Previous Article Next Article Table of Contents
The NOAO CCD Mosaic Imager Project (1Mar94) (from NOAO HIGHLIGHTS!, NOAO Newsletter No. 37, 1 March 1994) Individual CCDs have hit their size limit with the Tektronix 2048 x 2048 chips we now have in use at both KPNO and CTIO. We have been working for several years on the development of CCD mosaics to surpass this size, the ultimate goal being an 8192 x 8192 array with 15 um pixels. Because this is an ambitious project, we have set an intermediate goal of producing several 4096 x 4096 arrays using the same procedures that are required for the larger mosaic. The mini-mosaic development has proceeded well for the last couple of years. Two-side buttable 2048 x 2048 CCDs with 15 um pixels were fabricated at the Loral Fairchild foundry. We contracted to Michael Lesser of Steward Observatory the task of thinning and packaging the CCDs. This part of the process turned out to require a substantial amount of research and development. In the meantime, we jointly developed a mounting scheme that would accommodate either thick frontside-illuminated or thin backside-illuminated CCDs, and Lesser packaged four thick chips for us to assemble into a prototype 2 x 2 array of CCDs. This prototype was put into a standard KPNO universal Dewar, and has seen engineering time on both the 0.9-m and 4-m telescopes. It has been run with both our old CCD controller and one of the new CTIO ARray CONtrollers (ARCON), which we will be installing at all our telescopes next summer. The chips used in the prototype are of poor cosmetic quality, but they have allowed us to learn a great deal about how to assemble, operate, and reduce the data from a device this large. The gaps between the individual CCDs are about 600 um. The rows and columns are lined up to a couple of pixels and the whole array is flat to about 15 um RMS. This is fine for the f/7.5 focus of the 0.9-m, but is marginal at the faster 4-m prime focus. We believe that we can achieve flatness of 5 um RMS on the science grade device. The reduction of data from the mini-mosaic is a time consuming operation. We have found it easiest to keep the data from the four chips separate through the bias subtraction and flat fielding operations. We then assemble the four individual images into a large composite image using astrometric solutions derived from observations of an open cluster. We have not seen any evidence for changes in the relative geometry from run to run, and so the hope is that users of this device will be able to use a predefined transformation. Because the devices are of poor quality, it is desirable to take a number of frames with different centers, register them, and combine them. The figures are two mini-mosaic images taken at the KPNO 0.9-m telescope. They are a single image of the full moon, and a combined image of five exposures of Abell 262, showing that the gaps and cosmetic defects can be successfully removed by this process. [Figures not included] Our hope is to finish up three thinned mini-mosaics in the next six months. These will be deployed at KPNO, CTIO, and at the WIYN telescope, where the 0.15 arcsecond pixel scale should be a good match for the superb image quality expected. At the same time, we have begun work on the large 8192 x 8192 mosaic imager. We have initiated a run of 2048 x 4096 CCDs at Loral Fairchild. These are three-side buttable and the eight best chips will be assembled in a 4 x 2 format to produce a square CCD array five inches on an edge. The design of the dewar, mechanical interface, shutter, and filter mechanism are well along. Watch the Newsletter for further details. Mini-Mosaic and Mosaic Specifications Mini-Mosaic Mosaic Format 4096 x 4096 8192 x 8192 Pixel size 15 um 15 um Readout time 150 seconds 150 seconds (goal) Scale at 4-m PF 0.30"/pixel 0.30"/pixel Scale at 0.9-m f/7.5 0.43"/pixel 0.43"/pixel Field at 4-m PF 20' square 41' square Field at 0.9-m f/7.5 29' square 59' square Todd Boroson
Previous Article Next Article Table of Contents