Instrument Science Report STIS 2007-002
Changes in the STIS FUV
MAMA Dark Current
C. R. Proffitt
August 6, 2007
ABSTRACT
We describe the long term changes in the STIS FUV MAMA detector’s dark current and
hot pixels between the initial installation of STIS into HST in 1997 and August 2004, when
the failure of the STIS side-2 electronics rendered STIS inoperable. The typical level of the
dark current and the number of hot pixels both increased substantially over time. While
the low pre-launch measurement of 1.6 x 10−6 counts/hi-res-pixel/s persisted in a small
corner of the detector, much of the FUV MAMA was usually covered by a glow that
increased in intensity through each daily SAA free period when the detector was in use.
The rate at which the bright glow appeared and intensified increased with time. This glow
was too unpredictable to make removal of a mean glow image in the OTFR pipeline feasible. The number of hot pixels on the FUV MAMA detector had also been increasing over
time. In 1997, only 97 out of the more than 4 million pixels had a dark rate of more than
10−4 counts/s. By 2004 about 2500 pixels were at this level. Observational and data
reduction strategies for mitigating the effects of the increased dark current and the
increased number of hot pixels are discussed.
Introduction
The behavior of the FUV-MAMA dark current was initially described in a report by
Wayne Landsman in 1998 (STI IDT quick report R057). The FUV MAMA detector was
initially found to have a very low intrinsic dark current (1.6 x 10−6 counts/hi-res-pixel/s)1,
but suffered from a broad intermittent glow centered in the upper left quadrant of the
detector. Initially, this glow was present only occasionally, but the frequency of occurrence
Copyright© 1999 The Association of Universities for Research in Astronomy, Inc. All Rights Reserved.
Instrument Science Report STIS 2007-002
increased over time. At the time that STIS suspended in August 2004, this extra glow was
almost always present at some level, and was sometimes as much as 50 times brighter than
the base dark current. In this ISR we will discuss the spatial distribution of the FUV
MAMA dark current, as well has how the occurrence of the glow correlates with a variety
of detector parameters and observational conditions.
FUV MAMA Dark Current in Coadded Dark Exposures.
For each of the STIS MAMA detectors, two 1380 s dark monitor exposures were normally taken each week. A few additional dark exposures of between 300 and 1200 s were
occasionally taken as part of post-servicing mission verification or other calibration programs. For our analysis here, we use 704 separate dark monitor exposures taken as parts of
calibration programs 7093, 7604, 7950, 8504, 8843, 8426, 8920, 9615, and 10034. For
629 of these there is engineering telemetry available which allows us to determine how
long the MAMA power supplies had been on at the time of each observation.
Even at the highest dark currents shown here, the large majority of pixels in an individual frame contain no counts. It is necessary to sum more than 100 images to get a mean
value of 1 count/hi-res-pixel. When this is done, we can see clear changes is the distribution of the enhanced glow over time. Figure 1 shows the mean glow during four different
epochs of STIS operations. The glow not only became brighter over time, it had become
significantly broader and clumpier. The clumps were typically about a factor of two
brighter than the surrounding diffuse glow. At later epochs, the glow in the area surrounding the clumps also increased, and the clumps became somewhat less prominent.
Table 1. Average characteristics for the FUV MAMA dark current at different epochs.
Epoch
(MJD)
# of
images
averaged
total
exposure
time (s)
Mean dark
rate (10−5
cnts/pix/s)
Mean glow
rate (10−5
cnts/pix/s)
Mean
non-glow
rate (10−5
cnts/pix/s)
# hot
pixels
>10−4
cnts/pix/s
# hot
pixels
>10−3
cnts/pix/s
50566 - 51028
131
178803
0.21
0.36
0.161
87
3
51035 - 51491
126
173880
0.45
1.08
0.165
428
13
51542 - 52044
141
194580
0.56
1.24
0.161
1431
25
52102 - 52318
62
82020
0.72
1.67
0.172
3584
88
52326 - 52766
120
161280
0.63
1.54
0.156
2607
62
52772 - 53220
125
172500
0.65
1.65
0.154
2489
46
1. Each STIS MAMA has a 1024 X 1024 anode array grid which defines the normal (lo-res) pixel
scale for these detectors, but the raw data is normally recorded in a sub-sampled 2048 X 2048 format
(hi-res pixels). Non-repeatability of the high-res flat fields usually makes it advisable to rebin the final
images or spectra to the lo-res sampling, but when discussing the dark current and especially hot pixels,
it is often useful to consider the higher resolution format.
