. Having an established reputation as a world leader in high precision photometry led to Dr. Henry's recent involvement in another science area of obvious interest to NASA: detection of extra solar planets. Follow-up photometry for planets, provisionally detected in radial velocity studies, is very useful in ruling out stellar variability as a source of the radial velocity signal. More importantly, for the short 3-5 day systems now being detected by the Marcy and Mayor groups, the chances of transits occurring given random system orientations on the sky is fully 10% for each planetary system. We may expect a radial velocity detected planet will be found transiting its host star as Dr. Henry examines just a few more candidates. A transit would fix the inclination allowing a direct (with knowledge of the same for the host star to -10% from general stellar structure considerations) mass determination from the radial velocities. The transit depth will provide a measure of the plallefs radius allowing determination of its density .The APT project is uniquely positioned for both fundamentally important long-term results in the solar-stellar, Sun-Earth areas and in the exciting short-term prospects for extra-solar-planet research.

The solid payoff in astronomical photometry results from having dedicated telescopes on which stable instrumentation can be maintained in support of long-term goals. An automated spectroscopic telescope nicely complements established capabilities in CASS and holds the promise of facilitating forefront research that could not be approached with multi-user facilities associated with large telescopes. The AST2 is being developed to support strong science programs and could become a valuable resource for providing optical spectra in support ofNASA space-based ultraviolet or infrared observations.

Operationally, the APT is a relatively simpler system as the remote control commands are fewer and the data volume is limited in size. As the automated stellar photometry utilizes standard phototubes, the data output is a simple data stream of photometric measurements and easily moved by available modem communications. The CASS has established the ability to remotely operate five small telescopes (and very soon to be eight!) in a routine manner. The center is now stepping up to much more complicated remote control systems as they move from a single cell detector to large format two dimensional detectors.

An Automated Imaging Telescope (AIT) is approaching operational status. It will utilize a lkxlk CCD and has the potential or producing an even larger data volume. The primary program is to monitor a selected list of galactic clusters for an extended period of time to search for planetary systems. This is a much more complex project as the data is two dimensional in nature. Detector operations, data recording, data reduction, archiving are likely to very large in volume. Indeed the AIT project promises to be a pathfinder for AST2. Currently, the AIT telescope is at the Arizona site. The AIT detector with controller is being prepared by a contractor located in Phoenix, Arizona. While not in the scope of this focused review, the review panel recognizes that AIT has the potential of both helping and hindering the AST2 project. As AST2 continues, careful attention must be paid to the tradeoff of developing the AIT just ahead, or in parallel with, AST2. There are likely to be conflicts of resources between the two projects.

As the fourth year of a five-year grant begins, the AST2 has moved forward due to the focused efforts of Dr. Joel Eaton. The telescope design is completed and most of the telescope hardware has been fabricated. An existing 2-meter, f/l.5 primary with mirror cell and a secondary mirror have been delivered. Assembly of the telescope is well along at the Tennessee State Hanger located at the Nashville Tennessee Airport. Site preparation at the Fairbom Observatory, located very close to the Arizona-Mexico border, is complete and the telescope building has been erected. Housing for the spectrograph is yet to be obtained, but a working design concept is in hand. The spectrograph design is at concept stage.

Recently, Dr. Frank Fekel has joined the AST2 team. He is focused on the science goals of the project and is addressing the data reduction, analysis and archiving issues. In the area of data analysis, the AST2 project has made a wise decision to contract with Dr. Jeffrey Hall of Lowell Observatory who has developed excellent software for analysis of data from similar echelle spectrographs. This is likely to be a cost and time effective means of obtaining software for use in the reduction and analysis during the routine phase of AST operation. Moreover, the illL software provides adaptive software analysis for testing the CCD performance and diagnosing spectrograph performance during the instrument checkout phase.

