IMACS Commissioning Report to the Magellan SAC Performance and Operation 2+ years after installation Alan Dressler and Tyson Hare, April 8, 2006 1. Capabilities 2. Optical performance: Image quality and throughput 3. Mechanical performance: Devices and Flexure 4. Mosaic CCD Camera performance 5. Operating software performance 6. Telescope interface, including Shack-Hartmann 7. Multi-slit masks, design and production 8. Software observing tools 9. Summary of remaining issues and future work 10. Data reduction pipeline: COSMOS 11. Support Agreement: Tasks, responsibilities, personnel I. IMACS capabilities: 1. IMACS has the largest field of an imaging spectrograph on any 8-m class telescope. IMACS f/2: 670 sq arcmin; IMACS f/4: 236 sq arcmin Comparisons: VIMOS on VLT: 224 sq arcmin DEIMOS on Keck 85 sq arcmin GMOS: 30 sq arcmin I. IMACS capabilities (cont.) 2. IMACS has the largest range of spectral resolutions of any spectrograph on an 8-m class telescope • f/2 grisms: R = 450, 600, 900 (1”, [email protected]: 150-l, 200-l, 300-l) • f/4 gratings: R = 1350, 2650, 5150 (1”, @5000A: 300-l, 600-l, 1200-l R = 3875, 7975 (1”, @7500A: 600-l, 1200-l R = 9375 (1”, @4500A, 1200-l 2nd order blue) • Multi-Object Echelle (MOE) R = 21,000 (0.6” slit, @6000A) • Low-Dispersion Prism (LDP) R = 20 (@8000A) to 150 (@4000A) • Comparisons: VIMOS R = 180, 600, 2500 DEIMOS R ~ 1100, 1650, 2200 (1” slit @6000A) GMOS R = 335 to 2200 (4 gratings available) 1. IMACS capabilities (cont.) 3. Multislit Spectroscopy IMACS “bread and butter” is multislit spectroscopy, which has now been carried out by dozens of observers at both f/2 and f/4. • Powerful mask design software: Ken Clardy has developed an elaborate, well-documented, easy-to-use program for designing slit masks. It has been distributed, and used, at all Consortium Partner institutions. •Typical Multi-slit masks are 100-1000 objects covered by 1” x 8” slits, but many variations have been made by users. • Laser-milling machine in regular use at Las Campanas. Software provided by Ken Clardy, with some modifications made at LCO. Some glitches (cutting 8000 slits on one mask), but the bottom line is that approximately 1000 slit-masks have already been cut, and the operation is almost routine. (What problems persist are bascially machine-related). • Excellent mask alignment software (Mark Phillips). Alignments of ~0.1 arcsec in RA and DEC, ~0.005° obtained in 5-10 minutes Comparisons: DEIMOS & VIMOS, smaller fields, fewer dispersion choices IMACS Cluster Building Survey - Dressler, Oemler, Gladders, Poggianti IMACS Deep Survey -- McCarthy & Collaborators 1. IMACS capabilities (cont.) 4. Integral Field Unit (IFU) -- contributed by Durham University 0.20” sampling, two 1000-fiber bundles cover 5”x7” (f/2) or 4”x6” (f/4) ==> f/2, f4, means that full range of resolutions available IFU observations may be mixed with mutli-slit observations -- IFU requires 3 of the 6 slots in the slitmask server. Comparisons: GMOS -- similar unit supplied by Durham DEIMOS: none VIMOS: 54” x 54” at 0.67” pixel (~6000 fibers) 1. IMACS capabilities (cont.) 5. Multi-Object Echelle MOE consists of an echelle grating and cross-dispersing prism that mounts in one of the positions of the IMACS Disperser Server. MOE provides a 2.5 pixel resolving power of R=21,000 (0.6” slit). Each echelle spectrum is composed of 9 orders, from roughly 3400 Å to 9500 Å. Typically, 7 or 8 objects in the 15’x15’ field can be observed with full spectral coverage. An observation of RGB stars in Carina with the 480nm-780nm blocking filter (4 orders) covered 20 stars, and as many as 100 stars can be done in a single order. Comparisons: ESI on Keck. Very similar specifications, but without multi-object capability MOE spectrum of a red-giant star in the Carina dwarf galaxy, showing the detection of the challenging Eu II line. 1. IMACS capabilities (cont.) 6. Maryland-Magellan Tunable Filter (PI, Sylvain Veilleux, co-I, Ben Weiner) SDL-Queensgate Fabry-Perot Etalon, 150-mm dia will enable narrow band (6-40Å) over fields ranging from 