Recommendation #1, on CCDs for Megacam
Recommendation #2, on grens for Megacam
Recommendation #3, on the choice of glass for the wide-field corrector
Recommendation #4, on the data pipeline
Recommendation #5, on project schedule and management
Recommendation #6, on queue scheduling
Recommendation # 7, on joint programs and proprietary time
Endorsement #1, on targets of opportunity
Larry Millard joined the Operations Group on April 14 as Site Operations Manager. Larry has many years of experience with Alcoa as team leader and has extensive safety training experience. Unfortunately after working for several months at the summit on a day-crew schedule he has not found it possible to acclimatize to summit conditions and has decided to leave CFHT as of November 14. Larry will be assisting in the recruitment of his replacement. The first of three candidates will be arriving for interviews next week. In his short term with us Larry has gained the respect of all the staff. His skills, in particular related to the Safety Plan, will be missed.
In reviewing the duties of the Site Operations Manager it has become obvious that the implementation of a safety plan by this individual, over and above site management responsibilities, is not likely to work out. We are therefore looking for a consultant to help implement the CFHT safety program, which the Site Operations Manager will then, hopefully, be in a position to maintain. This is likely to be a two year effort.
Site operations has been increasingly involved in major maintenance and upgrade projects in recent years with the result that summit routine and preventive maintenance has suffered. In an effort to alleviate this condition we have promoted Roger Wood and re-assigned him from headquarters to the summit operations.
Robert Anderson joined the staff as head of the new Electronics Group where he is responsible for both the operations and development staff from the former electronics group as well as the detector development staff. This change was implemented to reduce the management-to-staff ratio in electronics and to free up detector engineers to concentrate on detector development rather than working on group management.
SueAnn Healey joined the software group on July 7 filling the position left open with Mark Weber's departure. Apart from learning CFHT software systems, SueAnn has been working on the implementation of the filter wheel control for KIR. (The departure of Mark Weber has created significant discontinuity in the work started on detector software and has affected instrument support for Coude, FTS, and MOS-OSIS.
In evaluating future staffing plans we see major impacts by the MEGACAM project on both the Operations and Software Groups. Accordingly, we have decided not to fill the position in the electronics group left open with William Rambold's departure, but to use this position to bolster manpower in software. The position will be advertised in the next month. William's departure affected ongoing projects such as the 'Gumball' upgrades and OASIS support and significantly increased our concerns for staff familiarity with a wide range of important instruments.
To support the Operations Group we have identified external consultants to assist in the design and finite element calculations needed for the rebuild of the Prime Focus Upper End.
With the anticipated completion of the project phase of the TCS-IV Project at the end of April 1998 the TCS-IV Group will be dissolved and William Cruise will be attached to the Electronics Group where he will maintain overall responsibility for TCS operations, and will be in a position to assist with major electronics projects. At the same time Peregrine McGehee will return to the Software Group where he will provide TCS support as well as work on software projects.
Persistent problems with shutter motor alignment and motor mount damage resulted in the mechanical failure of 2 bolts and a bracket on a shutter motor mount in early July. The bolts and bracket fell from the shutter to the observing floor. Inspection of motor mounts and associated bolts showed that the motor mount securing hardware is extremely crude. Bolt holes for example were oversized and flame-cut. Maintaining motor alignment and continual brake adjustments have required weekly routine intervention by the summit crew for the past year. Binding of shutter guide rollers, which then transmit undesirable high loads to the drive motors, was discovered in early summer. All accessible rollers have been serviced. Maintenance on the remaining rollers is scheduled.
A project to replace the dome shutter motors and controls was initiated in early 1994 following a failure of the shutter due to a jammed eddy-current brake. During assessment of the problem at that time we decided to replace both the 8 shutter motor / brake assemblies and their related controls with more reliable systems. Important related issues where ongoing problems with the eddy-current brakes and the fact that motor electrical controls were located on the moving shutter which prevented access for trouble shooting should the shutter controls have failed while the shutter was open.
These upgrades, started in 1994, but which were mostly completed this semester, required :
- replacement of existing cord reels with motorized cord reels and new cables
- installation of Allen-Bradley motor controllers (1 per motor) in a new enclosure located off the shutter on the inside skin of the dome and accessible from the dome catwalk.
- installation of motor control interfaces using an Allen-Bradley SLC 500 series programmable controller
- fabrication and installation of new motor mounts on the shutter to accommodate new motors and to improve maintenance access
- replacement of wound-rotor motors with Allen Bradley induction motors with shaft mounted disk brakes. (One motor is supplied with an integral encoder to provide shutter position information.)
The relocation of motor controls combined with the automated system monitoring provided by the new equipment will dramatically reduce the time required to troubleshoot and repair the system in the future in the event of a failure.
The improved cord reels where installed last year as was the initial physical installation of the upgraded control panels. Upgrades completed this semester included the fabrication of new motor mounts, the installation of 8 new motors and the implementation of PLC-based motor controls. Training of summit staff on the use of PLC-based controllers was undertaken in support of this and other related projects. The design and fabrication of the new motor controls and motor mounts was undertaken by CFHT, while the physical installation of the motors and motor mounts on the shutter was contracted to Coast Steel Construction.
Installation of the new systems started on September 5 and was completed successfully on September 26. Our initial plan was to remove 3 old motors and to install 4 new motors (the shutter has been operating on 7 motors for a number of years) over several days of regular telescope operations, to switch control circuitry during a 2 day/1 night scheduled telescope shutdown, and to then remove/install the remaining 4 motors again over several days of regular telescope use. The ability to operate the shutter on 4 rather than 8 motors during the change of controllers was checked in advance.
The loss of 2 + scheduled observing nights during this period resulted from the failure of one of the old motors while the dome was in a half-open condition. Access to the failed motor and to the motor controls at this point was extremely difficult since normally all maintenance access to the shutter motors takes place at the top of the dome with the shutter nearly fully open. Incorrect wiring documentation for sections of the old motor control circuitry which we had not planned on using, but which was required to regain motor control didn't help. Much of the lost observing time was caused by water contamination of the telescope hydraulic oil which resulted from heavy condensation inside the dome while the shutter was open.
CFHT and Gemini have agreed on joint access to the weather monitoring data from the CFHT weather tower. Because of severe weather and the effects of repeated lightning strikes most of the monitoring equipment on the CFHT tower has reached the end of its service life. The agreement with Gemini provides that CFHT will select and Gemini will purchase new weather tower sensors which CFHT will install and maintain. Access to the data by Gemini through existing summit network connections will be worked out jointly between CFHT and Gemini. The upgraded weather tower equipment should be in place on the tower no later than the summer of 1998.
The CFHT dome needs to be repainted. It was last painted in 1990. We are investigating
the use of Lo-Mit aluminum-based paint for the dome structure. This paint is
used on the Gemini dome. The CFHT dome is currently coated with a TiO2-based
white paint. Lo-mit paint, as the name implies, has a reduced emissivity and
lower absorption than TiO2 paint at 10 um and was selected by Gemini to optimize
their dome's thermal environment. Visually, if we change paints, the CFHT dome
will be a dull silver color as opposed to its current bright white. To take
advantage of potential cost saving by painting near the same time as Gemini,
painting of the CFHT dome could start as early as the end of this calendar year
provided the aesthetic changes to our dome are not deemed objectionable.
The use of larger than normal aluminum clips on the coating chamber's tungsten filaments required test firing and refitting of the chamber. Firing was successful with no aluminum drops evident. The next scheduled use of the chamber is for the UKIRT primary mirror in May 1998 and recoating of the CFHT primary mirror in August 1998.
CFHT will be hosting a 2 day symposium on Reflective Coatings for Large
Telescope Optics in mid-November. One day of meetings will be held at a
local hotel followed by a day at the summit touring coating facilities
at CFHT and several of the new telescopes.
