This page provides news about the YALO Telescope, ANDICAM, and the web
pages of interest to observers, proposers, and consortium members. Entries
are in reverse chronological order (most recent news first).
- 2002 September 24:
YALO Project End of Operations
- After more than four successful and highly productive years of
operation, the YALO Project was completed and observing operations ended
on 2002 September 24. This move will make way
for the start of the SMARTS Project, and the move of the ANDICAM to
another telescope as part of that venture early next year. This website
and its observing records will be kept online for the foreseeable future
as a service to all YALO users.
On behalf of all of us who have worked on the YALO project since its
inception, we would like to thank all of the astronomers whose
imaginative use of this unique facility have made it an unqualified
scientific success. We are among the first such projects to demonstrate
the scientific utility of this mode of observing, and to do it on a
shoe-string budget. The YALO operations model will be the basis of our
future ANDICAM operations with SMARTS. We are not so much ending YALO
as moving on to the next logical step along these lines.
In particular, the YALO project would not have been possible were it not
for the tireless dedication of our two observers, Juan Espinoza and David
Gonzalez. Thanks guys!
On to SMARTS...
- 2002 June 27:
CCD Problems - an Update
- Shortly after getting back into operation after two weeks of
terrible weather at CTIO, the substitute observer Sergio reported that
the ANDICAM CCD had stopped working. The last night of CCD data was
June 5. On June 6, CCD images read onto disk only contained what looked
like noise+bias signal regardless of integration time or degree of
illumination, and with none of the usual artifacts from hot areas or
blocked columns. We also saw no cosmic rays. We could occasionally get
non-bias signal if we blasted the detector with light, but nothing
By June 11, a series of software tests conducted remotely by OSU
personnel with help from Juan Espinoza determined that it was not a
problem in the data-taking software. Bruce Atwood was on travel, but
nevertheless got in touch with electronic technicians at CTIO and
initiated a series of tests over the phone to try to isolate the problem
in the CCD readout electronics. By June 14 it was clear that there was
no obvious solution, and they had failed to pinpoint the problem on any
of the readout boards. While there are no firm conclusions as yet, it
is increasingly likely that the CCD detector itself is malfunctioning.
At this date we are looking at options for verifying the fault and
replacing the CCD if necessary. The latter requires a significant
amount of effort. In the meantime we are proceeding with IR-only
observations. More news will be posted when we learn more.
- 2002 Feb 8:
Back in Operation
- Bruce Atwood arrived at CTIO on Feb 5 and replaced the shutter on
Feb 6. Shutter tests on the night of Feb 6 and afternoon Feb 7
verified that we have a working shutter. Juan executed an old
(pre-breakdown) observing program to verify that everything was back
Regular science operations resumed on the night of February 8.
Many thanks to Bruce Atwood, who made the rescue run, and to Oscar
Saa and the technical support staff at CTIO for all of their
- 2002 Jan 28:
ANDICAM Offline: CCD Shutter Dead
- The CCD camera shutter finally died a definitive death during
testing at CTIO. Oscar Saa and Bruce Atwood were unable to
revive the shutter, but did determine it was not an electronics
failure. The prognosis is a total mechanical failure. ANDICAM
will remain offline while Bruce Atwood prepares to travel to
CTIO to replace the shutter. The earliest we could be back online
is Thursday, Feb 7.
- 2002 Jan 14:
CCD Shutter Problems
- The CCD camera shutter in the ANDICAM has been sticking open
intermittently. The problem first appeared in mid-December
following the recent filter change, but then went away for
the most part. Most CCD images of the last few weeks have been
unaffected, but when it does strike the problem is kind of obvious.
The main symptom is that you will see bright stars streaking vertically
near the middle of the CCD, usually looking like a line of beads extending
above and below the star images. The brighter the star the brighter
the streak. The cause is the shutter sticking part-way open during
readout. The effect is particulary damaging to flats and bias images.
[Example of an affected I-band image].
During routine tests to try to determine the nature of the problem
and its duty cycle during the Christmas/New Years's break, the
problem seemed to have cured itself. A regular regime of tests
was instituted, and late during the week of January 6-12 we
resumed taking dome flats with the CCD camera.
The problem resurfaced on the night of January 13/14 with a vengeance.