10-arcmin x 27 arcmin (full monochromatic spot diameter) to the full IMACS f/2 field. Wavelength coverage is from 5000-9200A. Frequency-switching, synchronized with Mosaic CCD camera chargeshuffling enable wide-field surveys to unprecedented depths. Ideal for deep redshift surveys (Ly), tracing dark matter with planetary nebulae, studying star formation and gas outflow in merging galaxies (including not only H, [O III], [S II], but also for fainter [O I], [Fe VII), [Fe X] over a wide range of redshifts. Successful testing completed in Pasadena by Veilleux and Weiner in Feb ‘06. Pre-ship review conducted in March (passed). Will be shipped to Magellan soon and possible commissioning run in May. MMTF should be available in semester 07A for general proposals. Comparison: None on an 8-m class telescope. 1. IMACS capabilities (cont.) 7. GISMO - Gladders Image-Slicing Option for IMACS High-multiplexing advantage in a 4’ x 4’ central field of IMACS: Through a series of mirrors in the focal plane, and lenses above the plane, 16 central areas (each 2’ x 0.5’) are fully separated to utilize the entire 8K x 8K detector. 5-8 times multiplexing advantage, enabling for small field spectroscopy (HDF, HUDF, etc.), centers of galaxy clusters, globulars in nearby galaxies, very deep fields… Progess: Optics now being delivered, mechanical design advancing. Available: semester 07B Comparisons: no such device available elsewhere 1. IMACS capabilities (cont.) 8. “Other” Imaging -- little used, especially f/2. Problem with f/2 baffling Centerfield slit-viewing spectroscopy: Enables traditional slit-viewing spectroscopy for brighter objects -- not necessary to use Mosaic CCD Camera for acquisition. Not much used, but Centerfield Guider has proven essential for on-axis Shack-Hartmann testing, and flux calibrations are routinely done with 7” hole in slit-viewing mode. Centerfield acquisition also used for setups, particularly IFU. LPD (Low-Dispersion Prism) -- very low dispersion mode, R~ 30-150 -- built by Scott Burles, commissioned in Jan ‘06. Very high multiplexing (thousands of spectra per exposure) 1. IMACS capabilities -- Summary IMACS is the most versatile spectrograph on an 8-m telescope (on any telescope?). It has a wide range of dispersions, a unrivaled large-field capability for multi-object spectroscopy, and several “accessories” (IFU, MOE, MMTF, GISMO) -- two of which are unique. All of the available modes have been used already, in just 2 years. By the way, IMACS cost ~$6M, compared to ~$10M for DEIMOS and $11M ($91M for 8 instruments) for VIMOS. IMACS receives a fraction of the support that DEIMOS and VIMOS do. 2. Optical Performance Imaging quality: f/4 camera Design was for 0.15” FWHM central rad < 5’ degrading to 0.20” rad ~ 12’ The goal was to achieve ~0.30” images in “perfect” (0.25” FWHM) seeing. Pinhole test in Pasadena: pinholes ~ 0.30” FWHM -- unable to confirm, but “unresolved,” a good sign. Best image taken at telescope by Dave Osip, 0.45”-0.50” over entire f/4 field, 15’ x 15’ This suggests that the image was taken in ~0.45” seeing, with the camera performing within design specifications. Need a better seeing night to push this further. Note: The telescope images also indicate that the collimator performs as expected, that is, it adds negligibly (~0.10”) to the image size. 2. Optical Performance (cont.) Imaging quality: f/2 camera The f/2 camera was designed to have image quality of 0.25” on axis, ~0.30” off axis (rms). It was meant to achieve ~0.4” images in “perfect” seeing. (The sampling of the camera is 0.20”/pixel.) Images of the pinhole mask (whose images are ~0.30”, smallest the laser could cut) show that the spec was met in the central (half-radius) part of the field, but that the images degrade (elongate) to ~0.45” in the very corners. The best image has been taken by Dave Osip, who achieved ~0.50” FWHM in the central part of the field degrading to 0.70” FWHM in the outside. Hartmann-mask tests have been taken by Osip and Phillips. These show a clear tilt of the focal plane that is the largest aberration, specific to the f/2 camera optics. The f/2 camera design has been entered into ZEMAX by Marla Geha and will be used by Alan Uomoto to look for sensitivity of the f/2 optics (decenter, tilt) that could account for the test results. 