2
Instrument Science Report STIS 2007-002
.
Figure 1: The average count rate images from 131 STIS FUV dark monitor images taken
between April 1997 and August 1998 (MJD 50566 - 51028 in the upper left), 126 images
taken between August 1998 and November 1999 (MJD 51035 - 51491 upper right), 141
images taken between December 1999 and May 2001 (MJD 51542 - 52044 lower left),
and 125 image taken between May 2003 and August 2004 (MJD 52722 - 53200 lower
right) are displayed using a common brightness scale. Also shown are the regions defined
for measuring the glow brightness in the upper left (hi-res-pixel subarray [400:800,
1200:1600]), and the dark corner in the lower right (hi-res-pixel subarray [1800:2000,
20:220]).
The mean characteristics of averaged darks at different epochs are listed in Table 1.
Values in this table are for the hi-res (2048 x 2048) format used for raw data files. The
glow region count rates are measured in the sub-array [400:800,1200:1600], while the
3
Instrument Science Report STIS 2007-002
non-glow region counts are measured in the sub-array [1800:2000,20:220]. These regions
are also marked in Figure 1. The number of hot pixels has also increased. The evolution
over time of the distribution of the number of pixels with more than a given dark rate is
shown by Figure 2. Initially the growth rate at all count rate levels was quite high, but in
the last few years the distribution began to evolve more slowly.
Figure 2: The total number of pixels equal to or greater than a given count rate is plotted
as a function of count rate for several time periods.
FUV MAMA Temperature History
We will show later that the FUV dark glow increases as the FUV-MAMA detector warms
up. However, before investigating how the glow correlates with instrument conditions, we
will first discuss how the FUV MAMA temperatures varied over the course of STIS
operations.
The STIS MAMAs’ high-voltage power supplies (HVPS) were normally turned off
during any orbit affected by the South Atlantic Anomaly (SAA). High energy particles
would produce an excessive count rate were the HVPS left on during the SAA passage.
Each MAMA’s low voltage power supply (LVPS) was also turned off during the deepest
SAA passages because of concern that scintillation events in the optical isolators might
accidentally turn on the HV in the middle of an SAA passage. The HVPS was only turned
on during orbits that are completely unaffected by the SAA. As a result, the STIS
MAMAs were typically only available for observations for 5 to 6 orbits each day. The
amount of time between the LV and HV turn-on varied considerably, depending on the
details of how HST’s orbit intersected the SAA on a particular day.
For the FUV MAMA there are two temperature sensors that provide useful information about the detector temperatures. The value labeled by the telemetry mnemonic
OM1CAT provides a measure of the FUV MAMA electronics, while OM1TUBET measures the temperature of the MAMA tube itself. The OM1CAT value is inserted in the
4
Instrument Science Report STIS 2007-002
header of each STIS FUV MAMA dataset, but as the OM1TUBET provides a better estimate of the temperature of the micro-channel plate, the available telemetry values were
interpolated to the time of each dark monitor observation and are used for most of our
analysis.
Figure 3: The OM1CAT temperature value at the time of each of the dark monitor observations is plotted as a function of observation date. The vertical dotted lines mark Feb 6,
1998, and May 21, 2001 (when the STIS side-1 electronics failed). The solid lines show
linear fits to the temperature value in each of these time periods.
The OM1CAT values from the header of each FUV dark monitor observation are
shown as a function of observation date in Figure 3. Taken as a whole, these OM1CAT
values show a systematic increase with time of about 0.5 degrees per year. However, a
closer look shows that a large part of these changes were due to discrete changes in the
thermal environment rather than continuous slow changes. After about Feb 6, 1998, the
lower envelope of the temperatures increased by more than a degree, and after the switch
to the side-2 electronics in mid-2001, the heater configuration of STIS was substantially
different. Between these events, there was little systematic change of the OM1CAT value
over time.
