CCD Mosaic Imager Status Report
(from KPNO, NOAO Newsletter No. 49, March 1997)
The image is a single 5 minute exposure of M33
taken at the 4-m.
Images from the eight calibrated CCDs have been
transformed into a single, geometrically corrected (for CCD alignment
and optical distortions) image. This image is a combination
of 5 shifted exposures which have been registered and medianed. Trails
from extremely bright stars remain but all other defects and gaps
vanish. The two images are shown at the same size of 8Kx8K though the
full combined image is 9Kx10K. Image reduction and construction
courtesy of Frank Valdes.
[the images are on madrona.tuc.noao.edu, available via anonymous FTP,
cd pub, get m33_single.eps.Z, get m33_final.eps.Z.]
Since the most recent Newsletter article on the CCD Mosaic Imager, we
have had Mosaic testing runs at the 0.9-m in January and the 4-m in
November and January. The software has advanced, the system is more
reliable, and we have conducted numerous tests.
The 4-m Corrector
During the January engineering run, we performed an initial evaluation
of the new corrector and its atmospheric dispersion compensator (ADC).
Although the seeing was variable during the run, excellent subarcsecond
images were obtained near the zenith with (or without) the ADC in use.
Images also were taken up to 70 degrees from zenith in B. At the
larger zenith distances, the ADCs improved the images as
expected, although the seeing at the time was about 1.3 arcsec so the
improvement was small. The software to operate the ADCs allows the
observer to use them in a transparent tracking mode for long exposures,
or to set them to a null position where no correction is applied.
There remains one problem with the corrector that we expect to have
largely eliminated before the April 14-17 engineering run. When using
narrow-band interference filters, a faint image of the telescope pupil
falls on the CCD and has a diameter of about 10 arcmin. Depending on
the bandpass and construction of the filter, this reflection typically
manifests itself at a 3-5% level above the background. It arises from
an internal reflection off the front surface of the rear element of the
corrector despite the use of an extremely good anti-reflection (AR)
coating. Our investigation suggests that similar 4 element correctors
currently in use should experience a similar effect and tests performed
recently by Alistair Walker with the CTIO 4-m confirm this analysis.
For the purposes of the Mosaic, our approach to solving the internal
reflection is to re-coat the critical surface with an AR coating
specifically optimized for superb performance over the wavelength range
where most interference filter work is done: from 5000 to 7000 A.
To extend much beyond this range, interference filter users
will have to rely on flat-fielding to minimize the effects of the
reflection. During subsequent engineering runs, we will be exploring
observing strategies to minimize the effect.
Filters
We currently have B, V, and R large-format filters for Mosaic.
We are in the process of ordering several additional large-format filters.
Depending on the pricing and vendor delivery schedules, we hope
to obtain all or most of the following by June: Kron-Cousins I-band,
H-alpha (80A FWHM) and 4 redshifted filters at velocity
intervals of 1800 km/s, [OIII] 5007 (50A FWHM), and a
continuum filter at 5300A (250A FWHM).
Data Pipeline
For each read of the Mosaic array, 128 Mbytes of data are generated.
The readout time currently is 110 seconds (with one amplifier per CCD),
but the time to handle the data and save it to disk requires an
additional 4 minutes. Only then can one display the image and perform
typical quick-look interactive analysis to assess the data quality. We
consider this overhead of 6 minutes unacceptable.
The IRAF group has developed a strategy to improve the throughput of
the Mosaic data pipeline significantly. A data messaging system,
operating across computers, will take packets from the Arcon CCD
controllers and ship them to our recently acquired Sun Ultra-Sparc 2.
Clients on the Ultra will unscramble the packets from all the different
CCDs/amplifiers, save the data to disk in an efficient new FITS format,
and simultaneously display the data during readout to a new real-time
image display. During the readout/display process, data can be accessed by
all IRAF quick-look utilities (e.g., IMEXAMINE) to assess the data
quality (e.g., focus, background level) even before the full image is
finished reading out.
Preliminary tests of the messaging system, the fast ethernet link
to the Ultra, and the new data format all appear promising. The new
display system has not been developed yet, but should be available
in the Fall. In addition, the data pathway from raw images to astrometrically
corrected and combined images has been verified. These tests utilize
the new IRAF astrometry package, a new multi-amp version of CCDPROC,
and the new multi-image FITS format that will be available under the
next version of IRAF (V2.11).