2. Optical Performance (cont.) A sensitive test, suggested by Shectman, is to measure to take out-of-focus images (doughnuts) Dan Kelson has written some excellent software to do the analysis: The size of the outer ring is a measure of focus, The shape of the outer ring is a measure of astigmatism, and The decentering of the secondary hole measures coma. 2. Optical Performance (cont.) Imaging quality: f/2 camera The problems with the pinhole images were noted in Pasadena, in the shop, but no resources were available to pursue it and there was some hope that proper illumination of IMACS by the telescope would improve the situation. It is hoped that a simple cause can be found and corrected (such as moving one of the lens groups), but the problem may not be reparable through the optics. For example, a wedge in one of the two (strong) aspheres would not be correctable. However, significant improvement might be made just by tilting detector focal plane. The effect of this on performance is bothersome, but not very significant. 2. Optical Performance (cont.) Throughput of IMACS, including telescope (M1,M2,M3,ADC) 2. Optical Performance (cont.) Throughput of IMACS is lower than anticipated, peaks at ~21% for spectroscopy (peak) Expected Actual Perez measured M1 x M2 x M3 x ADC: IMACS optics (20 air-glass) 0.60 0.80 0.62 0.70 IMACS grism, grating, filter 0.75 0.60 Mosaic CCD Camera 0.85 0.84 0.31 0.22 throughput MAGIC IMACS f/2 B V R I 27.35 27.29 27.52 26.83 26.85 26.82 27.12 26.35 -0.77 -0.47 -0.40 -0.47 (0.89 x 0.82 x 0.89 x 0.95) Note: LDSS-3 reaches 30% with similar type grisms! These filters have been scanned by Gabriel Martin direct filter CCD 0.20 0.24 0.22 0.14 0.78 0.80 0.87 0.38 0.59 0.69 0.60 0.85 Assume telescope, CCD same, remove ADC (5% loss) ==> IMACS optics are 0.70 (30%, not 20% loss) Tyson Hare and Dressler are planning a laser test that will measure the IMACS collimator, f/2, and f/4 cameras, but even if something is found, throughput can only rise by 3-4% -- CCD and grisms? 3. Mechanical Performance: Devices • Guiders -- excellent, very few problems • Shutters -- excellent, essentially no problems • Slit-mask server -- good until late in 2004, at which point lots of (sometimes serious) positioning errors occurred. Problem has been traced to optical encoder on the translation stage, and a new one has been ordered. • Filter servers -- poor at first, good for the last year. f/2 filter server has been working without problems for at least the last year. Recent problem with f/4 filter server -- can’t make the uphill push! -- new small air cylinders ordered. Workaround in place, and service planned for May ‘06. • Disperser Server -- First year, some problems latching up. Latches were serviced (synchronized) early-2005, then very good for a while. After grating/grism change tool in use, problems again. Diagnosis: need to reinitialize the unit after installation. Should be okay again. • Grating Tilt-mechanisms -- good, but reports of zero-point drift. Will now be checked at the beginning of each f/4 run (zero-order test), to see if the problem can be isolated. • Dewar Focus (Mosaic Camera) -- working, but not much margin (and probably deteriorating). Rebuild when second Mosaic Camera become available? Need to empirically determine focus = ƒ(temp) 3. Mechanical Performance: Flexure • Flexure from the focal plane to the Mosaic CCD Camera Already small, corrected by open loop control of piezo stage, but bolts added (Osip and Hare) to mount of Mosaic Camera to optical cameras have reduced the flexure. Now: 3 x 5 pixel “ellipse” for f/4, 2 x 3 ellipse for f/2 Note: Even without control these are comparable to size of seeing disk in median seeing for a complete rotation of the instrument -- not bad! New flexure tables for f/4 ready for test; new f/2 