The OM1TUBET values (Figure 4), show a change in February 1998 similar to that
seen for the OM1CAT values, but in mid-2001, when the switch to the STIS side-2 electronics was made, the OM1TUBET values show very little change. This contrasting
behavior of OM1CAT and OM1TUBET and the discontinuities in 1998 and 2001 are
shown even more clearly when the difference between these two temperature values is
plotted as a function of observation date (Figure 5).
5
Instrument Science Report STIS 2007-002
Figure 4: The same as Figure 3, except that the OM1TUBET telemetry values interpolated to the time of each FUV MAMA dark monitor exposure are used in place of the
OM1CAT values.
Figure 5: The difference between the OM1CAT and OM1TUBET temperature values at
the time of each dark monitor observation is plotted. As in Figure 3, the dotted lines mark
Feb 6, 1998 and May 21, 2001, and the solid lines show a linear fit to the data in each time
interval.
Turning on the LVPS caused a substantial increment in the power used by the STIS
instrument. Whether or not the HVPS is on made only a modest difference in power consumption. The MAMA tube temperature correlates very well with the length of time the
LVPS has been on, and the daily cycling of the power supplies caused a daily oscillation of
6
Instrument Science Report STIS 2007-002
the temperature, with the tube warming up throughout each day’s period of observational
availability (see Figure 6 and Figure 7).
Figure 6: The correlation between the FUV-MAMA tube temperature as given by the
OM1TUBET and the length of time the detector’s LVPS had been powered at the time of
each FUV dark monitor observation is shown. The size of each symbol increases with the
date of the observation. The open green circles mark values from before Feb 6, 1998. The
blue circles show data from between that date and May 15, 2001. The orange symbols
show data taken after the switch to the STIS side-2 electronics in mid-2001.
Figure 7: All FUV-MAMA tube temperature (OM1TUBET) telemetry values for the
period between Sep 6, 1999 and Oct 13, 1999 are plotted as a function of time. While the
NUV MAMA was safed, the FUV-MAMA tube was 4 - 5 C cooler than during the surrounding period.
The four points in Figure 6 that fall well below the trend of tube temperature vs. LV
on-time were taken during a period when the NUV-MAMA detector LVPS was off for
several days. The reduced heat output from the NUV-MAMA LVPS caused a noticeable
7
Instrument Science Report STIS 2007-002
reduction of the FUV-MAMA tube temperature during this period in the fall of 1999
(Figure 7). As we will see later, this episode provides a very useful test of how the external
thermal environment affects the FUV dark current.
Brightness of the FUV MAMA Dark Glow in Individual Exposures
In Figure 8 we plot the mean count rate in a 400 x 400 hi-res-pixel sub-array [400:800,
1200:1600] centered on the bright glow region for each of the FUV dark monitor observations. The location of this region was shown in Figure 1. The brightness and frequency of
the glow has clearly increased dramatically over the lifetime of the STIS instrument. The
very low global dark rates commonly seen early in STIS operations were seen much less
frequently at later times. The glow clearly increased as the detector warmed up during
each SAA free period of orbits. The distribution of darks over the SAA free period is not
always uniform. The apparent gap in bright glow observations centered near MJD 51750
was the result of the dark monitor observations being systematically scheduled near the
beginning of each SAA free period when the glow brightness was low. When the schedule
builders were reminded to distribute the dark monitor observations more uniformly over
the SAA-free period, the higher glow count rates were again seen.
Figure 8: The mean dark rate in the region of peak FUV glow ([400:800, 1200:1600] in
hi-res-pixels) of dark monitor observations is plotted as a function of observation date.
The symbol size of each point increases with the amount of time that the detector HV had
been on during that SAA-free period. The dotted vertical lines mark Feb 6, 1998 and May
15, 2001.
8
Instrument Science Report STIS 2007-002
Figure 9: The mean dark rate in the lower right corner [1800:2000,20:220] of the dark
monitor images as a function of observation date is shown. Symbols and lines otherwise
have the same meaning as in Figure 8. Note the change in the scale of the abscissa from
Figure 8.