Here are images of a flat-field screen using the entire CCD of the
shutter stuck open in two positions:
Mostly Open and
A series of tests has been initiated to try to characterize the problem.
We at OSU are also working to get a new shutter prepared and to make up
detailed instructions on how to replace it should that be required.
Since this involves opening the vacuum vessel, we anticipate about
3 days of downtime will be required for a replacement, and want to
be sure it is needed before proceeding.
There is no "fix" for images clobbered by this artifact, but since
it tends to mostly affect the boundary between the two CCD halves,
it does not (usually) affect the region of the CCD that is shared
by the IR channel (where the observers are instructed to center
- 2002 Jan 14:
Web Server Downtime
- The YALO web and ftp servers will be unavailable from noon EST
on Tuesday, January 15 until 9am EST on Wednesday January 16.
The downtime is needed to move the YALO web pages and associated
observer files over to the new department web and ftp server and
to turn off the old server.
- 2001 Dec 6:
Johnson U Filter Replaces Stromgren u Filter
- The Stromgren u-band filter in the CCD filter wheel (position 2) has
been replaced with a filter that reproduces the Johnson U bandpass.
This filter completes the UBVRI set, and has substantially higher
throughput than the Stromgren u filter. [rwp]
All web forms and device tables at the telescope have been updated
to reflect the change.
- 2001 Feb 19:
Stromgren Filter FOV
- In response to requests from AURA users, I have prepared a "finder"
image for the Stromgren filter set on the CCD channel.
The Stromgren filters are only 25-mm diameter, and thus only
partially illuminate a round spot centered on the ANDICAM CCD. The
effective FOV is 500-arcseconds (1660 pixels), with some
vignetting, and a little off center towards the top of the image as
viewed by IRAF unflipped.
The orientation of the FOV on the sky is shown in this
screenshot of a star field
imaged in the Stromgren y filter (300s total integration). This is
as the image would be displayed by IRAF (XVista users would see E/W
inverted vertically). The circle shows 500-arcsec diameter, and the
noisy left half of the CCD has been masked.
Because of the persistent noise problems with the left half of the
ANDICAM CCD, users requiring precision photometry should locate their
targets in the rigth half of the CCD, within the 500-arcsec circle.
- 2000 Nov 27:
Changes to the Report Archives
- Security changes to our web server at OSU required me to modify the
nightly report archives to create "index" files
for the reports. Our server no longer permit access to "bare
directories". Nightly observing reports have been indexed,
and the reporting scripts modified to re-create the index files as
required. Nightly trouble reports will be indexed like this
- 2000 Sept 28:
CCD POSTCLEAN set correctly
- In taking sequences of short (20s) CCD images, we noticed that the
first image of each set had many more cosmic ray events than all of
the preceeding images. On checking the IC settings, we found that
POSTCLEAN, which somewhat counter-intuitively sets the number of
pre-exposure erase cycles had been set to zero (0). This means that
the CCD was not being erased before each image, and if the first
image was preceeded by a long period of deadtime, it would have
greater cosmic rays and effective dark time.
- POSTCLEAN has now been set to 1, which will have a minor (2-3 second)
impact upon the total execution time, but well within our canonical
5-10% estimate envelope, so the execution-time calculator has not
been changed. You should notice that the first bias frame of a sequence
or first short-frame of a sequence not preceeded by another image
should be cleaner.
- We do not know why this variable was set to zero, or by whom.
- 2000 Sept 12:
New Web Tools
- New web tools are available for YALO investigators on the
Phase II Observing Program Pages:
These tools should simplify making program modifications and
tracking the progress of your observations. If you encounter
any problems, please contact Rick Pogge
- Observation Template (Obs) file editor.
- Simple observing log search tool.
- 2000 Sept 8:
CCD Software Upgrade: READ THIS!
- We have upgraded the software on the CCD channel data-taking system.
This institutes a number of changes:
- We can now readout the entire 2048x2048 CCD.
- Bias and overscan must now take into account 2-amplifier
readouts (details below).
- The CCD gain has been reset to its correct nominal values, but...
- this affects the readout time, and hence all observation execution
times need to be revisited for on-going programs.
We'll treat each of these issues in turn.
- Full-Frame Readout
- Hooray! An error of 1 bit (!) in the CCD sequencer map was responsible
for the garbage on the left-half of the ANDICAM CCD. This was found and
fixed and we can now readout both halves of the detector. This restores
the full ~10-arcmin field-of-view of ANDICAM.
