The MITLL2 Dark Current Problem
Jessica Chapman, AAO, February 9, 1999.
1. The Problem
Dark currents measured with both the TEK and the MITLL2 CCDs during UCLES
and UHRF observations have been consistently higher than the dark currents
previously published on the web, based on measurements made in the Epping
lab. Here I do not discuss the MITLL3 which has not so far been used for
coude observing (as far as I know).
The raised dark levels are a major problem for observations of faint
sources where the dark current becomes the limiting factor in determining
the signal/noise ratios. This problem is most severe for UHRF observations
as a very large number of pixels (up to 1200) are integrated (spatially)
to form the final spectrum.
As an example, in his recent Jan 99 run using UHRF@1E6 + MITLL2, Ian
Crawford measured dark currents with the MITLL2 which ranged from 2.5
e-/pix/hr, approximately 12 hours after powering the CCD up, to 0.72 e-/pixel/hr
after 3 days. These are a factor of 9.3 and 2.7 times the published values.
These higher levels effectively washed out his observations of 8th magnitude
stars. Indeed it appears that a dark current of around 1 e-/pix/hr on the
MITLL2 would restrict UHRF observations at 1E6 to sources brighter than
approximately V=6. Ian's data appear to be consistent with a dark
current measurement I obtained last year giving a dark level of 1
e-/pixel/hr 1.5 days after the MITLL2 was installed.
High dark currents have also been a serious limitation to UCLES observations
of faint sources and this has raised several complaints from users
over the last couple of years. Here the spectra are extracted over a much
smaller number of pixels and the dark current levels have limited
observations of sources fainter than around 15-16th magnitude (Brian may
have more precise information on this).
2. Dark Current Sources
The raised dark currents are caused by:
The dark current on the TEK CCD initially drops fairly rapidly but takes
at least one day to reach its lowest value. Previous data I have show a
dark level of 1.5 photons/pix/hr after 25 hours. This compares with an
expected best-level of 1.0 photons/pixel/hr.
the room flourescence - this has been known about for many years.
The coude room walls appear to 'glow' for 2-3 days after the room lights
have been turned off.
The long clean-out time of both the TEK and the MITLL2 CCDs.
The clean-out time has not been systematically measured on the MITLL2.
However, tests by John Barton with the MITLL3 show a clean-out time of
2-3 days. For Ian Crawford's run, the dark current (measured with the camera
shutter closed to avoid room glow) had dropped to 0.72 photons/pixel/hr
after three days. This is still more than a factor of two higher than the
level of 0.3 e-/pixel/4000s measured by John. Its not clear to me why the
level was still so high or whether the best measured value of 0.3 e-/pixel/4000s
could in practise be reached.
3. Impact on Operations
3.1 Long Term Solution
It seems to me that the only long term solutions are to:
(i) Paint the coude walls (I gather that paints are being investigated).
(ii) In future acquire a CCD for coude observations that has a much shorter
clean-out time - so that the nominal dark level can be reached during
a single day. Do any CCDs have this characteristic?
3.2 Short Term Solution
For the present I offer the following advice:
(i) It would be really helpful to do some systematic tests of the MITLL2,
to properly measure the dark level with and without the camera shutter
closed, over an interval of several days or more. This would give a better
understanding of both the full clean-out time and the level of room
glow. John Barton also advises that tests be done of the dark current
linearity with time. At present we have very limited data for the MITLL2.
(ii) At present users are advised that BEST dark level they
are likely to reach with the
MITTL2 CCD is 0.7 e-/pixel/hr, and that the dark current will be
higher than this if the CCD is installed on the day of observations.
(ii) For UCLES observations of faint sources (> 14th magnitude?)
TEK should be installed at least one day in advance of the run and
MITLL2 should be installed several days in advance.
(iii) For all UHRF observations, install the TEK at least 1 day in
advance and the MITLL2 several days in advance of the run. This is
absolutely critical for faint sources (>~ 6th magnitude) but I'd
advise this is done routinely for all UHRF observations.
(iv) Once installed the CCDs should be run on a no-break supply and
should not be switched off! If the CCD power is turned off for any
length of time, the dark level is reset back to the starting value
again! The CCDs must also be kept at a stable temperature.
For both the flourescence and clean-out problems it's clear that the
the CCD has to 'settle down', the better the outcome will be.
Report by Jessica Chapman (firstname.lastname@example.org)
Converted to HTML by Chris Tinney