Three short engineering shutdowns in April, May and June were used to complete the maintenance on the remaining "bird's head" locking mechanisms which lock the telescope's three upper ends to the telescope top structure. During this period the geometries for all "bird's head" lock segments and the locking mechanisms on the upper end handler were measured and then adjusted for correct operation. This work, which successfully solved problems of lock mechanism binding, has resolved mechanical clearance problems experienced over the past many years during upper-end seating operations. It has taken over 18 months and a significant amount of work to repair these problems since the failures which led to the f/35 accidents at the end of 1995.
Water contaminated the telescope hydraulic oil during severe winter weather early in the semester and again during the efforts to replace dome shutter motors in September. A canvas awning and metal gutters were designed and installed over the telescope's south journal bearing oil drip pan to prevent water which leaks through the top of the dome from entering the oil system. Similar measures are under consideration for the large open return-oil tray under the north horseshoe hydrostatic bearing. The solution to the real problem - water entering the top structure of the dome during bad weather (even when it is closed) - is a high priority but has proven in the past to be difficult to solve.
Investigation of oscillations in declination which were dramatically recorded in the tracks of geostationary satellites on UH8K camera images in semester 1997I led to modifications of TCS III data logging programs and to the installation of a PC-based data logger to provide long-term telescope position records at a rate of 10 Hz. Despite a considerable effort to catch observe the large declination 'glitches' seen on these images the problem has not been recorded again.
Two other sources of telescope instability have been clearly identified. The first is an oscillation with a 10 to 20 second period, mostly in declination, of approximately 0.3 arcsecond peak to peak amplitude. This motion is the likely cause of previous complaints of elongated images. Although we have not identified a permanent solution, whenever the problem occurs the Observing Assistants have been asked to issue a TCS command to reduce declination servo loop gain to eliminate the oscillation.
The second problem is a sporadic declination offset of up to 0.8 arcseconds amplitude lasting up to several seconds of time. These offsets occur several times a night. The problem appears as a sudden change in the declination incremental encoder counters and may be associated with noise from the declination incremental encoder. We have switched to a second incremental encoder but have not had the opportunity to determine whether the change has isolated the problem.
The AO Bonnette continues to operate efficiently without major problems. The Fabry- Perot etalon used with the AOB-FP mode has been repaired by Queensgate and successfully lab tested in the AOB.
The new V, R, and I band beamsplitters, intended for use with OASIS, are still being fabricated. We are hoping for delivery in November.
Continuing efforts to eliminate increasingly smaller light leaks in the MOS and OSIS spectrographs has resulted in the additions of bellows between the camera boxes and the grism boxes, replacement of missing screws from various locations, addition of opaque tubing on the nitrogen purge fittings, and addition of a seal between the grism boxes and the Octagon. It is now possible on both MOS and OSIS with the STIS 2 ccd to take 10 minute dark exposures with dome lights on without recording any stray light.
Scattered light seen on long direct imaging exposures has been greatly reduced by the installation of masks at the internal pupil image. Similar masks are being fitted fort each of the grisms.
A problem resulting from swapped guider APD signal cables was discovered in May during lab tests. Once corrected, the system was re-tested on the mezzanine and during engineering time on the sky and worked well. Observers however continue to report mixed success when fast guiding is connected to the TCS at large Cassegrain Bonnette rotation angles. Further lab and sky tests are schedule in order to put this frustrating problem to rest. However, decoupling fast tip/tilt guiding from TCS low-frequency guiding by using the Cassegrain Bonnette autoguider allows continuous observation with OSIS fast guiding for at least an hour. For the last OSIS-V runs the observers were able to maintain their objects within 0.5 arcsecond wide multi-slits for many hours.
Clarification of several OSIS setup forms in the Pegasus session have made the use of OSIS fast guiding more efficient. One particularly vexing problem associated with changes to the OSIS firmware control code which turned guider high voltage supplies 'off' when they were commanded 'on' was solved.
Telescope offset commands required for precise centering of objects in mask apertures continues to work well at all bonnette rotation angles.
Several procedural problems associated with software preparation of OSIS runs which led to setup delays and missing control features are being addressed by the software group.
OASIS arrived disassembled but otherwise intact and in working order at the end of May, 1997. Integration of the 2 imaging and 4 Tiger modes started with the arrival of the Lyon staff in early June and lasted until mid-month. The remainder of the optics and the ccd detector were aligned without major problems.
The entire first commissioning run, June 23 - June 26, with OASIS mounted on the AO Bonnette, was lost to poor weather. Time was well spent during this period however on tests of image quality, flexure, and light leaks.
Optically OASIS performed as predicted. Significant light leaks are evident around the lower part of the camera assembly however.
Flexure tests using the AOB's artificial stars show image motions to be less than 0.07 arc- seconds per hour in imaging mode and less than 0.6 micro-lenses pupil diameters per hour and less than 0.3 pixels per hour in the Tiger modes.
Preparations for the second engineering run in August showed that OASIS is optically and mechanically stable. The engineering observing run from August 7 to 1, 1997 was very successful. OASIS was tested on the sky in all of its TIGER modes (3 mm, 8.5 mm, 12.5 mm, and 22 mm enlargers), but not with all possible combinations of resolutions and wavelength regions. It was also tested in the 62 mm imaging mode. The results of this engineering run are given in the "OASIS Test Report" by Gilles Adam.
The following points still need to be addressed :
- The Gumball spectral calibration source is much too faint for use with many spectrographic modes in OASIS. Flat field exposures, which are needed many times a night, last anywhere from one minute to an hour depending on the mode.
- Observer interface software has yet to be delivered and tested. The engineering software delivered with OASIS has performed well.
- The OASIS camera focus occasionally shifts by approximately 1 mm for no apparent cause.
- The mechanical fit of a large-format CCD dewar needs to be tested.
- Some controlled devices intermittently failed to complete their actions properly
- Light-leaks need to be located and sealed
- Data reduction software for Tiger mode has not yet been delivered
Apart from the fabrication of masks for MOS and OSIS, LAMA was used to fabricate pupil image masks for the microlens array used in OASIS and pupil masks now mounted in the KIR camera.
The Gumball upgrade project is proceeding well. The optical design has been finalized and the optics have been received. The Fabry Perot etalons for use in conjunction with continuum lamps to provide improved spectral coverage were received in early September. Mechanical and electrical control upgrades are underway. We plan to have upgraded calibration optics in place and operational in MOS/OSIS and the AOB by the November MOS run.
The new Loral 5 CCD was in use on Gecko for the October observing runs. Tests indicate that Loral 5's QE is better than Loral 3 by a factor of 2 to 3 times for most wavelengths and that Loral 5's QE appears to be stable. The onset of fringing with this detector is seen at ~6800 with the fringing reaching a plateau of about 26% (peak-to- peak) at ~8500 . (See further details under 4.2 Detectors - Loral 5)
Initial indications are that image quality (slit image size) is degraded with Loral 5 compared to Loral 3 by ~20%, most probably as a result of the MTF degradation expected with thinned CCDs. We have not seen a clear wavelength dependance to the degradation. Further characterization is needed.
An HTML file data-base is now available for entry/recovery of past spectrograph configurations
Dial indicators have been added to the mosaic grating mount to measure grating roll.
The FTS BEAR-mode system software has become much more reliable - observers now report that they have confidence that a long scan will be completed without the system stalling. (See below the significant improvements related to Redeye Wide upgrades)
The new IR2 beamsplitters have been received and accepted. They have not yet been put into service.
CFHT's current operational detector are :
We have on loan or use on an ongoing basis the following non-CFHT detectors.
Severe readout failures characterized by pickup noise, many hot pixels, and finally a general failure of the first bank of ccds occurred during the entire May UH8k run. The problem was not solved during the run. The following UH8k run in June started with the same problems. After a full night of troubleshooting in the prime focus cage, CFHT staff found a slightly unplugged power supply board on the ccd controller which was resulting in flaky voltages. The problem has not reappeared since.