In contrast, the mean dark rate in the lower right hand corner (Figure 9) shows little
change over time. A small number of points do fall above the main trend. These points
seem to fall either very early or very late in the daily MAMA operational cycle. These
may represent observations that were affected by high energy particle hits near the SAA.
The increase in the FUV dark current as the tube warms up leads to a strong correlation between the tube temperature and the glow (Figure 10); however the glow is not
simply a function of the tube temperature. The tube temperature correlates best with the
length of time the LVPS has been on (Figure 6), while, as was originally pointed out by
Brown & Roberts (2002), the brightness of the glow at a given epoch depends most
strongly on the length of time the HVPS has been on (Figure 11). However, the rate at
which the glow increases with HVPS on-time has been increasing dramatically over time.
If we instead compare the glow brightness with the length of time that the LVPS has been
powered (Figure 12), the correlation isn’t nearly as good; even when the LVPS has been
powered for several orbits, the glow only starts to increase once the HVPS has been turned
on.
9
Instrument Science Report STIS 2007-002
Figure 10: The dark rate in the high glow region of the detector is plotted as a function of
the FUV MAMA tube temperature as given by the telemetry temperature value
OM1TUBET. The size of the symbol plotted increases with the date of observation (i.e.,
larger symbols represent more recent images). Symbols have the same meaning as in
Figure 6.
Figure 11: The correlation of the glow brightness with the time the detector HVPS has
been powered on is shown. At a given epoch of observations, the brightness increases
approximately linearly with time, but the value of this slope has increased substantially as
STIS has aged. The symbol size increases with the date of the observation, and the colors
of the symbols have the same meaning as in Figure 6.
10
Instrument Science Report STIS 2007-002
Figure 12: The same as Figure 11, except that the correlation of the glow brightness is
shown as a function of the time that the detector LVPS has been powered.
So does the temperature matter at all? Or does the glow simply depend on the observation date and the amount of time that the HVPS has been on? The NUV safing episode
discussed above provides an opportunity to separate the effects of temperature from the
other parameters, due to the cooling effect it had on the STIS FUV MAMA tube
(Figure 7).
Only four FUV-MAMA dark monitor observations were taken during this NUV safing. In Figure 13 we compare the observed glow brightness as a function of HVPS on-time
for these observations with other dark monitor observations taken during the weeks immediately surrounding the NUV safing. It appears that the cooling did reduce the brightness
of the glow, but it is also clear that the glow brightness is not simply a function of the tube
temperature alone. While the glow increased more slowly when the detector was cooled, it
did increase even though the detector remained well below the temperature at which the
glow appeared in observations taken while the NUV MAMA was operating, (see
Figure 14 and also the discussion in Brown and Roberts 2002).
11
Instrument Science Report STIS 2007-002
Figure 13: The FUV MAMA glow rate as measured in dark monitor observations is plotted during a 37 day period centered on the 1999 safing of the NUV MAMA. FUV dark
monitor observations taken when the NUV MAMA was off are plotted as open symbols.
Data taken during the surrounding weeks when the NUV MAMA was operating are plotted as filled symbols. The label next to each point gives the OM1TUBET value interpolated to the time of that observation. Also see Brown & Roberts (2002).
Figure 14: The glow brightness for the same observations shown in Figure 13, are plotted
as a function of FUV MAMA tube temperature. Open symbols show the FUV dark monitor observations taken during the NUV safing, while filled symbols show the observations
taken in surrounding weeks.
12
Instrument Science Report STIS 2007-002
FUV MAMA Hot Pixels
For the FUV-MAMA we define a hot pixel as one which shows more than 10−4 counts/s/
hi-res-pixel. The data for the averaged images listed in Table 1 make it clear that the number of hot pixels has increased significantly over time, but since typical dark monitor
exposures are only 1380 seconds long, it is necessary to average over many monitor
images to reliably determine which pixels are “hot”. This makes it difficult to determine
how the number and brightness of hot pixels correlates with time or temperature, and
whether there is a correlation between the hot pixels and the anomalous intermittent glow.
To study such links we will consider the number of pixels with 4 or more counts (i.e., ≥
2.9 x 10−3 counts/hi-res-pixel/s) in each individual 1380 s monitor image.
