A 3.5m Status Report
3.5m Telescope Scientist
Department of Astronomy
University of Washington
postscript version of this document
This report has 3 interrelated objectives: 1) to summarize the current status
of the 3.5m and its instruments, 2) to identify current performance
limitations, and 3) to set forth a set of action items for addressing these
The document is intended to provide a framework for technical discussion within
the ARC community, so that we can move rapidly towards making more productive
scientific use of the telescope.
I want to stress a few important points at the outset. First, the diagnostic
data presented here were obtained by ARC scientists and site staff working in
close cooperation. The degree of support, competence and dedication of the
site staff is extraordinary, and they are one of the most
precious resources we have. Second,
science is being done, but we can do better. My personal
view is that we must do better in order to do forefront science in
the years to come. This brings me to my third point: this report is heavily
weighted towards my own view of how to approach the technical challenges that
we must overcome to be a truly first-class facility.
In my view the most important performance yardstick for the telescope
is the extent to which it contributes to scientific papers and
While there have been some notable successes, I do not think that the APO
3.5m telescope and its instruments are having as big an impact as we
We have made substantive engineering progress in the past year.
The initial diagnosis phase is complete, and it's time to start fixing
the identified deficiencies.
II. Telescope Performance
Some of the relevant design specifications for the 3.5m are reproduced in
Table I, taken from the documentation at the APO web site.
Although these numbers serve as a useful set of benchmarks we have to recognize
that the goalposts have moved since the specs were written. Our competitors
are routinely obtaining sub-arcsecond images assisted by active and adaptive
optics and aggressive thermal management schemes. The instruments on these
also advancing steadily.
Table I. Some Original 3.5m Performance Specifications
| Non-atmospheric seeing, all sources: ||0.29"|
|pointing to ||1" rms|
|open loop tracking ||0.2" rms over 10 minutes, to 1"
|maximum of time to reconfigure instruments ||2 minutes|
At the present time the 3.5m performance falls far short of these design
specifications. Much of the past year has been spent trying to understand
why, and that effort has been fairly successful. We have a much clearer
idea of the image quality budget on the telescope, and full instrument
characterization is well under way, as in the very nice recent work on
DIS from our colleagues at NMSU. A status summary for a variety of
categories follows, in no particular order.
Pointing and tracking
The control system of the telescope makes no fundamental distinction between
pointing and tracking. A poor pointing model produces poor tracking. Various
mechanical problems with the azimuth, elevation and rotator drive systems have
produced erratic pointing/tracking performance over the past year. Typical
"good" pointing produces rms pointing errors of the order 8-10 arc sec in the
instrument focal plane. Unguided tracking performance varies considerably, is
poorest near the zenith, and most users are generally reluctant to attempt
unguided exposures in excess of 10 minutes. This is not atypical performance
for a big telescope, but does fall short of the design spec.
Much of the engineering time in the past year has been spent working
on a variety of mechanical drive problems. This has been generally a successful
effort, and earlier problems with slippage and erratic pointing have largely
been overcome. We still occasionally have trouble with rotator overcurrents
but a program of preventative maintenance will, we hope,
increase the reliability of the mechanical systems.
The way to overcome tracking errors is to use a closed-loop guider. Equipping
the telescope with a decent guiding capability continues to be my top
priority. Using the current guider with a bright star (brighter than
14th or so in 10 sec exposures) stabilizes the image on the science
instrument to a few tenths of
an arc second, and works well. Unfortunately the current guide
camera has a variety of defects that significantly compromise our ability to
guide on faint objects. These include A/D converter anomalies,
sporadic hot pixels,
and inadequate binning capability.
Furthermore, the plate scale (0.14"/pix) and resulting FOV make
it unlikely that a suitable guide star lies at an arbitrary place on the sky.
Finally, the current guide camera has a thick CCD.
The overall result is that the only hope of finding an appropriate guide star
requires rotating the instrument and offset guider on the
sky. This makes guided multi-object spectroscopy nearly impossible. For guided
long slit spectroscopy the instrument seldom has the slit at the parallactic
angle, and we lose flux. Finally, the task of acquiring a guide star is
presently not very efficient. This task is facilitated by a nice utility
(available on the Web from the APO home page) that makes a guide star
finding chart for a given target field.