- Bias and Overscan Issues
- We readout the CCD with two amplifiers, one for the each of the left
and right halves of the detector. The horizontal readout registers
on the Lick/Loral-3 CCD have 16 extra
pixels for each readout amplifier that are read before the data pixels
in a line. These constitute "hardware underscan" pixels. After
reading each line of data, we readout an additional 32 pixels of
"overscan" at the end of each line. These latter are used to measure
the DC bias level during the readout of that line on that amplifier,
and constitute the traditional "overscan" columns familiar to most CCD
observers. Our data-taking system repacks the data and under/overscan
pixels so that the non-image pixels are to the outside (left and right)
of the data sections. This can make it look like the images have extra
overscan. The diagram below should clarify this.
[Schematic of a raw ANDICAM CCD frame
showing the locations of the data, underscan, and overscan regions.
[Typical Bias Image (161Kb GIF)
before subtraction of overscan bias. This image is shown with
exaggerated contrast, +/-20 ADU, to emphasize the difference in
the mean DC bias levels between the two halves of the
device. This difference vanishes after overscan subtraction. The
pattern noise on the left half is also visible, as are the bright pixels
in the first 10-odd data columns at left.]
With 1x1 binning, the raw image size is 2144x2048 pixels, and the
data and bias sections for each half of the device are:
The BIASSEC coordinates explicitly avoid the 16 pixels of
"hardware underscan" in each line, and only use the 32 pixels
of "overscan" to determine the DC bias level. Note that the
first 10-odd data columns on the left side are unusually bright,
and may be discarded with a TRIMSEC of [60:2096:1:2048].
With 2x2 binning the raw image size is 1072x1024 pixels, and the
correspondng data and bias sections are:
With 2x2 binning we now have 8 pixels of "underscan" and 16 pixels
of "overscan" for each side. The BIASSEC's above avoid the hardware
underscan as noted above. In addition, the first 8 data columns are
bad in the 2x2 binned images, and may be discarded by using
a TRIMSEC of [32:1048,1:1024].
The DC bias level is slightly different from one half to the other,
but this should comes out in the overscan processing. However, there
is a mysterious pattern noise (ripple) with a peak-to-peak amplitude of
~3-4ADU that is visible in the left-hand side of the detector in bias
(zero) images. This will be harder to diagnose fix, and does
not entirely come out in 2-D bias (zero) subtraction.
- CCD Gain:
- Back in November 1999 when we replaced a crashed disk in the
CCD controller, we did not notice that an outdated config file
was on that disk that inadvertently changed the CCD gain setting. As
a consequence, the CCD gain has been running high, ~8 electrons/ADU,
since November 1999, resulting in a higher effective readout noise
(nearly 20 electrons) due to digitization error. This was discovered
in Feb 2000 by Nick Suntzeff, and called to our attention.
We have reset the gain with the software upgrade. The latest gain and
readout noise numbers for the ANDICAM CCD are:
gain = 3.6 electrons/ADU
Please use these for all exposure time estimates.
readout noise = 11 electrons (rms)
- CCD Readout Times
- Changing the gain has the effect of making the readout time longer
than before. The new nominal readout overheads are:
1x1 binning: 80 seconds (74s readout + 6s pre-exposure erase/setup)
The observing template creation scripts have been updated to reflect
2x2 binning: 24 seconds (24s readout + 6s pre-exposure erase/setup)
- Observers who have been trying to reduce deadtime in dual-mode (CCD+IR)
observations may wish to revise their observing templates as necessary.
Note, however, that there is little point in trying to make the deadtime
smaller than ~20 seconds per template because of various unpredictable
system latencies that can erase any apparent reduction of deadtime.
- You can do this most simply by using the new obs file editor. After
logging into the Phase II Preparation tools, select "Edit Old
Observation Template Files" from the tools menu below, and follow
- Note that the IR array readout time is unaffected.
- 2000 Sept 1:
Automatic Observing Log Archiving
- Starting Sept 1, we are automatically archiving all observing logs
on the web at 10am the following morning. These are available
in the Observation Log Archive index.
Some logs may not be available right away if the network is
acting up. Missing logs usually appear the following day (unless
one of us intervenes by hand, unlikely after classes start at OSU).