What remains to be accomplished, and the current schedule:

Given the resources available, the review committee is very impressed with what has been accomplished. Initially, CASS intended to purchase a telescope, but found that the bid process produced responses which exceeded the budget or, if within budget, were not credible. Instead. CASS made the decision to design and build the telescope with university staff. This has produced a less costly hardware, but with an overall delay of the development schedule. The telescope development is well on the way. However several issues must be addressed concerning the telescope performance, cost and schedule (see Issues section).

The spectrograph is a major challenge to the AST2 project. Based upon the committee's collective experiences, the spectrograph is nearly equal in cost and complexity to the telescope. Our best guess is that the spectrograph will take two to four years to complete with the clock beginning at the time a full time person is recruited to take responsibility for the spectrograph. Indeed the spectrograph is likely to be the pacing item in the overall schedule for AST2.

Our best guess is that the AST2 will take at least three more years to reach full operational stage.

Issues that must be worked:

Many activities remain to be accomplished with this project. Aside from a very top-level schedule, the committee saw no management plan, cost nor schedule. The committee recommends that a timeout be taken and that a simplified management structure be put in place as soon as possible. The approach should be as follows:

1. Carefully define the science goals in terms of desired spectral coverage, desired resolving power, spectral purity, signal-to-noise ratio in a specified exposure time for a defined stellar spectral type and magnitude. The goals must also include overall stability of the AST2 as the science goals are to monitor relatively bright stars for long term spectroscopic variability .As remote control is planned and large data volumes will be handled, the science goals must specify how often these stars will be sampled, what information must be extracted, how the data will be stored, what must be archived and distributed.
   la. While not in the scope of this review, the committee notes that the AIT has very significant resource requirements both in the commissioning phase and the operational phase. A management plan for AIT should also be designed to ensure that the resources of AIT and AST2 do not conflict.
2. Translate the science goals into AST2 perfonnance specifications. These derived specifications would address perfonnance of the telescope, the fiber optic transfer system, the spectrograph, the detector, control and data handling systems, remote operations, data capture, storage, reduction, analysis and archive. Each of these specifications must be quantifiable in order to compare to perfonnance as cost, schedule and staffing will have to be traded off for each specification and for the overall AST2 perfonnance.
3. Build a work breakdown structure based upon the perfonnance specifications and project areas that demand resources. This structure will need associated costs, estimated staffing requirements and schedule. We anticipate that this structure and the resulting Gantt chart will flesh out many conflicts in staff time, and costlschedule conflicts.
4. Revise the current proposal to reflect the cost and schedule that comes out of this activity .An example of expected changes is the cost and schedule of the spectrograph. The camera alone is likely to cost a few hundred thousand dollars and will require a few years from design to completion. Dr. Mark Adarms has provided a top-level example of a Gantt Chart for the AST2 (Attachment A).

The perfonnance of the telescope optics is unknown at this time. While the AST2 documentation calls for an image quality of eighty percent encircled energy within 0.6 arc seconds diameter, an overall performance budget is not in place. Dr. Larry Ramsey has provided a discussion on telescope error budgeting and mirror testing (Attachment B) as an advisory for determining the overall telescope performance. At this time the actual optical performance of the primary and secondary mirrors have not been assessed. We recommend such be done, and that further tests be planned for determining the robustness of the support structures, tracking, focus, pointing and facility-induced seeing. At the component and mechanical level tests should be done before shipping the telescope to the Arizona site.

The committee recognizes that AST2 has insufficient staffing to get the project accomplished within the next few years. We recommend the following additions to the staff:

1. Astronomer/instrumentalist to lead the spectrograph development. This individual would be responsible for translating the spectrograph performance specifications to hardware and operations. This individual should have an interest and a share in the intended observational programs planned for AST2 once it is operational. An immediate task would be to contract with an optical designer possessing considerable experience in ground-based telescope echelle spectrograph camera designs. Few individuals exist and are in heavy demand. Dr. Ramsey has provided an advisory on the development and performance of the spectrograph (Attachment C).
2. Engineer/technician with opticaUmechanical experience. This individual is needed to assist Dr. Eaton in the assembly and testing of the telescope, evaluation of the optical performance, transportation of the telescope to Arizona, assembly at Fairborn Observatory and commissioning. This individual would also be needed for the spectrograph activity , which is likely to follow the telescope schedule a year or two later.
   We point out a major safety concern. To date, much of the telescope assembly has been done without a second individual always present. While there are other individuals in the hanger working on Tennessee-state-owned aircraft, there is not an individual immediately there checking to make sure lifting and moving large mechanical items is being done in a safe manner. In the event of an accident, it is not obvious whether trained personnel would be immediately available.
3. Sufficient staffing to make sure that an overall management/cost/schedule plan is in place and that the project is moving ahead with the appropriate resources. A concern by the committee is that in past years, end-of-year funds have been expended for other telescope hardware when staffing was falling short on AST2. Indeed, with the AIT coming on line, even more staffing issues are potential.

The spectrograph is of size and class comparable to major instrument projects undertaken by organizations with much greater technical infrastructure than CASS. Completion in a timely and cost-effective manner cannot be done with just Joel Eaton doing both the telescope and spectrograph. The following steps should be taken:

1. First and foremost, a management, cost and schedule, as listed above, must be developed based upon quantifiable measures of the science goals that translate into telescope and spectrograph performance requirements.
2. Once a set of science driven requirements are specified, the spectrograph concept(s) need to be developed to a sufficient level that cost and performance trades can be done.
3. Another CASS staff member should be assigned primary responsibility for the spectrograph to relieve Dr. Eaton to focus on the telescope.
4. The spectrograph design should be pursued by an outside consultant (or consultants), whose work is monitored by CASS staff. This consultant should have experience in the use of optical fibers and spectrograph design. Once this step is complete, a technical approach should be selected for completion of the spectrograph.
5. Working with outside consultants, the services of an optical designer for the camera should be procured.
6. The spectrograph project should be reviewed at the concept and preliminary design stages at least. A critical design review should also be considered.
7. No further procurement of spectrograph hardware should take place until after a preliminary design review has been signed off.

The detector, controller and the telescope control is a very significant activity, likely to consume much resources as the hardware is purchased and the software developed. Getting optimal operation of the CCD is not a trivial activity as many timing and voltages settings must be tested to ensure good linearity range, minimal readout noise, appropriate voltage bias and no grounding loops. For both the AIT and AST2 this will prove to be a challenge. However there are a growing number of astronomers, instrumentalists and technicians with experience in this area, especially in southern Arizona. A plan to get the detector and controllers will have to be fleshed out.

We have three concerns for the commissioning and operational phases:

1. Achieving a fully remote operational stage and maintaining a remote program is likely to be staffing intensive at the Arizona site. Planning of the logistics, housing and review must be worked.
2. Checking out the detector operations and the telescope operations will not be trivial. The AST2 project needs to scope out what can be done at TSU and what must be done at the Arizona site to make the most cost-effective, staffing and schedule approach. During the spectrograph commissioning phase there will be a need for active interactions between someone who can use and modify the echelle data analysis software, for the purpose of optimizing both the software itself, but more importantly in these early stages optimizing the spectrograph setup as well. The need for data analysis expertise will arise in the earliest stages of detector and spectrograph testing and continue well into the first few years of routine operations.
3. Remote operations will require much data transmission. High-speed data lines are likely to be very costly to install with very significant ongoing costs. Tradeoffs of data transmission and mailing of data tapes should be done. The committee understands that cost of installation ofa high speed line between Fairborn Observatory and University of Arizona may be approaching $1 OOK, and that with the recent deregulation of communications fees, the annual costs are not known, but very likely they will be high. This may lead to a short-term solution of mailing data tapes with a long-term solution of real time data when necessary .Here technology is changing so quickly that a conservative approach should be considered for initial operations with annual reconsideration.

