As science operations were already under-
way, we had to make the best of the situa-
tion while trying to find a technical solution
for the problems. Eventually we found the
charge bleeding problem in the controller
backplane, which we replaced during the
telescope shutdown in August 2014. The fix
brought the charge transfer efficiency (CTE)
measurement back to ~ 0.999999 and there-
fore within specifications.
With the bleeding fixed, the “banding” issue
remained, and by February 2015, we saw
more serious complications when it coupled
to a column of hot pixels that accumulated
spurious charge on amplifier #5 (on CCD2).
With longer exposure times and binning,
the saturation of this column became worse,
causing the background level in the section
of the CCD on the same amplifier to devi-
ate from the normal level by up to 25%. The
best strategy was to avoid using the region
of amplifier #5 as much as possible. An in-
vestigation of the historic trend of this effect
showed that the problem had become pro-
gressively worse, particularly for long expo-
sures and full frame binned readout.
After lengthy investigations we identified a
solution for the banding problem by imple-
menting new ARC47 Rev.E video boards —
which still had to be modified in order to
offer the same good readout noise perfor-
mance as the previous boards. Satisfactory
results were finally reached in May of 2015.
Mysterious Events
In May 2015, another problem cropped up,
namely a CTE issue affecting CCD1 in Nod &
Shuffle (N&S) data. CCD2 and CCD3 were not
affected at all. Again we formed an ad-hoc
tiger team to work on a remediation plan.
Complicating matters, GMOS-S, our most
highly-demanded instrument, was in near-
continuous use. Then what no one expected
happened: by the end of July 2015, the CCD1
CTE problem spontaneously disappeared,
74
GeminiFocus
without any intervention, and N&S spectros-
copy programs were resumed.
Although the detectors behaved well from
September 2015 until June 2016, we were
not out of the woods yet. Following a ther-
mal cycling event of the cryostat in June
2016, a new vertical structure appeared in
the bias of CCDs 2 and 3, as well as repetitive
sharp horizontal lines. A few months later
these structures became even stronger, now
seriously affecting the science quality of the
data; in particular, due to the increased noise
over significant parts of the CCD array. Once
again we looked for measures to minimize
the effect and carried out many tests follow-
ing advice from the chip manufacturer… but
nothing seemed to improve the situation; it
seemed desperate.
Then one day in February 2017 a fault on a
compressor used for the cooling of GMOS-
S caused the cryostat to warm up. We all
feared the worst and were anxious to see the
first bias frame come out of the instrument.
To everyone’s surprise, the bias looked nor-
mal, without any of the vertical banding or
horizontal stripes!
A Happy Ending?
Well… yes and no. At this moment the detec-
tors are performing well, and we are monitor-
ing their behavior; but clearly something’s
not right as more than once serious problems
have come and gone without any clear indi-
cation why. We need to take stock of the situ-
ation and determine the best way forward.
LSST: Gemini South´s Neighbor
Comes Closer to Reality
New large telescope facilities always attract
attention, and the Large Synoptic Survey
Telescope (LSST), currently under construc-
tion on Cerro Pachón, is no exception. In
December 2016, for instance, two eminent
astronomers — Ewine van Dishoeck (Pro-
January 2018 / 2017 Year in Review