CFHT added a temperature probe in the ccd electronics rack connected to Luppino's camera temperature controller and ran cables so that the focal plane and electronics temperatures can now be monitored in real-time from an RS-232 connection.
Shutter failures between May 5 and 9 run resulted in lost observing time. The drive shaft of the shutter cocking motor bent in operation. It has not failed since the shaft was subsequently straightened. To protect future observing runs CFHT has purchased spare shutter motors.
One night was lost in June to spurious noise problem which was addressed by re-seating several connectors in the camera control electronics box.
Failure of the fiber optics interface electronics resulted in lost observing time during Luppino's run in September. A spare communications board was borrowed from the IfA after which no further problems were reported. The CFHT spare was not initially delivered but will be in the next weeks.
To improve camera reliability CFHT staff and Gerry Luppino looked into the possibility of mounting the shutter mechanism for the CFH12k camera on the UH8k camera for future runs. This option turned out not to be physically possible.
To provide better protection for observers in the case of technical problems and to familiarize us with shutter operations, likely failure modes and solutions which have worked in the past, Luppino and CFHT technical staff spent a day in May working together on the camera at the summit. The occasion was used to repair and upgrade several shutter-related mechanisms. There have been no problems with shutter operation since.
Vignetting of the camera by the guide probe is no longer a problem. The guide star selection program however appears not to work reliably. The source of the problem is being investigated.
The camera control software status / command window now makes control available through either a command line interface or with Pegasus-style commands.
Shortly after solving initial dewar leak problems it was evident that this ccd would not be appropriate for general observatory use since the top half of the ccd suffers from severe image smearing in the columns when exposure levels above 65k e-. The rest of the device has a full well of 80k e-.
Loral 5 is the first CFHT ccd for which dual amplifier readout has been implemented. Both the lower left and lower right amplifiers have been used. The preamplifier circuitry will handle simultaneous dual channel readout. This feature is not currently available since we restrict use to a rather small subarray of 400 x 2k pixels near the bottom edge of the device.
Loral 5 is a platinum flash gate device which can be desensitized through exposure to hydrogen ions (water vapor). Although the device does not require UV flooding we have found that exposure to ambient pressure O2 at 85 C for an hour followed by vacuum pumping appears to leave the detector in a stable, high QE state. Changes in QE during lab tests distributed over several weeks have been under 5%
Loral 5's characteristics are listed in the table below.
Characteristics of Loral 5
Readout times including system overhead are now 2.8 seconds in the BEAR mode compared to 4.5 seconds seen previously. The readout clocking sequence can now be readily optimized for the differing needs of OSIS and BEAR - minimum read noise and maximum sensitivity for OSIS, and minimum read times in the case of BEAR.
Interferograms generated in FTS Bear-mode now have substantially lower long term drift and higher signal-to-noise than seen before as a result of the new preamp mounted on Redeye, which is giving much more stable output signal levels than the previous system.
A problems that plagued previous FTS BEAR-mode runs was a randomly variable level of overhead or dead-time between steps in a scan. This injected a large noise component due to variable thermal background coming from the interface optics and the variable inter-step latency. Because of changes in the Redeye control software BEAR- mode now has an extremely stable dead-time between scan steps which has eliminated this important noise source.
The software to implement double sampling per pixel to reduce read-noise by factor 1.4 has been demonstrated to work in the lab and is available at the telescope on an as- requested basis. The cost to the observer is an additional 0.2 second in readout time.
Pickup noise at the telescope remains a significant problem. The use of an insulating standoff to reduce pickup on OSIS in itself did not help but is a first step in isolating the camera from this noisy environment.
During the OSIS observing run, image motion attributed to flexure inside the Redeye cryostat, amounting to 2 pixels in 4 hours, were seen. We have no immediate plans for addressing this problem..
Delivery of the EEV science device is now expected in January 1998, but we expect to have the engineering-grade device in UBC's hands by December.
DetI is a new general-purpose detector control software system developed at CFHT for engineering-level interactions with detector systems. The front end has been rewritten to take advantage of a new interface tool called 'Director' developed in the software group. Director provides enhanced acquisition system flexibility both at the user level and for the underlying architecture. The core functionalities of DetI still remain.
Director is a shell around DetI. DetI is now a single agent running on a independent machine which can now be fully dedicated to detector data acquisition. This has to be compared with DetI's previous structure which used distributed interface handlers and servers.
Features of DetI :
We have designed a Class 1000 clean room to be installed in Room 71 of the Waimea headquarters building. The facility is intended for use with all CFHT detectors but the principle impetus is to provide lab space for work on the CFH12 k camera and MEGACAM.
Room 71 is currently the mechanical design office. An initial proposal to use Rm 70 for the clean facility did not work out because of the limited space available and the cost of building modifications required to provide improved access. Rm 70 will be converted to a mechanical design office instead. We should be able to order basic material for the clean room in 1997 with installation taking place in early 1998.
The DS3 communication link which utilizes the fiber optic connections installed in semester 97I by GTE between Waimea and the Summit is now in use. All equipment were selected, purchased and installed between May and October. After several days of tests, the connection was put into service on October 16.
Several communication and performance tests were conducted in early October. These tests indicate performance levels in both directions within 5% of the nominal 45 Mb/second rate specified for a DS3 link.
The CISCO routers used to support the DS3 connection were selected by CFHT staff because of their flexibility. The router provides the option of imposing 4 separate T1 channels beside its basic ATM connection. A pair of Fast Ethernet connections may be needed to fully support connections to the net.
The OASIS software control and TCL/Tk-based engineering user's interface were incorporated in a regular Pegasus session before the first OASIS observing run. All the engineering modes were tested on the sky during the August observing run. General user modes will be installed and tested during the next observing run.
Writing data to the disk in the background while CCDs are being read has saved roughly 30 seconds per full mosaic readout.
Archiving of UH8k data is now automatic and is handled in the same manner as all other detectors at CFHT.
A shutter control time delay has been added to the software to allow for the shutter rewind time. (Undocumented shutter rewind delays led to lost observing time in the past.)
Intermittent readout failures caused by software running on the Unix machine have been fixed.
In preparation for the CFHT 12k an adaptation of the NOAO message bus system has been implemented and tested. This software will allow us to take advantage of tools already developed by the IRAF group, and uses the FITS multiple extensions image format.
Director is a wrapper for generic command line interfaces. Developed at CFHT using standard libraries from the Free Software Foundation(GNU), Director provides a safe user environment and allows for text status display, easy scripting and hooks for communication with graphical user interfaces. Director currently is used with UH8k and KIR control software, will likely be used with MEGACAM, and has the potential for use with other CFHT detector systems.
O/S upgrades of the summit HP computers continue as time permits.
In July one of our Sun servers, which hosted the CFHT anonymous FTP site, became
unusable although it remained booted up. Initial analysis suggested a Denial
of Service attack in which someone continuously floods a service, such as ftp,
until the attacked machine can no longer handle the load. We stopped anonymous
FTP service at that point and made a more in depth analysis of the situation.
The result showed that pirated PC software was being stored from outside CFHT
on the CFHT anonymous FTP site and then advertised. Apparently, so many people
rushed to download this software that it had the effect of a Denial of Service
attack. The anonymous FTP site had been in service for many years with no problems.
However, this incident demonstrated that the configuration needed to be changed
to reflect modern security concerns. The primary change which has been implemented
is that users outside of CFHT can no longer retrieve files they have stored
on our FTP site. Retrieval of files is now performed in a different area in
which outside users have no privileges except retrieval.