The overall analysis:

The committee is enthusiastic about the science goals of AST2 and very supportive of the effort. With revision of management plans and additional staff, we feel that the fully operational activities of AST2 can be achieved in the two to four year time period. We recommend that NASA HQTRs continue to provide resources to do such.

As AST2 is in a critical development phase, we recommend that technical reviews be scheduled for one day every six months in an effort to provide guidance and experience to a very challenging, but highly rewarding endeavor.

1) Introduction

The concept of a 2 meter automated spectroscopic telescope is an exciting one with tremendous potential to make both intermediate and long term contributions to time in a variety of areas and domains of astrophysics. While the science programs and top level philosophy reflect the combined thinking of the CASS staff, the telescope project to date has been carried principally by Dr. Eaton. He has a concept that is fundamentally sound and has clearly worked energetically, if not heroically, to implement it. However there are significant threats to the implementation that can be significantly reduced or eliminated by applying some basic systems analysis tools. It is the scope of this brief document to discuss the AST2 error budget and the ancillary need to assess the performance capability of the telescope optics.

2) System error budget

The primary mission of the AST2 is high-resolution spectroscopy. As such the ultimate performance figure of merit is the signal-to-noise (SIN) in the output spectrum. In this situation the image quality error budget that forms the foundation for most telescope system analysis, is only important to the extent that it effects the system transmission budget. All other transmission quantities being fixed, the (S/N) will vary with the fraction of the light in the image that is delivered by the telescope into the fiber. This in turn depends on the telescope optical image quality, guiding and focus accuracy, as well as atmospheric perturbations such as seeing and differential refraction. For the AST2 to be competitive the telescope system goal should be EE(80) into the 2.5 arc- second fiber at a median level. Further, the broad bandpass of the spectrograph demands that this EE(80%) requirement be met at the limits of the spectrograph bandpass when the image is dispersed by atmospheric differential refraction.

In order to facilitate understanding this process, I have constructed an imaging error budget based upon my (incomplete) understanding of the AST2 system. This error budget is summarized below. All items are in terms of EE(80). The telescope optics numbers come from document #5 supplied to the committee which called for a total delivered image quality EE(80)= 0.6 arc-seconds. The division between the primary and secondary is somewhat arbitrary. The telescope systems component is meant only to illustrate what contributes. The focus numbers come from assuming I pixel error in a f/8 focal plane where the pixels are 15 microns. The tracking numbers are derived from document #5 control system numbers. The facility number is an estimate for the residual seeing induced by the building, spectrograph room etc. Finally, I have assumed a median I arc-sec FWHM seeing typical of mid continental sites which translates to EE(80) = 1.9 arc-seconds. The differential refraction number comes from explicitly calculating the number for 370 to 700 nm at a zenith distance of 30 degrees; likely a typical number. This latter number is decreased from about 1 to 0.75 arc-seconds, for example, if the short wavelength limit is 410 nm. It is extremely important to understand that differential refraction must be considered when using a circular aperture such as a fiber as one cannot maximize the transmission by aligning the slit along meridian circle. The resulting 2.34 nominal EE(80) illustrates that the design numbers cannot be greatly exceeded without degrading the spectroscopic potential of the AST. As such it is critical that the AST team take the time to construct an error budget that reflects the current state of the system so they can rationally allocated resources early to those areas that will lead to the greatest performance improvement.

Table 1: Strawman AST2 Error budget


Total nominal                                                    2.34
EE(80%)

Telescope Optics                                        0.60
    Primary mirror                           0.50
       Diffraction             0.058
       Figure                  0.400
       Support                 0.300
    Secondary Mirror                         0.32
       Figure                  0.250
       Support                 0.200

Telescope Systems                                        0.65
    Focus                                    0.21
       Encoder                 0.150
       Sensing algorithm       0.150
    Tracking                                 0.36
       RA drift/guide error    0.060
       DEC drift/guide error   0.060
       RA jitter               0.250
       DEC jitter              0.250
    Pointing                                 0.50
       Centering               0.500
       Encoder                 0.150

Facility                                                0.10
    Facility-induced seeing                  0.10

Atmosphere                                              2.16
    Median seeing                            1.90
    Differential refraction                  1.026

3) Assessment of the optical system performance status.