tables in process. • Flexure between guiders and focal plane Most data from multi-slit alignments and observations, where fields are followed for hours. General agreement now that alignment of objects to their slits holds at < 0.2” per hour when IMACS moves through modest rotation, but misalignments of several pixels, typically 0.5”, when crossing the meridian. Since there is closed-loop rotation guiding now, this should be zero if the guiders are not moving with respect to the slit mask. We believe that this is true for the Principal Guider, but that the Shack-Hartmann guider, which guides rotation, is too soft (may have loosened up) because of its greater weight and double-stage. Tyson and Dressler will be testing this in May or June of ‘06. 3. Mechanical Performance: Flexure (cont.) • Bolton and Schechter see 0.2” per hour motion of object on IFU as object is tracked for several hours. Bolton’s observation did cross meridian, so big change in gravity vector did occur. This may be related to drift in multi-slit alignment (and SH guider?), but it may not be. The IFU unit is much heavier than a slit-mask (30 lbs instead of 5 lbs) and may itself be shifting -- the total motion shown below is 0.2 mm! Dressler thinks he got performance better than this in the lab, but it has not been remeasured since IMACS installation and, at any rate, was not considered “a requirement” -- thought was that exposures would be centroided and stacked together, and that they would be ~0.5 hour, while the data shown below are one-hour exposures. Find the problem (likely) and fix it if we can (maybe). 4. Performance: Mosaic CCD Camera • Overall assessment: excellent performance, no major malfunctions of electronics (in normal operation) in 3+ years • One CCD fried, and replaced -- cause unknown, but may be related to problems with the Cryotiger cooling system and working close to the cold head when the dewar was open. • Early experience with dynamic range and charge transfer led to fine tuning of the system by Thompson and Burley in May ‘05 (some further work done March ‘06). Nuisance problem of failing to “reset DSP” may be fixed -- we’ll see. • Major damage to electronics caused by glycol leak -- electronics box flooded. Boards replaced. System continues to leak at fittings, despite several fixes, so has been left disconnected (no active cooling). New stainless steel hose and fittings (hard coupling) ready for installation. • Read noise is high on 2 of the 8 chips: some board swapping and voltage adjustments will be attemped. • Swapping the Mosaic CCD Camera between f/2 & f/4 continues to be delicate operation because of Cryotiger cooling system -- swaps have been kept to a minimum. New camera for f/2 in process. 5. Performance: Operating Software (Birk) • Overall assessment: excellent. Very few bugs have been found over the last 3+ years, considering the complexity of IMACS. These have been attended to quickly and efficiently by Birk. Users rave about simplicity of operating IMACS. • New capability for Nod-and-shuffle introduced in 2005. Now in routine operation, working fine. • New software for MMTF has been produced and will be added mid-2006. MMTF will be operated from within the same CamGUI. • Continuing refinements on TO-IMACS-User interface. Plans to add a correction of atmospheric dispersion into the guiding system. 6. IMACS Interface to Telescope: Two major problems that complicated commissioning IMACS 1) No ADC/corrector for first 9 months 2) Difficulty of aligning telescope for wide field -- tilts in the focal plane delivered to IMACS. First problem made it difficult to diagnose second problem, and together they made it very difficult to make and hold good images. Second problem seems to have been (mostly? completely?) removed by controlling the tilt of the secondary mirror as Baade is moved in elevation. (Extensive work by Osip and Perez.) Images at IMACS track images with guiders at Clay very well over the last year. Remaining problems with Shack-Hartmann correction using IMACS offaxis guiders • Different templates are required at different rotations of IMACS -should only need one! • Just by rotating IMACS (changing gravity angle), we can introduce an astigmatism signal into the SH correction! -- the real puzzle! 