TCS IV has supported its first full observing nights on a trial / test basis on AOB with MONICA, and on OSIS. The nights with AOB and MONICA were very successful. Only a small amount of dedicated engineering was necessary, after which observations were made with about 90% of the efficiency of TCS III. The main stumbling blocks were lack of star catalogs and lack of an observing log and some limitations of the user interface. The work with OSIS went well except for guiding. TCS IV was not able to guide on the slow guiding corrections sent by the OSIS computer. This has been an ongoing problem with TCS III, and the problem may not be on the TCS side. Further investigations are planned.
TCS IV heavily embraces the concept of virtual telescopes. We currently support a main beam virtual telescope and a guide beam virtual telescope. Use of this system, where the beams are treated as separate telescopes from the input, or user, side has been very successful. It provides us with several advantages, such as automated guider pointing, coordinated main beam and guider offsets, and the ability to have the main beam and guider use objects with different coordinate specifications. This last capability resulted in the easy implementation of non-sidereal object observations.
Command and status links with data acquisition and visitor instruments were implemented. The data acquisition link uses EPICS channel access with direct network connection to the TCS IV EPICS Input Output Controller. With a new tcsh front end built by the software group TCS IV provides the same services and interface to Pegasus applications as the present TCS III link. The visitor link uses RS-232 and provides a limited subset of the TCS III visitor link functionality. These links are critical to support of observing with visitor instrumentation.
A telescope pointing problem was injected into TCS IV during routine code updates. This happened after initial successful pointing testing in 1996. Unfortunately, due to the error in the 1997 Astronautical Almanac, bad weather and lost testing time, it was not found for several months. Once the problem was identified it required two weeks to uncover the source of the problem and to correct it. During this work a new technique was developed, using original TCS III pointing test data, to fully verify system pointing on the simulator.
Much of the project work for the last semester has been devoted to user interface development. The use of TCS IV in various observing scenarios has been studied by the group and discussed with observers and observing assistants. A paper has been written to guide the user interface development. Many of the necessary user interface forms have been tested and improved.
A plan to use commercial databases for the system's object catalogs and observing log suffered a setback when the purchased product presented insurmountable problems on the HP 9000 platform. A search has been made for suitable products which are available within the project's budget. We have been considering sybase and Informix. We are also evaluating a freeware product, postgreSQL, which may end up in the final system. As a result of this setback the currently operating system does not yet have a star catalog system. It has also delayed the start of work on the observing log.
A final, high level decision was made to not improve the prime focus bonnette control system. This resulted in a TCS IV project decision to support the PF bonnette through its present CAMAC interface. As CAMAC had been removed from the TCS IV project during earlier design stages, it was necessary to reintroduce CAMAC into the design, and to reconsider the transition phase of the project. As a result CAMAC hardware has been purchased, and EPICS CAMAC software installed. The dome control will be done through CAMAC during the transition phase, which will simplify that switch over. TV status displays will be implemented using the same CAMAC TV display generators as used by TCS III. This will remove the requirement for a VME TV display module and software.
Due to the unexpected delays with the KIR project and associated work load at the University of Montreal, there have been no major developments in the OSIS-IR project since May. Some progress has been made in Montreal with the optical design. CFHT has issued draft contracts documents to the University of Montreal. The project is awaiting detailed costing and negotiations on the technical specifications before the final contract can be developed.
KIR has recently completed a successful 4 night engineering run (Sept 21-23). KIR has not yet passed acceptance tests since issue of filter wheel control, internal flexure, and spurious signal pickup remain. Nonetheless, the results from this first engineering run indicate great promise for KIR which we fully expect to be available for science starting in December, 1997.
Vacuum integrity of the cryovessel is excellent. Hold time for the outer LN2 chamber in an upward-looking configuration is in excess of 20 hours. This should correspond to a total system hold time of at least 24 hours since this is a two vessel cryostat.
The cryostat still requires a thorough disassembly, cleaning, and baking before it is release for general use. We will review possible benefits of gold-plating the outer radiation shield as well as blackening of additional internal surfaces.
Internal electrical wiring at the filter wheel assembly and the detector fanout board need to be reworked. A number of improvements are needed in the filter wheel area to improve reliability, decrease out-gassing, and to improve electrical performance.
Noise pickup in the detector acquisition system due to wiring from the preamplifier to the focal plane fanout board will be addressed by reducing the size of long service loops and reducing multiple ground loops. An attractive solution utilizing flexible flat cable similar to that to be used on the CFH12K is being investigated.
Flexure of the internal dewar mechanics significantly exceeds the specification of 0.5 pixels per 15 degrees. The University of Montreal has identified a promising simple solution which should be in place for the December observing run.
The filter wheel drive problems have been an ongoing concern for sometime and will be addressed before the December run. The filter wheel unit is currently at the University of Montreal for modifications to upgrade the stepper motor. Difficulties with motor control and an associated optical encoder which encodes the wheel position between discrete filter positions has delayed implementation of high-level software control of the wheel, and full-integration of KIR into the CFHT environment.
Optical performance is quite good with diffraction rings clearly evident in J,H and K. Image Strehl ratios are currently being calculated. The dewar window is scratched and will be replaced.
Detector and electronics
KIR used the University of Montreal engineering array for the October observing run. The CFHT science-grade device will be in place for the December run. Current detector performance shows a gain of 3 to 3.6 e-/ADU, 1% or better linearity, full well better than 100ke- and read noise of 25 to 35e-. Read noise at the telescope is dominated by interference pickup which appears to be coupling to the preamplifier input through internal dewar wiring. System noise (clocking included) with the focal plane array output replaced by an equivalent resistive load drops to under 5e-. The detector read noise specified by the manufacturer is 10e-. With improved internal wiring and grounding we should achieve detector-limited performance.
The current KIR pixel rate is 13us/pixel. This is at the limit of the SDSU v.1 architecture. The HAWAII FPA has a 1us/pixel maximum readout rate. We hope to upgrade the data acquisition electronics in the future to match the FPA performance. The current KIR preamplifier has been designed to provide detector- limited performance at rates exceeding this pixel rate.
The new VME memory boards with an on-board ALU have behaved flawlessly and provide very efficient image subtraction required for IR arrays. The memory boards will also provide image co-addition in the December run.
Additional work needs to be completed for filter wheel control electronics, and to implement software-accessible temperature measurements at the radiation shield, the filter wheel, and 'getter' surfaces. A KIR shutter driver needs to be fabricated to replace the one borrowed from MOS/SIS spares, and shutter state sensing hall-effect switches need to be integrated into the system. Since KIR is an infrared camera and sees "heat", tests remain to determine if the shutter solenoid is a heat source which affects the detector.
The detector acquisition software (G3V4) and user interface (DetI) have worked well. The system is easy to use, flexible, and robust. The interface to AOB is functional and is maturing daily. The detector DSP code has been reliable and is powerful, especially with the added features for several readout modes and configurations. The combined acquisition software system (DSP code, G3V4, and DetI) provides many tools that have been previously missing, and adds power, flexibility and efficiency to our image acquisition process. Significant work remains to be done to provide a fully integrated environment with the AOB and the TCS. Tools for focus and alignment as well as observing support are not yet implemented. This has slowed the observing process. These tools are currently under development. We also lack a multiple sample readout capability for the detector. This is a common infrared detector noise reduction technique which is expected to be available this fall, along with high level filter wheel control. Many of these features will be implemented before the December observing run.
Most project efforts since April have centered on the exploration of options and initial design studies. The principle issues at this point are :
Many of these issues are inter-related. Firm decisions for some may have to wait until staff at DAO have the opportunity to start conceptual designs. Several issues raised at the last SAC meeting have been addressed and are covered below. Since May CEA has concentrated its efforts on concept design issues related to the camera head, filter Juke Box, shutter mechanical layouts and computer hardware array control. DAO has been contracted for the optical design and fabrication quotations for the wide field corrector, and for mechanical studies of optical element deformations. CFHT has concentrated its efforts on the definition of software / hardware interfaces to the camera system and on optical design studies in support of DAO's efforts. OPM has continued to explore the tip/tilt plate and driven-lens options for image stabilization.