The perfonnance of the telescope optics is central to the success of this project. At this point in time the complete telescope optical system has been delivered but the optical performance of the components has not been assessed. It is vital that the as-built perfonnance of both the primary mirror and the secondary be detennined as soon as possible. The highest priority should be the primary mirror. This is a significant and burdensome task to execute at a any level. I would highly recommend that the AST2 project consider contracting the U of A or a commercial outfit such as Rayleigh Optical (www.rayleighoptical.com) and send their mirrors out to be measured. If that is not possible I propose two tests of increasing complexity.

Anecdotal information from Kent Honeycutt indicates that this mirror was specified to yield 1 arc-second images. I do not know if that is FWHM, EE(50), EE(80) or if they apply to the mirror or mirror plus cell. The top-level goal of any test is to ascertain if the images meet the requirements derived by analysis of the error budget. Optical testing is discussed in detail in Malacara's "Optical Shop Testing"(ISBN 0-471- 01973-9). I have an old edition (1978) and the newer edition is even better. The tests

should be done for the mirror independent of the cell. Without this information, one will never know how well (or poorly) the mirror support system functions. There are two classical ways to test in this way; with the mirror horizontal in a sling mount or vertical setting on a uniformly supporting surface. I would recommend the latter using a thick carpet to best isolate the mirror from the non-uniformity of a concrete floor.

A) Hartmann Test: The theory as well as discussion of the implementation of this test is described in Malacara. A rig will have to be constructed to get access to the mirror radius of curvature. A simple mask can be constructed of aluminum and made in sections for handling. The data can be recorded on a CCD or frame grabber and the spot location found using IRAF (may the heavens forgive me).

B) Interferometry: Undoubtedly, the best way to assess the optical performance of the primary mirror in or out of its cell is with an interferometer. There are several commercial set-ups that can be purchased and this would be the recommended path. However, I have employed the set-up in Figure 1 successfully. It has the virtue of being simply constructed from catalog components. A beamsplitter cube, prism and lens are cemented together as shown in Figure. The outgoing beams (Figure 1 top) are formed from a single spatially filtered and collimated laser beam. The reference beam passes straight through the beamsplitter onto the mirror. It is returned (Fig 1 bottom) by the mirror under tested and goes through the L1 which images it at the location of the filed lens. The assumption here is that the laser beam samples a "prefect" surface; not unreasonable for a few mm patch. The outgoing beam diverted by the beamsplitter is expanded by Ll to overfill the test mirror. The return beam from the test mirror is diverted through the beamsplitter and is co-focused with the reference beam creating the interferogram. The resulting interferograms can be analyzed with any commercial package. I used one from WYCO (now Veeco Instruments in Tucson).

1) Introduction

The AST spectrograph will be one of the most powerful stable spectrographs on a 2-meter class telescope in the US. Indeed, a scientific strength of the AST2 is bringing that capability on-line with a telescope optimized to explore time domain astrophysics. There are well identified and appropriate science programs that drive the spectrograph technical requirements. The overall technical approach of a white pupil spectrograph is also appropriate in this case. Not only does it allow an economical solution to achieving high resolution, but it also provides an excellent platform for eventual high stability observations for precision observations such as in the search for planets or precision line profile analysis. That said, the spectrograph remains the least developed aspect of the AST2 project. Yet is of equal importance as the telescope in overall system performance.