As can be seen in Figure 15, the number of such very hot pixels has increased very
substantially over time, but, even at late times, there are still individual images that show
very few hot pixels. The number of hot pixels correlates approximately with the amount of
time that the HVPS has been powered in a way similar to that seen for the bright dark
glow, and there is a correlation between the number of hot pixels and the strength of the
glow (Figure 16), suggesting that the glow and the hot pixels may result from the same
physical mechanism. However, this relation has changed over time. Through 1999 the
ratio of the number of such “4 count” pixels to the glow brightness was roughly constant
(see Figure 17), but then the ratio began to increase systematically until mid-2002. After
that time, the scatter in the ratio increased substantially, and the number of hot pixels at a
given glow level began to decrease on average.
Figure 15: The number of pixels with four or more counts found in each 1380 s dark monitor image is plotted as a function of observation date. The size of each symbol increases
with the amount of time the HVPS has been powered.
13
Instrument Science Report STIS 2007-002
Figure 16: The correlation between the number of very hot pixels and the glow region
dark current is show for each dark monitor image. The size of the plot symbols increases
with the date of the observation.
Figure 17: The ratio of the number of pixels with 4 or more counts in a single 1380 s
exposure to the excess of the glow region rate over the dark corner rate, is plotted for those
observations where the glow rate is more the 1.5 times the dark corner rate. The size of
each symbol increases with the amount of time the HVPS has been powered.
Individual hot pixels do change significantly over time. As an example, we plot in
Figure 18 the counts seen in monitor observations for a single, arbitrarily selected, hot
FUV MAMA pixel located at x=1563 and y=961. Between about MJD 51541 and 52186
this pixel had a mean dark rate of 5 x 10−4 counts/s, but before and after this period, the
mean dark rate was < 1 x 10−5 counts/s – lower by a factor of 50!
14
Instrument Science Report STIS 2007-002
Figure 18: The counts measured in pixel (1563, 961) of FUV-MAMA dark monitor
observations.
Another illustration of the time evolution of hot pixels is shown in Figure 19. The distribution of count rates for the 338 pixels that were flagged as hot in the mean of darks
observed between MJD 50853 and 51491 is compared to the distribution of the count rates
for the same set of pixels between MJD 52113 and 52534. Despite the nearly 2 year gap
between these two epochs, many of the hot pixels observed in the earlier epoch still
showed enhanced count rates when compared with the 3 x 10−5 count/hi-res-pixel/s mean
rate for all pixels, but the majority have faded substantially below their earlier rates and no
longer exceed the 1 x 10−4 counts/s threshold used to define hot pixels. FUV MAMA hot
pixels, it seems, can persist for years, but often fade over time.
One hot region of the detector is worthy of special note. Beginning in late 1999, the
pixel at the high-res coordinates [1767, 191] began showing an extremely high dark current, with count rates as high as a few tenths of a count per second (Figure 20). Some of
surrounding pixels also showed smaller, but still enhanced count rates. This enhanced dark
current abruptly ceased about a year later, although a significantly enhanced count rate
was seen again at this location in 2001. Flat field images do not show any evidence for a
significant decrease in sensitivity at the location of this super-hot pixel (i.e., it hasn’t
burned a hole in the detector cathode).
The count rate seen in this pixel was far higher than anywhere else on the FUVMAMA detector. If we exclude the 5x5 region centered on the super-hot pixel, the next
brightest pixel seen in any individual image had only 27 counts in a 1380 s exposure.
15
Instrument Science Report STIS 2007-002
Figure 19: The distribution of count rates for the 338 hot pixels found in the mean of the
dark images taken between MJD 50853 and 51491 (solid line), is compared to the distribution of count rates for the same pixels between MJD 52113 and 52534 (dashed line).
While, most of these hot pixels have faded significantly, many remained significantly
brighter than the mean count rate for the detector during that period of 3 x 10−4 counts/hires-pixel/s.
Figure 20: The measured counts as a function of time are plotted for the central pixel of
the super-hot region that developed in late 1999.
The FUV-MAMA hot pixels may represent very localized defects in the micro-channel plate array. They do not appear to represent a danger to the long term functioning of
16
Instrument Science Report STIS 2007-002
the detector. However, the steady increase in their numbers, and in the possibly related
glow, have caused a gradual degradation of the ability to detect and measure faint astronomical sources.