The project has suffered from the lack of a full-time project manager. Formal technical specifications are just being started. Many specifications needed to finalize formal contracts await the results of preliminary studies. We plan to have a project meeting early in the New Year to discuss these and related issues.
Two optical designs have been established - one based on Ohara BSL7Y glass, and a second using Schott BK-7 glass. The designs are essentially identical.
The glass homogeneity specification for the corrector elements - mostly for the first element - is an important issue since the homogeneity grade required is likely to be better than standard Grade A glass and may therefore seriously drive glass cost, availability and delivery. This is not an easy specification to establish. Studies are currently underway.
Optical transmissions for both correctors and a variety of antireflection coatings are shown in Figures 2 - 5 of Appendix 4. These curves include the effects of the 4 corrector lenses, a tip/tilt plate, a glass blank in place of a filter, and the camera cryostat window. The most realistic curves are those labeled 'Mixed' which represents system transmission using MgF2 AR coatings on all optical surface except for the 6 internal surfaces of the corrector which have dual layer MgF2 plus sol-gel coatings.
Grens and grism designs
Grism (grating on prism) and grens (grating on lens) designs for field spectroscopy have been explored at dispersions of 20 A/pixel and 40 A/pixel and are reported in Appendix 5 The use of a grism in the filter wheel, which is operationally attractive since it offers the potential for stable corrector opto-mechanics, appears not to be viable since image quality suffers badly and grism wedge angles are sufficiently high that grism edge thicknesses (up to 45 mm) greatly exceed the space available in the filter assembly.
The use of a grens on the other hand, with a grating located on the last surface of a wedged lens - the last lens of the corrector - appears more promising. Design spectral image quality is 0.5 arcseconds or better. However zero-order images needed for guiding are on the order of 2 arcseconds in diameter. Precise guiding with such images may be problematic. We are investigating options for a single element 'corrector' which could be introduced into the guiding optics to significantly improve zero-order image quality. Note that using an undersized prism and grating on the last element of the corrector to provide an undispersed guide star does not appear to be an option since the zero-field beam diameter at the grens (~300 mm from the focal plane) is large.
A further complication for use of a grens is that gratings of the requisite size (roughly 480 mm in diameter) are not available, but could potentially be generated at considerable cost by the Richardson Grating Lab or other facilities capable of generating custom master gratings.
A grens will require specialized handling equipment for semi-regular swapping between the last corrector element and the grens with the attendant risk to both optical elements. Our suggestion at this point is to provide for an exchange of the last corrector elements in the WFC mechanical design but to otherwise not pursue this option further. If there is a time in the future when stability of the corrector optics becomes less of a concern to the science community, then a grens option might be pursued without extensive disruption of the system.
The dimensions of the prime focus environment have had to be made slightly larger than the outline of the current PF cage in order to accommodate the camera and filter Juke Box assemblies. No attention has been paid in these figures to other environment requirements such as connector locations and cable routing.
As can be seen in these diagrams CEA now has a design for a Juke-Box filter exchange mechanism which accommodates up to 8 15 mm thick filters. Since the original rotating half-disk shutter design is still proposed, filter exchanges will only be permitted while the camera shutter is closed. Otherwise mechanisms will collide.
There is an ongoing conflict between the space needed at the front of the camera and the available field for the guide probes without vignetting the ccd array. If the distance from the ccd array to the front of the camera (really to the top of the guide probe pick off mirror) can be kept below roughly 100 mm, then the guide probes have a sufficiently large vignetting-free field to have a reasonable change of finding pair of guide stars. If this distance grows to 120 mm the ability to find a pair of guide stars starts to be compromised, especially with 15 um ccds. (See Figure 1 of Appendix 7 for details)
CEA currently cannot fit the dewar window, the filter, their closed-shutter calibration light source, and the shutter inside a space of 100 mm from the focal plane. The solution they are currently investigating is to mount a reflective diffuser in one 'filter' cell in the Juke Box and to use the diffuser together with an external filtered light source for day-time pseudo-flat field detector calibrations. If this arrangement will work CEA has a design which fits the remaining equipment within 105 mm of the camera focal plane.
SAC, at its last meeting, asked CFHT to explore options for more complete focal plane tiling by mounting additional ccds on the edges of the rectangular array originally proposed. Although they have not yet embraced the concept, CEA has attempted to incorporate these additional detectors in the dewar design. There appears to be no fundamental problem with adding these ccds but they clearly have an impact on system cost and size, and particularly on the cooling system heat loads. It remains to be seen whether optical filters large enough to accommodate the longer ccd array will be available. A drawing of the larger array is shown in Figure 2 of Appendix 7
We have had very successful discussions with CEA on the software and computer hardware interface of Megacam to the CFHT acquisition environment. The proposed interfaces provide a high performance connection to allow for efficient handling of the Megacam image data and should also provide for an efficient acquisition environment and quick-look processing capabilities.
The computer connections between Megacam and the CFHT acquisition system are shown in Appendix 8. Megacam will have either 1, 2 or 4 channels of data being sent from the camera. The exact number will depend upon the performance of internal acquisition systems with the worst case being a channel for each 'bank' of CCDs. Each channel will have its own PCI/Fibre Channel (FC) connection to the CFHT summit computer environment. Fibre Channel is a high performance technology capable of transmitting data at 100 MB/sec. Each FC from the camera will connect to an FC switch to which a high performance Sun HPC 450 (4 processors and 2 GB of memory) will also be connected. The Sun FC interface is also PCI-based. This architecture should have no problem in handling the expected data rate from Megacam of 30 MB/sec.
Megacam will also have a standard Ethernet connection to the CFHT network. This connection will be used for command and status information and will be independent of the data connection described above.
Megacam will use the same software architecture for sending data as CFHT is using for the CFH12K camera. For CFH12K we are using the NOAO Mosaic software developed by the IRAF group for use with their mosaic camera. (This software is likely to be used by other telescopes with CCD mosaics). Data is sent from the camera using a software message bus and a DCA (Data Capture Agent) on the receiving computer which captures this data and writes it to disk. The data is sent sequentially as it is readout so that the data is written to disk very shortly after the camera has finished reading (assuming the disks can handle the data rate). At the same time the data is also captured by an RTD (Real Time Display) which displays the Mosaic data as it received. The RTD will be able to do some simple on-the-fly processing of the data (e.g. subtract a different bias level for each readout) to make the display more useful. The display will consist of at least two 'screens', one which is a view of the whole mosaic and one which shows a zoomed portion of the mosaic.
This RTD will be integrated with IRAF and CFHT plans on making an acquisition environment integrated with IRAF which will provide quick-look tools to assess data quality.
We plan on using FITS Image Extension format for the data which keeps all of the data for each exposure in a single file with the data for each readout (64 readouts for 2 readouts/chip) kept in a separate image extension. This is the same format used for NOAO and other Mosaic cameras and is used by HST for NICMOS data.
CEA will write the data acquisition software for Megacam which will run on a VME/VxWorks machine as a command line driven program. This will allow CEA to fully develop and test the acquisition system without depending upon CFHT developments. When Megacam is operated at CFHT, it will be operated using the new GenX environment which will provide an integrated acquisition environment which will allow for efficient, scripted operation of the instrument and the TCS. A prototype of this software is being developed for use with CFH12K.
The device production for the 6-member MIT/LL consortium is well under way, with the first draft due to take place shortly after the time of writing.
A first test batch of devices on 6 inch, high-resistivity wafers has also been produced. Each batch constitutes 12 wafers of 6 devices Megacam would order two of these batches, on normal EPI silicon (high resistivity is harder to handle and has a lower yield), at $400k per batch, less 10% for two batches. Given the usable yield of around 50%, this would more than satisfy Megacam's need.