2) Current status of the spectrograph

The design for the high-resolution spectrograph on the AST has been developed to the early concept stage. It is a 200 mm beam diameter echelle system where the effective pupil on the echelle is re-imaged onto the cross disperser grating by two parabolic pupil transfer mirrors. This creates another pupil on the cross disperser and minimizes the size of that element and the camera system. The general performance of an echelle system was modeled by a computer program by Baton. Two alternative designs with slightly different B and y angels were studied. Apparently these were used to develop a strawman mechanical layout mechanical layout that was presented at the review but not in the pre-review package. However there has been no ray-trace of the system. The echelle grating has been procured and delivered. In addition the pupil transfer mirrors have also been ordered.

I have done a simple analysis of the proposed designs using numbers supplied by Eaton. Using the theta=0 and gamma=3.00 layout, the resolving power with a 200 um fiber is 28,000. Using the alternate theta=0 and gamma=3.00 layout, the resolving power with a 200 um fiber is 30,600. The maximum resolving power in this latter configuration with critical sampling is R=DeltaLam/Lam=61000. I do not see how the R=90,000 mode discussed in the document circulated before the review works with the 500 mm camera. The two basic modes should be R=28,000 and R appx 60,000.

The assumption is made that the f/8 fiber fed will result in a f/7 fiber output for the collimator. That is undoubtedly true at some level of throughput. However, I suspect that will be much less than 95%. The actual number depends critically on a) how the fiber is prepared and mounted b) the details of the fiber cabling and c) the length of the fiber cable. As discussed in the previous paragraph, the basic resolution with a 200 um fiber is appx 30,000. The higher resolution is to be achieved by placing a slit (appx100 mm) in front of the fiber. There are several options here and a proper trade study driven by science goals and throughput needs to be done. It is non-trivial to place a slit in form of a fiber. Indeed, this is currently a major current design problem for the instrument I am building for the Hobby-Eberly telescope. Alternatives such as a separate fiber with a permanently attached slit or an optical transfer system should be investigated. The high- resolution mode may well benefit from a fiber slicer. Overall, the fiber feed system is a place to go horribly wrong and loose a lot of light. Thus, it deserves substantially more analysis at the top science mission and systems definition level than it has been given. The total throughput of the spectrograph is likely to have more impact on the system performance than the telescope. As such it also demands careful analysis of the

trades between various performance attributes; resolving power, spectral coverage and transmission. Just as in the case of the telescope where we used an image error budget, the spectrograph must have both an image error budget and a transmission budget. Without such tools rational design decisions are impossible. For example, consider the trade in bandpass and throughput. If the science goals demand one to optimize performance in the blue (<400 nm) a significant price is paid in the red (>450nm). Figure 1 shows illustrates this trade by comparing the surface reflectivity component of a transmission budget for a blue optimized system using aluminum and a red optimized system using a protected silver coating such as Denton FS-99. In this analysis I have used 5 mirrors in the spectrograph, a collimator mirror, 2 pupil transfer mirrors, a fold mirror, and a cross disperser. The differences can approach a factor of two. I also note that if the science drives one to optimize in the blue, the baseline cross disperser is a poor choice since its theoretical blaze efficiency drops below 50% at wavelengths less than 400 nm.

In the material on the spectrograph sent before the review there was a brief description of the spectrograph and pupil mirror mounting. There was a phrase saying something about mounting the pupil and folding mirrors on tombstones. I took this to mean that they would not be on the optical table. This has consequences for the radial velocity stability of the spectrograph that should be thoroughly investigated before implemented.

The greatest challenge for the spectrograph is the camera. The white pupil system described will require a state of the art all transmitting camera. The optical properties of the system will preclude any early use with a "surplus aerial surveillance camera". That is due to the nature of the final focal surface produced by the pupil transfer mirror system that is cylindrical in nature. The way round this is to use a single large spherical pupil transfer mirror. The McDonald 2.7 meter 2-D coude spectrograph uses this approach with a prism cross disperser and I have done this on the Upgraded Fiber Optic Echelle with a grating cross-disperser. There are performance & cost trades here that need to be looked at carefully. In either case a robust camera design is required. This will be neither cheap nor fast but is essential to the overall system performance.