The objective is to enable guiding in essentially any patch of sky. This task
is already being pursued vigorously as a top priority. A new camera with a thin
chip is on order, and improvements in the guiding code were recently installed.
We plan to use the astigmatic nature of the out-of-focus guider images
to monitor telescope focus and adjust accordingly. Status: Active,
- Upgrade the guider:
- thin CCD
- focal reducer
- new camera
- monitor focus with guider
I consider delivered image quality to be the most important figure of merit
telescope. The smallest FWHM I know of is 0.7", achieved in a 30 sec
exposure in r
with the DSC in no wind near the zenith. This is an exception, the DSC often
reduces images in the arc sec regime while the best images on DIS are 1.3"
On GRIM, typical seeing is around an arcsec.
Table II lists my best estimate for an image quality budget for the 3.5m, based
on a variety of diagnostic tests. These contributions add in quadrature and
aggregate effect is
to be compared with the $\sim $0.3'' value from the design specifications.
Table II. Best Guess Image Quality Budget (units are arcsec)
|20 Hz. drive oscillation ||0.2|
|primary servo ||0.1 at zenith, 0.4 at 2 airmasses|
|top end instability ||0.2 to 5+, depending on wind|
|enclosure and mirror seeing ||(unknown)|
- Procure new secondary. Status: This is in progress.
- Fix primary mirror servo instability. Status: This is underway.
- Replace secondary support cage, with one like 2.5m version.
Wind-driven vibration of the top end is a major problem. The secondary support
cables have been retensioned, and temporary cross-bracing has been installed on
the secondary support structure.
- Fix 20 Hz elevation drive oscillation. Status: inactive
- Install diagnostics for seeing and thermal environment, run
comparison seeing monitors, acquire image quality frames on ongoing basis.
There are significant scattered light problems, and this has prevented a number
of projects from making use of 3.5m data. Pinhole images show a great deal of
scattered light off the cylinder that surrounds the tertiary, and from
the inner surfaces of the Nasmyth port. This is in part responsible
for gradients across flats and rotator-dependent effects.
This is one of the major impediments to extracting good data
from images on the telescope, and fixing it is a big job.
- Better baffling! Status: inactive, but
next priority once guiding task is finished.
III. Instrument Performance
Achieving an understanding of the limitations to the telescope's performance
has been my main focus, as this will benefit all instruments and
all users. This task, and following up on the things we have
learned, occupies all my 3.5m time. The instruments have not received
their share of attention, and the comments that follow are spotty at
best. There is a clear need for scientists who bear primary
responsibility for keeping up to date on each instrument.
- Appoint ``instrument scientists'' (one per instrument)
from the user community who will work with the site staff to
keep documentation up to date, monitor performance, and
identify needed fixes/improvements.
This is basically the workhorse instrument on the telescope. DIS
seems largely robust and reliable, but the red side's 800 x 800 TI CCD is a
dinosaur. There are also a variety of charge transfer problems.
Reliability of moving parts is much improved.
The recent tests of throughput seem to show a major deterioration in
overall throughput over the last 12 months, and the data from the
lens monitoring program seem to support this. See the recent writeup
on DIS performance for details.
One positive step in the last year was the installation of the slit viewing
camera on the DIS. This allows users to drive objects in to the slit and to
verify that they stay there.
- replace + rework detectors and readout chain.
- Fix throughput.
- Devise new technique for cutting reflective aperture masks.
I am personally less familiar with this instrument. Users seem broadly
its performance, but I will leave it to others to comment on its status.
While not officially a facility instrument, the DSC is effectively our
The detector is ``cosmetically challenged'' and there are some low level charge
transfer problems. The combination of the drift scanning capability of this
instrument and the agility of the telescope control system make a powerful
- Decide on DSC status and if appropriate integrate it
better into the Observatory. Remark interface? Status: sporadic activity.
The echelle spectrograph is being finished, its delivery timetable is not
known to me.