MIT/LL is particularly interesting given the possibility that we may be able to order devices with an anti-blooming drain. Built into the channel stops between CCD columns, this feature draws off excess charge produced by bright stars, thus preventing the effective loss of imaging area due to blooming. Typically, much more information is lost to blooming than to cosmetic defects in a science grade device.
The characteristics of devices tested so far can be inspected at:
Devices 66-2 and 72-2 are on high-resistivity silicon. The fringing observed in these devices is only slightly reduced compared to the EPI devices, which is disappointing given the thickness of the devices (~40 micron). Read noise is typically 2-3 electrons/pixel, CTE greater than 0.99999 and full well ranges from 110ke to 180ke. MIT/LL have the highest output node sensitivity of ~16 uVolt/electron.
On the negative side, MIT/LL CCDs do not have the UV sensitivity required by the Megacam project. Furthermore, the current devices show a "brick wall" flat-field pattern, produced by the laser-annealing process, which can have an amplitude of up to 70% in the blue (320 nm) and which renders flat fielding problematic. While they are working to address both of these issues, which are of concern to other MIT/LL consortium members, there is as yet no clear timescale for their resolution.
MIT/LL do not provide packages for their devices and will expect us to design, build and supply them. They will affix the CCDs to the package.
MIT/LL works on a best-effort basis, although they have shown considerable willingness to produce extra runs if one shows an excessive failure rate. In this sense, MIT/LL are much more prepared to act in a collaborative manner than the other vendors. Nevertheless, MIT/LL are legally required not to compete with industry (they are funded by the US government). This means that Megacam must have requirements of its CCDs which cannot be met by other vendors. The anti-blooming drain fulfills this requirement.
Of the three vendors under consideration, MIT/LL clearly has the most advanced production facility and this is running at far below its maximum capacity, which is encouraging for our confidence in their ability to produce large numbers of devices, once started, at a high rate.
EEV are currently in great demand and attempting to satisfy a number of orders. They have delivered several devices to the original contractee, RGO. The first of these suffered from a low-level (<100 electrons) deferred charge problem which they have since solved. They will complete this order in Feb 1998. These devices are known as CCD 42-80's and have 13.5 micron pixels. RGO have recently published new QE curves for these devices, with coatings optimized for blue and for red (although the blue curve is only marginally inferior to the red at ~650 nm and is superior elsewhere):
All EEV devices come with a unique "dump drain" which allows rapid clearing of charge from the device without clocking the serial register. EEV CCD amplifiers have sensitivities intermediate between MIT/LL and SITe (~4 uVolt/electron). EEV already have a four-side-buttable package design.
EEV have delivered the first thick 2kx4kx15 micron pixel CCD (CCD 44-80) of a large order (20 devices) to ESO. Thinning of wafers is currently under way. SAO have ordered 6 CCD 42-90's (2kx4.5k 13.5 micron pixels), and Gemini have ordered 12 more. CFHT has one of these devices on order for use in MOS. A number of smaller orders are also in place. EEV have shown willingness to collaborate with buyers to within their need to protect their industrial secrets which includes details of the output amplifier structure.
The most obvious obstruction to Megacam employing EEV CCDs is cost. EEV have quoted US$40k per grade 1 CCD 42-80 when ordered in quantities of 30 or more. However, they have recently stated that they will offer grade 2 devices at US$25k each. Grade two have twice as many cosmetic defects as Grade one devices. All other EEV device characteristics fall within the Megacam specification, including blue response. The fringing amplitude peaks at ~40% in the red, which is a little worse than MIT/LL's.
Our confidence in the ability of EEV to deliver the devices necessary for Megacam is tempered by the considerable delays in the delivery of RGO's order and the fact that their production line has suffered two major breakdowns in the past 12 months. On the other hand, they do work to contract and not best effort.
SITe are latecomers to the 2kx4k game - a result of a catastrophic production line failure early this year. However, they have recently delivered the first two of 32 devices to NOAO. Read noise of 3-4 electrons and CTE is very high. Full well is ~90ke/pixel, and an earlier flatness problem has been eliminated to within Megacam specifications. (A third device was also sent to NOAO, extra to contract and with an unknown problem.) SITe also have two other orders for 2 devices each from unknown US and Japanese buyers.
SITe's current price is US$38k per grade 2 device. They do not yet have a design for a four-side buttable package. There is some concern that the maximum parallel transfer rate is a little slow, limited to 5kHz, which is slower by an order of magnitude than both EEV and MIT/LL, but just falls within Megacam specifications. SITe's UV response, with the standard AR coating, is better than MIT/LL but not EEV's. Their UVAR coating is better, but is not competitive in the red.
SITe have shown willingness to collaborate with buyers to within their need to protect their industrial secrets which includes details of the backside passivation.
Device grading: It is difficult to compare manufacturer's grade specifications, as they have different ideas as to what constitutes a defect, and particularly as testing takes place at higher temperatures than will be employed with Megacam. In practice, both EEV and SITe are willing to be flexible and have stated that they will work with us to determine which devices are acceptable.
Call for bids: If we go for a call for bids from EEV and SITe, we are effectively closing the door on MIT/LL since that would be a public admission that Megacam can go ahead with devices from private industry.
Yield: in the cases of EEV and SITe, yield is a secondary issue, but could slow us down if they run into problems. This is the advantage of a commercial contract over MIT/LL. On the other hand, MIT/LL's yield is exceptionally high and, as stated, if they run into problems, they are willing to make extra runs at their own cost.
After recent discussions with IfA we foresee no obvious reasons why the CFH12k camera shouldn't be available for initial tests on the sky in mid-semester 1998 I. The current plan calls for completion of the remaining mechanical assemblies (with the possible exception of the filter wheel) by early 1998. Assembly and testing of the controller electronics should be completed by February, 1998 and at least 8 thinned ccd's should be on-hand by March, 1998 to provide a functionally useable - although perhaps incomplete - focal plane. Basic user software developed at CFHT will be available by January, 1998.
B,V,R,I and H` filters have been ordered. Delivery is expected by December, 1997. The delivery date is not guaranteed however since fabrication depends on Schott's ability to provide large sheets of colored glass of non-standard size.
The dewar outer can, focal plane housing and LN2 vessel are all fully fabricated. The dewar window has been ordered and received. Damage to the fill neck assembly caused by severe acid corrosion as a result of a gold-plating operation required that this assembly be dismantled, parts refabricated and sent out for e-beam welding. One this assembly has arrived back at the IfA in late October the assembled dewar shell will be sent to CFHT for vacuum integrity tests.
The shutter mechanism has been fabricated and mostly assembled. This unit will be sent to CFHT for testing by the end of October.
The focal plane assembly is mostly designed. Raw materials are on hand. Fabrication of the basic mechanical parts, many of which are of pure Molybdenum, will start in November. Remaining issues at this time are the routing of cables, and design of flex- cable interconnects. Fabrication should be complete by early 1998.
Detailed design of the filter wheel has not started. The wheel design will likely be based on a 'Geneva' mechanism similar to that used in another camera at IfA. Design and fabrication of this unit will be one of the last project activities.
The additional weight of the CFH12k camera over the UH8K camera will likely mean that the Z' focus motion of the Prime Focus Bonnette will stall unless some weight reduction of other weight relief is provided. The camera weight can be reduced by approximately 45 Kg (100 lbs) by locating all power supplies on the PF Cage walls. If this is not possible, then a passive counter weight system may be needed in the cage. At this time upgrading the load capacity of the PF Bonnette would be our least favored option.
SDSU v2 CCD Controller
The IfA has received a total of 3 SDSU v2 controller boards for the CFH12k camera (1 clock board, 1 timing board and 1 acquisition board) which is enough, together with a standard utility board, to assemble a completely functional test system. These boards are currently at CFHT for tests. The remaining 17 boards (including 1 spare each of the 4 types) should be available by early 1998. Delivery of boards is currently not an issue. IfA has not pushed hard to receive the addition boards since they will not be in a position to use them in the next few months.