7. Multislit Masks: Design and Production Clardy has developed a mature program for designing multislit masks. The program has been distributed over the Consortium and upgrades are made available on a regular basis. Thoroughly documented. 7. Multislit Masks: Design and Production -- remaining issues Clardy program also writes the SMF files that drive the laser cutter. and Some refinements may still need to be made in order to facilitate cutting N > 2000 object masks and reducing machine failure (or allow the slit mask cutting to resume at an arbitrary point, simply. In general, the Convergent Prima laser milling machine has performed brilliantly -- quality of slits is very high. Realignment of the optics was needed in mid-2005 -- company sent a technician. New components ordered to allow Magellan personnel to do this. Ongoing problem with maintaining a good air supply. Action? Cutting procedure fails at some point every 4th the 5th mask or so. Needs to be restarted in order to complete. Seems to be due to problem in the way the autofocus feature works. Still trying to diagnose the problem or make a better workaround. Excellent interface made to allow uploading files on the web (Phillips, Osip), and a good bookkeeping system is also on the web. submission of masks: Things are better, but few observers meet the “month-in-advance” requirement. More typical is 2-3 weeks. This remains a major problem for Magellan operation! 8. Software observing tools -- website Very poor preparation for observing tools at beginning of commissioning. The situation has been rectified slowly, thanks to the efforts of Magellan staff, who have put almost all this together. 8. Software observing tools -- “cookbooks” Of particular note: Excellent mask-alignment tool by Phillips -fast, accurate, simple -- many compliments 9. Summary of Remaining Issues and future work Prioritized by Dressler, discussed with Hare, Perez, Phillips • Understand and hopefully fix SH guider operation -- one template! • Second Mosaic CCD Camera -- also more sensitive, we think • Make the autofocus work better -- empirical calibration as ƒ(Temp) • Understand IMACS throughput and increase, if possible to 50-60% (goal: reach 30% or greater including telescope) • Add baffle(s) to allow f/2 imaging • Improve reliability (less maintenance) of Laser Cutter for slitmasks • Diagnose source of f/2 image aberrations and improve, if possible 10. Data reduction pipeline: COSMOS A complete data reduction pipeline for mutli-slit observations has been written, distributed, and documented (not extensive, but usable) by Gus Oemler. Difficulty of use: moderate. Efficiency: excellent. Quality of result: good. Oemler has been responsive to questions and suggestions. Dan Kelson and Greg have taken responsibility to “keep” and improve the software, interfacing with users as problems arise and modifying the code has required. 11. Support Agreement & other requirements Facility Instrument Support Agreement has been written using template provided by Matt Johns. Reviewed and modified by Dressler, Hare, Perez, Phillips, Thompson, Martin, & Osip on April 13. Agreement covers responsibilities, documentation, communication between Magellan and IMACS team, setups, procedures, maintenance, repair, etc. Up-to-date IMACS User Manual from Dressler. Extensive documentation of IMACS systems from Hare. Spare parts are on site. List of continuing issues has been made and prioritized. Action Item for the SAC Report to the Council that IMACS has finished its commissioning period and is now considered a Facilities Instrument, as described in the Facilities Instrumentation Procedure. Some new capabilities are still under development, and there are still some problems, but the Magellan staff and IMACS team commit to continue to work these issues with present level of effort or greater.