The 8K X 8K CCD Mosaic has been developed jointly by NOAO staff in La Serena and Tucson. It will be shared between Kitt Peak and Cerro Tololo, and is scheduled to come to Tololo in late 1998. Work will start soon on a Mosaic Clone so that each observatory will have its own large format imager, primarily for use on the 4-m Mayall and Blanco telescopes.
This project has involved large scale cooperation between two geographically separated divisions of NOAO. It has provided valuable experience in an age when instrumentation projects are growing to such a size that intercontinental collaborations are increasingly likely. For the last 18 months, this project has had the dubious privilege of being the most resource intensive instrumentation effort at CTIO. Eighteen of CTIO's technical and scientific staff have worked on the project at some time, with Roger Smith managing. An additional EE was contracted to work full time on production and testing.
The project was divided between the North and South American teams at the interface between the dewar and CCD controllers. The Instrumentation Projects Group (IPG) in Tucson procured detectors and produced the cryostat, detector mount, filter, and shutter mechanisms, while CTIO produced the CCD controllers and associated software. The IPG simultaneously carried the additional burden of building an atmospheric dispersion corrector for the Mayall 4-m telescope. The Blanco Telescope's ADC pre-dates this project; however, the prime focus pedestal will require strengthening to support the weight of the Mosaic Imager. The IRAF team is working on real time display software, an extended image format, and quick look reduction tools to replace the interim tools written by Steve Heathcote.
The Mosaic schedule required that most subsystems be developed in parallel. Our particular problem was that the Mosaic dewar and controller were being assembled in separate hemispheres and could not be tested together until shortly before the first telescope run. Instead of producing one big controller, we chose to segment the focal plane into 4 quadrants and operate each with a nearly standard quad-readout Arcon. This allowed us to test each controller with a standard lab dewar containing a pair of 2K X 4K CCDs similar to those used in the final Mosaic, and it avoided re-engineering the enclosure, backplane and supplies. The spares, a complete Arcon, do double duty as a lab system.
The decision to use multiple standard Arcons allowed us to run four systems in separate labs simultaneously and thus to harness the additional manpower required to meet the deadline. We concurrently needed to develop upgrades to the software, and to the Video, ADC and Voltage-Temperature-Telemetry cards, while also testing 50 boards, modules, or enclosures.
The transputer network, communication packet format, and command handlers already in use lent themselves to the expansion needed to handle the mosaic. The transputers nearest the Sun workstation were connected in a ring to support broadcast of commands from a single link. However, for data transfer the controllers operated independently (over separate links to the Sun) to maximize throughput and postpone the data bottleneck as much as possible.
Electronic mail, networking and teleconferencing were essential to our collaboration; but these were no substitute for staff travel and direct interaction for understanding personalities and developing a real team spirit. At least five nationalities were represented on the team. After the preliminary telescope tests on the 0.9-m telescope in May it became clear that the system integration time had been too little and too soon. We successfully reorganized, cancelling some intermediate telescope tests in favor of more lab work, and were back on schedule for the 4-m run in September. Aside from successfully commissioning the controllers, we have gained more experience in planning and managing time-critical production and development projects. Above all we now go forward with a sense of camaraderie and mutual respect for our colleagues down North.
Roger Smith, Dan Smith, Alistair Walker