A wide field imager (roughly 8K x 8K pixels) is in the design phase. A
2K x 2K version will appear at the site this summer. The CHAOS group has made
good progress in developing a very promising adaptive optics capability.
I think that the lack of a concrete plan for instrument evolution is
a major shortcoming.
- Hold a roundtable instrumentation summit meeting ASAP.
The telescope control software is in good shape, and strikes me as a
strength of the
system. The observing software (Remark) is now fairly robust as well.
We probably will want to develop a Unix workstation based version of the
remote observing software in the future. We should consider less labor-intensive
ways of moving and displaying images.
There is some software work to be done in supporting some of the
described above. The image quality diagnostics are presently inadequate. We
also do a poor job of monitoring telescope status (primary mirror LVDTs,
motor currents, servo system signals, etc).
We need to decide if the observing assisting tools that the user community has
developed will have their ownership and maintenance transferred to the site.
Many of these fill genuine needs and are more than just conveniences.
V. Observing Efficiency
Given the scheduling philosophy of the telescope, and remote operation,
we need to ensure that observing efficiency is maximized. This has many
facets, including but not limited to network connectivity, instrument
changes, acquisition of calibration data, and overall operational
efficiency. I think we are breaking new ground in many ways here,
and that we can take steps to increase the overall effectiveness of
the facility. The "action items" here are more nebulous than the
engineering tasks described earlier.
Network connectivity is erratic between the site and the member institutions.
I think we need to actively explore network alternatives. There is an
to propose for NSF funds to bolster our capabilities in this area, which we
The acquisition of darks, flats, and lamp frames is a source of concern to
many users. It would be very good to have different users taking calibration
data simultaneously, say stacks of darks and biases, in the afternoon. Remark
does not currently support this. Also the entire calibration scheme needs
to be reviewed and enhanced, if needed.
We now have multiple instruments mounted on the telescope, and we also
carry our at least one instrument change per night, which usually lasts 20
That's 20 minutes that could have been spent doing science. We need to
automated rotation of the tertiary mirror.
A fairly constant refrain from the user community that impacts observing
is the generally poor communication between segments of the user
community, both within and between institutions. We need to devise
ways to spread the word about evolution of the systems and instruments, and
to do a better job of documenting and advertising new utilities.
It is going to take an
excruciatingly long time to make the 3.5m a fully productive facility
with the existing level of resources.
At current funding levels we can barely maintain the existing capability, and
we're making only very limited progress in addressing the deficiencies outlined
I think we face a clear choice of either committing the resources to
realize the potential of the facility, or continuing what has been
(for most users) a frustrating battle against telescope and
I have made a preliminary attempt to associate rough costs with the
tasks outlined above.
This should be considered a very crude order of magnitude estimate
to establish the scale of funding that is needed in order to obtain
acceptable performance from the APO 3.5m telescope.
|Work under way, ARC funded:|
|fix primary mirror servo system|
|enclosure wheel replacements|
|Work under way, UW funded:|
|new guide camera ||(40K)|
|new image quality data acquisition system ||(5K)|
|slit viewer ||(v 1.0 completed, 5K)|
|tip tilt secondary ||(15K)|
|Outstanding priority tasks, unfunded:|
|baffling of telescope||15K, 2 mo|
|Secondary cage||20K, 2 mo|
|Elevation axis controller revision || 5K, 2 mo|
|rotation of tertiary mirror || 10K, 2 mo|
|guider revisions. new optics ||10K, 2 mo|
|new rotator drive system || 10K, 2 mo|
|Telescope diagnostics monitor system || 20K, 2 mo|
|collimation tools and procedure ||10K, 1 mo|
|new detectors for DIS, address throughput problems &100K, 6 mo|
|DSC: address electronics effects, new CCD || 80K, 2 mo|
|total parts costs: || 280K|
|total labor: 23 months at 100K/12 months ||200K labor|
|contingency || 80K|
|total || 560K|
Go back to UC Astro APO home page
last modified 6 June 1996
Any opinions expressed on this page have nothing to do with
the University of Chicago or the Department of Astronomy and