Test Description
Over the past several months, from April-2002 to Aug-2002, 8x8 monitor windows have been added to available slots in perigee pass observations. The positions of the monitor windows were fixed with respect to specific pixels, ignoring dither. To best monitor the dark current these windows were placed in regions known to be fairly free of stars.In all, 20 observations containing a grand total of 32 8x8 readouts have been completed, with 9 or 10 warm pixels per slot. There has been some repetition in the selection process, generating very long (~10Ms), but discontinuous time series for several pixels. The graphic below shows the positions (in row/col space) of the cataloged warm pixels on the ACA CCD.
Analysis
For this analysis we define a pixel as 'warm' if the 50th percentile dark current exceeds 100 e-/sec. This cutoff is somewhat arbitrary, but consistent with the recent dark current calibration. In order to accurately determine the transitions between dark current values, light curves for each warm pixel were examined by eye. An IDL routine facilitated the selection and recording of these 'jumps'.Below are representative light curves from single pixels. Blue tick marks represent the time stamp of a 'jump' while the red line indicates the median dark current between transitions.
When selecting transitions by eye a certain amount of error is inevitable. However, the desire was to pick out changes in signal amplitude that were significant to over the 3 sigma level and that maintained that new level for at least several hundred seconds.
It should be noted that statistics were compiled separately for pixels which maintained a steady dark current or contained a single (one-way) transition. The duration statistics here are strictly lower limits. Lower limits were also compiled for each pixels state at the beginning and end of observations.
Results
The plots below are compiled from the full data set of monitored warm pixels. The two most basic parameters of interest are the magnitude (the percent difference in dark current), and duration (the time between transitions). We see that short (~1ks) transitions dominate, but that long periods of stability (~50ks) can exist between state changes. Although jump sizes of up to 350 e-/sec have been observed, the vast majority (90%) of transitions are limited to less than a 25% change from their initial magnitudes.
In the scatter plots over the initial amplitude, we see what appears to be a population of very warm pixels. Further examination revealed this to be a single blazingly hot pixel sampled several times over. Pixels with dark currents over 600 e-/sec are in fact quite rare.
Here we have a histogram of the median dark current values for the tested warm pixels. According to our most recent ( MAY-17-2002) dark current calibration, we should see more cooler (~100e-/sec) pixel than appear in this analysis. Selection effects may be responsible, as 8x8 regions of the CCD were chosen to contain a high density of warm pixels.
Stable Pixels
There appear to be very few warm pixels that maintain a consistent dark current. Only 9 of the 89 cataloged pixels remained at a constant level through one observation. Of these, 4 were found to flicker in other observations.Data Availability
All data and analysis tools are available in /proj/gads6/beu/DarkCurrent.Brett Unks
Last modified: 12/27/13