Pipeline drk Reference Files.
As has been shown above, the brightness of the FUV MAMA dark glow varies in a complex way that depends on the date of the observation, the amount of time the detector highvoltage has been powered, and the temperature of the detector. While it would be desirable
to have the OTFR pipeline automatically subtract a dark image that contains an appropriately scaled and smoothed image of the glow, the accuracy with which the glow brightness
can be predicted for any individual exposure was judged to be insufficient to allow such an
automatic scaling and dark subtraction. So instead, the pipeline FUV MAMA dark file
was optimized for the case where the extra glow is absent, and any subtraction of the extra
glow must be done by the user. In the pipeline dark file, pixels flagged as hot (> 10−4
counts/hi-res-pixel/s in the mean dark) are set to their measured mean dark value, but all
other pixels in the dark file are set to a constant value of 1.6 x 10−6 counts/hi-res-pixel/sec.
This is a good estimate of the mean detector dark current when the extra glow is absent.
Pipeline drk reference files populated in this way were created and delivered for several
overlapping epochs to track the evolution of the hot pixels over time.
To aid users in making customized dark files to subtract from their images, mean
images of coadded FUV darks as well as images of the mean glow with hot pixels and the
pedestal removed are available separately from the STIS web site at http://www.stsci.edu/
hst/stis/design/detectors/fuvDarkGlow.html.
Strategies for Dealing with the FUV-MAMA Dark Current
Despite the long term increases discussed above, the STIS FUV-MAMA still has an
extremely low dark current. Many, if not most, users of the instrument can safely ignore
the dark current when analyzing their data and, should STIS be repaired during SM4,
when planning new observations. However, for planning and analysis of observations that
target very faint sources, an understanding of these aspects of detector performance is
essential.
Observational Strategies
The MAMA high voltage power supplies are turned off daily for those orbits that are
affected by the South Atlantic Anomaly (SAA), and the low voltage power supplies
(LVPS) are also turned off during the deepest part of the SAA passages. This leads to the
daily cycling of the MAMA detector temperatures and dark current. The lowest FUVMAMA dark currents occur in the 1st orbit each day after the high voltage power supply is
turned back on. If possible, it is highly desirable to schedule observations that are criti-
17
Instrument Science Report STIS 2007-002
cally sensitive to the FUV-MAMA dark current at that time. However, since there is only
one such orbit per day it is not possible to routinely schedule sensitive observations in this
way – especially those that require several orbits of FUV-MAMA data.
Given the strong dependence of the intermittent glow with position on the detector,
observers of very faint targets should consider if they can position their target in one of the
darker parts of the detector. For imaging observations, the best advice might be to move as
close to the lower right corner as the extent of the object to be observed allows. Observers
should remember to take into account the locations of known hot pixels and, if no ACQ
exposure is done, should also remember to allow for up to 2” uncertainty in the pointing.
For spectroscopic observations, STScI had defined new standard aperture locations
near the bottom edge of the detector. These are designated by the suffix “D1” and include
52X2D1, 52X0.5D1, 52X0.2D1, 52X0.1D1, 52X0.05D1, 25MAMAD1, F25SRF2D1, and
F25QTZD1. Figure 21 compares the mean dark counts in a spectral extraction at the standard location at the center of the detector to those that would be found in a spectral
extraction at the new aperture location.
Figure 21: The FUV MAMA mean dark current as a function of the detector column
number in a seven lo-res (1024 x 1024 format) pixel high extraction box near the standard
extraction position 3" below the detector center (dotted line) and in a box near the proposed pseudo-aperture position 6.8" further down (solid line) are compared. The data used
are an average of 125 1380 second dark monitor exposures taken between May 2003 and
August 2004. This illustrates the typical reduction in the dark current affecting first order
spectra that might be expected to result from using the D1 aperture positions.
Data Reduction Strategies
Even when the FUV-MAMA dark current is high, the count rates in individual pixels will
be subject to substantial fluctuations due to both small number statistics and the erratic
18
Instrument Science Report STIS 2007-002
variability of the dark current. This makes it inadvisable to simply subtract a mean dark
image.