As an example, images of M33 were combined to produce the image shown
above, along with a single image showing the defects and gaps that
disappear during the IRAF processing phase.
Mosaic During Spring Semester
We do have additional 4-m Mosaic testing scheduled for February, April
and July. In addition, three nights of shared-risk science have been
scheduled at the 4-m in June/July and seven nights at the 0.9-m in
June. We received eleven Mosaic expressions of interest for this time.
Six of these programs had technical requirements consistent with
current Mosaic performance, and we have solicited and received detailed
project descriptions from these proposers. We will finalize the
disposition of these shared-risk blocks shortly and notify the users.
Science-Grade CCD Upgrade
The largest remaining task in the Mosaic project is the upgrade to
science-grade CCDs from the current engineering-grade devices. As
described in the December Newsletter, we have placed an order with SITe
for thinned high-QE low-noise 2K X 4K CCDs. We do not know
when we will receive eight suitable CCDs from SITe and have them
optimized. Therefore, the operating assumption is that the current
engineering-grade CCDs will remain the Mosaic detectors for the fall
semester. However, we will do everything possible to expedite the CCD
upgrade. For example, construction has begun on a copy of the Mosaic
dewar. The science-grade CCDs will be installed in this new dewar in
order to minimize the Mosaic downtime needed for the CCD upgrade. The
dewar that is currently in use will become part of Mosaic Clone, which
will eventually be deployed at CTIO.
Applying for Mosaic Time
We invite applications for shared-risk Mosaic time in the fall
semester. Proposals should follow the normal application procedure.
The proposals will be given a science grade by the TAC and also
reviewed for technical compatibility with the current state of the
Mosaic Imager by the Mosaic team. Proposers should list what filters
they plan to use and the required photometric uncertainty at a given
magnitude. As described above, there can be no assurance of the CCD upgrade
occurring during the fall semester, but there is some possibility
that it could occur late in the semester. Therefore, proposers should
write their proposals for the engineering-grade CCDs; please feel free to
let us know how the program would benefit if the science-grade CCDs
become available. Programs that will have the greatest likelihood of being
scheduled will have high science grades from the TAC, technical
requirements that are consistent with the performance of the current
CCD Mosaic Imager, and, to a lesser extent, observing goals that will
aid in final Mosaic evaluation and commissioning.
Many users will need to choose between Mosaic and the standard CCD imagers
at the 4-m and 0.9-m (T2KB and T2KA, respectively). In the
September Newsletter (Volume 47, page 27),
we cautioned users on a
number of deficiencies of the current Mosaic engineering-grade CCDs:
they are unthinned and uncoated chips and therefore have poor sensitivity
in the blue; the Mosaic CCDs are much poorer cosmetically than T2KB or T2KA;
the effective readnoise can be as high as 30 electrons
over parts of the array. In
addition, we only have a limited supply of Mosaic filters at present
(see above). Finally, although the system is maturing, users should be
aware that Mosaic is not yet as user friendly or reliable as PFCCD.
Rough guidelines for using PFCCD versus Mosaic are as follows:
- The standard CCD imagers have fields of 14X14 arcmin at the
4-m with the new corrector (T2KB) and 23X23 arcmin at the 0.9-m (T2KA).
- Mosaic has a field of 36X36arcmin at the 4-m and 59X59 arcmin
at the 0.9-m.
- Any program requiring deep exposures in the blue, such as U or B,
should use T2KA/B.
- Any program needing specialized filters not listed above
should use T2KA/B.
- If the primary determinant of success is going very deep
and/or having small photometric errors, T2KA/B are likely preferable.
As an illustration of this, the count rates listed for Mosaic at the
4-m in the December Newsletter
are down by 78% in B, 54% in V, 38%
in R, and up by 17% in I compared to the rates measured for T2KB and
the new corrector/ADC.
- If the primary determinant of success is covering as much sky
area as possible, Mosaic is likely preferable.
Feel free to consult with any of the undersigned if you are unsure of
whether to apply for PFCCD or Mosaic for Fall 1997.
For updates on the progress of the Mosaic project, check out the
Mosaic web page
at http://www.noao.edu/kpno/mosaic/mosaic.html
Taft Armandroff tarmandroff@noao.edu
Todd Boroson tboroson@noao.edu
George Jacoby gjacoby@noao.edu
Rich Reed rreed@noao.edu
For the Mosaic Team ...