We installed thinned SITe science-grade CCDs in the CCD Mosaic Imager in June 1998. Preliminary results from an engineering run in July were already reported in the September 1998 NOAO Newsletter, but many changes have taken place since then.
First, the terrible news: one of the CCDs self-destructed. For unknown reasons, CCD #4 developed a large region of saturated pixels. A replacement CCD was installed and the defective chip was sent to SITe for a post-mortem.
Now, the good news:
1) The replacement CCD was installed in August with no problems and has been working perfectly (thanks to Rich Reed and Tom Wolfe).
2) We reduced the readout time from 4 minutes to 2.5 minutes with some minor hardware modifications and by adjusting the clocks (thanks to Roger Smith at CTIO).
3) In the last Newsletter, we indicated that an "on-the-fly" processing feature would be added to the display program. This functionality is required because some of the CCDs exhibit strong (10-15%) spatial sensitivity variations that make it impossible to display faint objects at high contrast. We have added this optional processing step (overscan and flat-fielding) to the display task (raw data is not touched) so that observers can evaluate their data immediately and effectively. That feature is now fully operational and works great (thanks to Frank Valdes, Mike Fitzpatrick, and Doug Tody of the IRAF group). In the process, Frank found and fixed a significant bug in the display task that was slowing down the display of incoming data.
4) Astrometric solutions are now installed for both the 0.9-m and 4-m setups for a variety of bandpasses (thanks to Lindsey Davis of the IRAF group).
5) Sensitivity from U through (SDSS) z' has been verified to be at the expected level. Count rates in electrons per second for a 20th mag star for the 4-m (0.9-m) at UBVRI are: 40 (2), 330 (14), 340 (14), 410 (16), 225 (9). The sharp-eyed reader will see that the ratio between the 4-m and 0.9-m is ~ 25 rather than ~ 17 as expected from the ratio of the clear apertures of the telescopes. (The U-band ratio of 20 is another story—-see below.) We are looking into the possibility that the 0.9-m is suffering from some unexpected loss (or the data were taken on a far inferior night than we thought), or that the 4-m has grown since we last measured it.
6) A new coating was applied to the last element of the 4-m corrector to replace a degraded Solgel coating that was exacerbating the presence of the "ghost pupil" (thanks to David Vaughnn for leading the Solgel research effort). With the new coating, the ghost severity has been reduced to a nearly imperceptible level (0.5-1%) in BVRI, and rises to ~ 1.5% at U. Narrow-band imaging, where the problem is most evident, is also much improved. Before the re-coating, images taken through a 50 Å (FWHM) [OIII] filter exhibited a ghost level exceeding 15%; it is now under 4%, and this is the worst case filter.
7) After 37 nights at the 0.9-m and 4-m telescopes with the new, thinned, science-grade CCDs, the upgraded Mosaic system has performed at an extremely high level of reliability. It appears that nearly all significant problems have been resolved. This is a non-trivial point because requests for Mosaic next semester total 230 nights!
8) The manual and all web pages have been updated to reflect the new CCDs and software (see the Mosaic web page: http://www.noao.edu/kpno/mosaic).
We are still in the throes of developing the ultimate U-band filter. Several attempts have demonstrated that we can achieve superb efficiency with a peak transmission of ~ 84% with no red leaks (thanks to Jim De Veny, Gary Poczulp, and David Vaughnn). This filter, though, uses liquid copper sulfate that can attack the Schott glass that contains it, leading to a loss of throughput and image quality. (The low 4-m count rate noted in item 5 is a consequence of this effect.) Further, any liquid poses a hazard to the instrument should a leak develop. We are continuing along several paths to obtain that elusive filter.
Hardware for the second copy of the Mosaic system was shipped to CTIO, received, and tested. Receipt, testing, and integration of the second complement of SITe science-grade CCDs, and modifications to the Blanco 4-m prime focus are the two most important outstanding tasks. Mosaic-II is expected to be on-line for testing and commissioning at the Blanco 4-m telescope by the third quarter of 1999.
There are two issues that may affect observing strategies:
1) Cross-talk ("echos"). Two CCDs are run from each of four controllers in the NOAO Mosaic. We have found that saturated stars in one of the CCDs produce "echo" images in the other CCD of the pair. In some cases, not-quite saturated stars leave faint echoes, too. The direction of the echo is one-way: from CCDs 2, 4, 5, 7 to CCDs 1, 3, 6, 8, respectively. We are investigating ways to reduce the amplitude of the echo (currently at < 0.1%), or providing software to flag affected pixels.
2) Binning: Data obtained with the current set of CCDs, if binned, is mysteriously destroyed. With the readout time reduced to 2.5 minutes, and the option to block average the data after the fact, most programs should not be affected very seriously. Note that the 0.9-m and 4-m pixel scales (0.42" and 0.26", respectively) are barely adequate to sample the seeing profile, and so, binning will usually degrade the spatial resolution of the images. Consequently, we recommend against the use of binning.
George Jacoby, Taft Armandroff, Todd Boroson
(for the Mosaic Team)