For most observations of faint targets, the first step in post-pipeline data reduction for
an observation of a faint target should be to identify and mask out the individual hot pixels. The hot pixels can be identified by using the drk reference file listed in the image
header. In some cases, it may also be useful to construct an average of dark monitor
images taken within a few weeks of the science observations to identify any transient hot
pixels. A image of the mean glow should then be obtained from the STIS instrument web
site, scaled to match the glow intensity observed in each science image, and then subtracted from the data. Depending on the scientific goals of the analysis, it may be useful to
smooth the data and/or glow image before doing this subtraction, but the hot pixels should
always be masked out before any smoothing. Images of the average dark image and average glow (with the hot pixels and pedestal removed) for different epochs are available
from http://www.stsci.edu/hst/stis/design/detectors/fuvDarkGlow.html.
A customized identification of hot pixels might also be useful. In the darker parts of
the detector, a threshold well below the 10−4 counts/s level adopted for the flagging of pixels in the pipeline dark might be useful for some data sets.
FUV-MAMA Dark Current in a Repaired STIS
Should STIS be successfully repaired during the next HST servicing mission, (SM4), in
the fall of 2008, the detectors will have been off for more than 4 years. Given our poor
physical understanding of the cause of the FUV dark glow, this makes it impossible to predict how the glow will have changed in the interim. A further complication is caused by
uncertainty in the thermal environment that will be experienced by STIS. Thermal modelling predicts that, due to the extra heat generated by the COS instrument which will be
installed during SM4, the STIS MAMAs may run 3 to 4 degrees C warmer than they did
during 2003 and 2004. Part of the plan for STIS repair originally included the installation
of a passive radiator, the STIS Cooling System (SCS), which would have reduced the
MAMA tube temperatures by about 7 C, allowing them to run 3 to 4 degrees cooler than
previously. However, the installation of the SCS is a lower priority than many other servicing tasks that might be done during SM4, and, at this time, it appears unlikely that the SCS
will be installed.
Figure 22 summarizes the FUV dark current behavior during the last 10 months of
STIS operations prior to the side-2 failure. If we assume that the detector has not changed
during its long safing, and also assume that the FUV dark current responds to cooling the
way it did during the NUV safing in 1999, we might expect that the absence or presence of
the SCS will change the FUV dark glow behavior by about +/- 30% from what is shown in
Figure 22, with a cooler detector giving a lower dark current. If the cause of the glow has
continued to evolve while the detector has been safed, then the glow might be a factor of
two brighter than it was in 2004.
19
Instrument Science Report STIS 2007-002
Figure 22: The FUV MAMA dark current is shown for dark monitor observations taken
during the last ten months of operations before the side-2 failure. To allow more direct
comparison with values given in the instrument hand book and other documentation, in
this figure we give the count rate and pixel locations using units of low-res pixels rather
than the high-res pixels used elsewhere in this report. Filled circles show the count rate in
a region centered on the peak of the dark glow (lo-res pixels [200:400,600:800]), while the
open circles show the average rate for the entire detector. Note that the lower right corner
of the detector (lo-res pixels [900:1000,10:110]) showed little or no increase over the initial low dark current values.
After SM4, it is anticipated that science programs observing point source targets that
are sufficiently faint that the FUV-MAMA dark current is of concern will be using COS
rather than STIS, due to the new instrument’s much greater FUV sensitivity and lower
dark current. The level of the FUV-MAMA dark current will be primarily relevant only for
observers who need to obtain FUV spectra or images of very faint extended objects. It
may be possible to manipulate the policies for cycling the MAMA power supplies (e.g.,
see Brown & Roberts 2002) or for scheduling observations to minimize the dark current
for the small number of observations where it is likely to be of concern.
Acknowledgements
We would like to thank Alessandra Aloisi and David Sahnow for their careful review of
this document prior to publication.
References
Brown, T. M. and Roberts, E. 2002, “Investigating a New Scheme for Cycling the MAMA
Low-Voltage Power Supplies”, STIS Technical Instrument Report 2002-02.
Landsman, W. 1998, “Characteristics of the FUV-MAMA Dark Rate” (Raytheon ITSS).
20
Instrument Science Report STIS 2007-002
21