CFHT has agreed to take on the design, fabrication, integration and testing of shutter controls, filter wheel controls, and focal plane temperature controls in order to speed delivery of the completed system. We have been asked by IfA to accept a delay of the camera delivery until April 30, 1998 on the understanding that the UH8k camera will be available in 1998 to the community at large without further reimbursement. The agreement is currently under discussion with IfA.
The first lottery for 12 detectors (6 grade A and 6 grade B devices) , from which CFHT will acquire 4 devices, is scheduled for mid-November 1997. The lottery schedule for the remaining devices has not been set.
Lockout-Tagout awareness training has been held twice for all staff and will shortly be repeated a third time to complete staff training in this important area. A draft Lockout- Tagout policy has been written (but not yet issued). Summit operations are well on their way to more careful adherence to these procedures.
A safety and hazard audit was completed at the summit by the Safety Officer.
Overcrowding and general disorganization in the basement area has been cleaned up with much of the material previously stored there transferred to more use-specific storage on the 2nd and 3rd floors.
A large quantity of old paint accumulated in the dome building over the years has been disposed of.
After a detailed presentation by the executive (see Section 1.7.3), we have reviewed the status of the Megacam project and have the following 5 recommendations.
UV sensitivity is one of the principal scientific requirements of the camera,
as stated in our last report. We are also concerned that the project remain
on schedule. If it is necessary to order CCDs immediately, because of delivery
delays or in order to fix the camera design, we recommend a call for bid with
this priority. We note that there are vendors who can supply CCDs with adequate
UV sensitivity and cosmetic quality. If CCD orders can be delayed until the
UV sensitivity and other properties of the MIT/LL devices are improved, then
they may be a better choice. SAC wishes to be informed and consulted throughout
the CCD procurement events.
For slitless spectroscopy, the incorporation of a grens appears the only way to achieve the required image quality. However, there are potential problems that may be insuperable. a) Guiding feasibility needs to be assessed more quantitatively. b) Insertion and rotation of dispersion direction need to be incorporated in the design. c) In view of the cost, the grens science programs need critical assessment in terms of instrument performance and scientific value.
As part of the overall assessment, and to minimise unnecessary Megacam effort, we request that CFHT obtain grens data in representative science fields with the existing WFC optics and the 12K camera, at the earliest practical opportunity. If successful, this should be advertised as a 12K camera capability.
We presently feel that pursuing the grens option should not be allowed
to impact strongly on the progress of the project.
We have reviewed the throughput of the WFC designs with wavelength for
the glasses BSL7Y and BK7. The BSL7Y is preferable for its UV throughput.
However, if the cost difference is a critical budget issue, the use of
BK7 appears acceptable for some or even all of the optical elements. SAC
wishes to be informed and consulted as this issue develops.
SAC considers that the acquisition of calibration data should be the responsibility of CFHT. SAC also considers that the basic pipeline processing (i.e. removal of the instrument signatures) should be done at CFHT, to avoid delays, and allow timely access to processed data during a run.
Because of the above issues, SAC is concerned that the project may be delayed, and stresses the importance of timely completion of the camera for scientific and budget reasons. In view of the heavy workload already on CFHT staff, it may be cost- and time-effective to seek a dedicated manager at CFHT.
The SAC re-expresses its serious concern about the delays suffered by the CFH12k and, even more dramatically, the OSIS-IR, projects. In particular, the scientific niche identified for OSIS-IR is very much time-limited, as similar instruments planned for the larger telescopes will soon outperform OSIS-IR on the CFHT.
SAC was informed that the team in Meudon involved in the project of the AOB-OSIS link is undergoing structural changes. This shall not impact on the project, which is still supported. The current design provides for:
A tentative schedule is end of 98-II semester for preliminary tests on the sky.
The SAC reviewed simulations, presented by CFHT, which illustrated gains in
efficiency that queue scheduling could provide. Because observing conditions
are not correlated with the scientific ranking of proposals, many highly-ranked
proposals currently receive less-than-adequate observational data. With queue
scheduling, observations would be better matched to the needs of individual
programs and the highest-ranked projects would be assured of receiving the data
they need. The actual implementation of queue scheduling, however, will require
the development of an appropriate scheduling algorithm. Also, it is clear that
flexibility is needed and that not all programs would necessarily benefit from
queue scheduling. Finally, queue scheduling would require a significant manpower
investment from CFHT. In view of these considerations, we make the following
The SAC recognizes that queue scheduling can bring substantial increases
in observing efficiency and productivity. However, its implementation will
be complex and will require the development of a new observing policy and
procedures. We therefore recommend that CFHT prepare a detailed plan, to
be presented to the SAC a month in advance of the May 1998 users's meeting,
for the implementation of partial queue scheduling of the CFH12k camera
beginning in semester 1999-I.
CFHT, with its large complement of instruments, has become a complex facility to manage. Therefore the executive director has asked SAC to start a reflective process on ways to simplify the operation of the telescope. Since the de-commissionning of foci is a possible avenue, SAC invited talks by two users of the f/35 focus. J.-P. Maillard and F. Roddier presented scientific results and a case for maintaining this top end. Obviously the science is interesting and the f/35 still supports competitive instrumentation at the present time.
The executive director described the current situation at CFHT. It is the only 4m class telescope with 4 top ends, 10 instruments and 24 different modes of operation...making of CFHT an appreciated multi-purpose telescope. However this situation brings complexity to the operation of the telescope for three different reasons: (1) the loss of personnel resulting from the layoff plan, (2) the high level of personnel turnover and (3) the implementation of the instrumention plan with its many new projects. Graphs were shown informing SAC of the approximate cost in manpower and of the scientific throughput of the different foci. It appears that maintaining a large number of foci is appropriate for CFHT at this time: (1) simply eliminating a top end like f/35 or coude would not have a large impact on costs and (2) bright time instruments like the FTS or the UH AO are needed for an efficient use of telescope time during periods of bright time for which there would otherwise be little demand.
The complexity of CFHT operations is due to its multi-purposeness and
the large number of new instruments in the making (instrumentation plan,
OASIS,...). These causes cannot be easily alleviated since the communities
have a wide variety of scientific interests and new instruments are needed
to maintain CFHT at the forefront of astronomical research. Hence a possible
way to decrease the load on CFHT staff is to plan instrument changeovers
within the framework of block scheduling. However the particulars of optical
astronomy (i.e. bright vs dark time) prevents the use of one focus per
semester. Some form of hybrid solution will have to be devised. Actions
like the de- commissionning of a little-used instrument may also take place.
The executive director will continue his work on the subject and a case
will be presented prior to the user's meeting next May where the matter
will be discussed.
The SAC re-discussed fully these issues, that had been already partly covered
at the May 97 SAC meeting, leading to recommendation #2 of the SAC May 97 report.
As a result of these new discussions, the SAC wishes to make the following recommendation:
The SAC recognizes that there are successful precedents for the handling of joint programs (proposals submitted to 2 or more of the agencies), and that continued and enhanced close cooperation between the relevant TACs eliminates the need for SAC to define detailed guidelines at this time. SAC notes with approval continuing efforts to closely coordinate the timing of TAC meetings so that such joint proposals, and indeed all applications, may be considered in the broadest possible context.
The SAC recommends that the data from any and all observing programs should have a proprietary period of one year from the time of data acquisition. Extensions may be applied for in the proposal, and will be subject to TAC approval. The SAC notes that approved extensions to the proprietary period must be communicated to the archive manager and should be appropriately implemented. All non-scientific data such as bias, dark, or flat-field frames have zero proprietary time.
Proposals may be sent to the appropriate TAC for science requiring immediate or early access to proprietary data. Such access may be approved for science goals which are distinct from the approved program of the observers, but must not unduly impact CFHT activities.
(Note: this recommendation supercedes Recommendation #2 of the May 97
The SAC again discussed the need for a wide-field near-infrared camera for the CFHT. The CFHT presently lacks any wide-field IR capability. When the OSIS-IR camera is built, this will provide imaging over a field-of-view of approximately 3.5 x 3.5 arcmin with a 1024 x 1024 HgCdTe array. But by the time the OSIS-IR camera is built, other telescopes will have similar capabilities.
Two of the principal motivations for development of a wide-field near-infrared
(1.0-2.5 microns) camera are:
Brief descriptions of some of the scientific motivations for wide-field imaging are given in the May 1996 SAC report. The excellent seeing in the near-infrared at the Mauna Kea is well matched to the uncorrected optical performance of the CFHT; images with FWHM approximately 0.4-0.5 arcsec are expected at 2 microns. A near-infrared camera with pixels subtending 0.20-0.25 arcsec would be well matched to the expected image quality.
The SAC discussed whether a wide-field infrared camera should be designed for the f/8 focus or for the prime focus. The SAC felt that the speed of the incoming beam at the prime focus would make the optical design very difficult and the optics extremely expensive. The SAC felt that it was strongly preferable to construct a camera to be mounted at the f/8 focus.
In recommendation 7 of the May 1996 meeting, the SAC recommended the development of a 2048 x 2048 near-infrared camera with a field-of-view between 7 x 7 and 10 x 10 arcmin. The SAC reaffirmed this wish. The SAC felt that to remain highly competitive further into the future, the development of a more ambitious camera based on the 2048 x 2048 HgCdTe arrays, with a mosaiced design which would need to be mounted at f/8, would be preferable. However, no money source for the development of such a camera could be identified within the current budget.
David Hanes, the Chair of the sub-committee of SAC charged with considering the long-term future of the CFHT, reported that the committee had undergone important changes in recent months, for a variety of reasons. Jean-Rene Roy felt obliged to withdraw from the committee when he accepted the position of Canadian Gemini Scientist; he has been replaced by Simon Morris. Daniel Rouan has been appointed to the CFHT Board, and has also withdrawn. For other reasons, it seems likely that at least one more of the committee members will need to withdraw, and it is possible that the committee may need to be almost completely reconstituted in light of these developments and other external conflicts.
It was recognized, in ensuing discussion, that the first submission of the Committee (that which proposed the conversion of the CFHT into a very large segmented-mirror telescope on the present pier) and the initial engineering assessment which was commissioned (the Grundmann report) needed a general reconsideration in the context of a broadened mandate. The exciting prospect of a very large telescope optimised for the infrared has been very much weakened by parallel developments in support of the NGST, a facility which would make such a reconstructed CFHT uncompetitive.
SAC recalled that the original terms of reference of the sub-committee spoke very directly of a reworked CFHT telescope and stipulated that efforts should be made to salvage as much of the present physical facility as possible. There remains the worry that such a directive would inevitably lead to a telescope facility in which engineering compromises might lead to a hybrid design which proves less than optimal. Although future detailed studies may lend reassurance on that point, SAC now believes that a broader view is required, and has requested that the sub-committee, appropriately reconstituted, continue its exploration under a revised mandate, one which specifies more generally that the committee DEFINE THE SCIENTIFIC ROLE WHICH IS FORESEEN FOR THE CFHT IN THE LONG-TERM FUTURE.
The SAC would like to thank those members who have served on the committee but who must now withdraw, and looks forward to further developments. Open discussions are planned for the CFHT Users' Meeting.
A significant fraction of the SAC meeting was devoted to the preparation of the next users' meeting. We have agreed that a users' meeting in 1998 is very timely in view of the new instrumentation now being planned at CFHT, of the new modes of operations that we foresee for the future, and of the necessity to start thinking about the long-term future of CFHT.
We plan a CFHT users' meeting on May 18-20, in Quebec City, in conjunction with the CASCA meeting. The first day (May 18) will be devoted to presentations and discussions about the CFHT recent and planned instrumentation. On day 2, (May 19) we will discuss the new modes of operations and the new policies which are foreseen for the mid-term future, as well as the longer-term future of CFHT. Day 3 (May 20), which will be common with the CASCA meeting, will be devoted to presentations by Canadian, French and Hawaiian users of important scientific results obtained recently with CFHT.
A detailed agenda for this meeting will be prepared in the coming weeks. We will send shortly an open call for contributions to all users.
SAC has discussed the draft policy for targets of opportunity presented by CFHT, and endorses it.
The text of this policy is given in Appendix A.
Considering that new and important decisions will have to be made in a 6 month
timeframe, and that all SAC members will be present at the CFHT users' meeting
in Quebec City, we consider it both necessary and practical to plan a SAC meeting
in conjunction with the users' meeting in Quebec City. If this SAC meeting is
approved by the Board, we plan to have a first SAC session on May 17, 1998,
i.e. the day before the beginning of the users' meeting, then a second session
on May 21st, after the users' meeting. The TAC meeting will be held on the evening
of May 21st.
Retiring SAC members are David Hanes, Esther Hu, Pierre-Olivier Lagage, Gilles
Joncas, and Claude Catala. The SAC thanks all of them for their active participation.
Observing time at CFHT is scheduled well in advance based upon proposals from astronomers in the C, F, and H communities. This process works well for most observations but does not work well for time variable events, such as gamma-ray bursters, a SN in M31, etc., which cannot be predicted in advance. We are proposing that in exceptional circumstances the Executive Director (ED) be allowed to override the published CFHT schedule for these Targets of Opportunity (TOO). This situation is expected to occur typically once or twice per semester. Using this policy the ED would request that the scheduled observer, or CFHT staff astronomers, make observations that have a clear potential of making a significant contribution to the understanding of the relevant astrophysical phenomenon.
Take as an example the gamma-ray bursters that are very poorly understood currently. There is a great need to identify the optical counterparts of these objects in order to determine even such basic information as their distance. We do not know whether these objects are extragalactic, Galactic or if there are two or more populations. CFHT, with its excellent wide-field imaging capabilities (UH8K and CFH12K), is in a very good position to identify optical candidates for these sources. The initial position estimates from the gamma- and x-ray observations are on the order of tens of arc-minutes that makes the wide-field capabilities of CFHT very useful. Assuming that CFH12K, MOS, or even OSIS, is on the telescope, and a well-positioned burster source has been announced, the ED could declare this a target of opportunity and request the current observer to obtain one or more images of the field. For each of the following nights, this time could be requested in order to identify any variable object as a potential candidate. This information would then be transmitted via the appropriate distribution networks so that other observatories could follow up with spectroscopic or other observations.
1. This mechanism would only be used for true Targets of Opportunity where the ED believes that CFHT is in a unique position to make a unique scientific contribution or that the opportunity itself is unique such as a Galactic Supernova.
2. Reduction of the data would be made by CFHT staff astronomers and the ED would identify an astronomer responsible for data reduction before invoking this policy to ensure that the data is efficiently utilized.
3. The amount of time requested from an observer would usually amount to no more than a one hour, or roughly 10%, of any one night.
4. The data obtained under this policy would not belong to any one person and would be made publicly available as soon as possible. Any relevant information derived from the data such as positions etc. would be transmitted via the appropriate communication channels such as the IAU Circulars and e-mail exploders.
5. CFHT would require that any publication arising from data obtained under this policy include the name of the observer as co-author.
6. The ED would make a report at each SAC meeting on whether this policy had been invoked and if so a detailed statement on the impact on the scheduled observers, the observations obtained and any results that resulted from these observations.
7. In exceptional circumstances where an observer loses more than 15% of their scheduled time due to this policy they may be reimbursed from the normal Discretionary Time.