AUTHORS : A.R. Martel, G. Hartig, and M. Sirianni
Measure the WFC and HRC shutter accuracy and shading effects over a large range of exposure times with OTA-like white-light flat field exposures. This is an update to the Shading and Accuracy (Jun 2000) report.
The shutter of each detector consists of two blades (or sides) located in front of the CCD entrance window. One blade blocks the optical path to the CCD and when commanded, sweeps uniformly across the detector by 90 deg to open the aperture and expose the CCD for the commanded integration time. When the exposure is complete, the shutter rotates by another 90 deg in the same direction so the second blade covers the aperture. A single exposure then requires a full 180 deg rotation of the shutter mechanism i.e. one blade opens the aperture and the other closes it. If the blades sweep at a uniform speed, all pixels will be exposed for an identical integration time. For a 0.5 sec WFC exposure, the shutter blades rotate continuously through 180 deg with no stoppage. A 0.6 sec integration time is not allowed.
The shutter sides are arbitrarily referred to as A or B. Two 180 deg shaft revolutions are required to move the 16-bit resolver through its full range of 0-65535 counts so two resolver positions separated by about 32768 counts are associated with each blade. These are tabulated in Table 1 from the Apr-May 2001 calibration campaigns. Unfortunately, the resolver positions were not written correctly in the image headers of our Feb 2001 RAS/HOMS data so our nomenclature is relative in the analysis presented below.
Table 1 : Resolver positions for the HRC and WFC shutter sides
There are essentially two shutter properties that require calibration : (1) Accuracy : Does the shutter remain open exactly for the commanded integration time ? Are the deviations between the "effective" and commanded exposure times a function of the exposure time and shutter blade ? What is the necessary correction in the count rates ? (2) Shading : Does the finite time of passage of the shutter blades over the CCDs leave striations or features on the image ? At what exposure times does shading become important and at what level ? Are the patterns repeatable for different exposure times and shutter blade ? These two properties must be calibrated as a function of the shutter side (or blade) and of the commanded exposure time. Both are tested by comparing short (<1 sec) and long exposure flat fields since any deviations will be relatively more important and easily detectable at the shortest exposures.
DATA AND INSTRUMENT CONFIGURATION :
While ACS was configured in RAS/HOMS at BATC, sets of white-light flat fields were acquired on Feb 27, 2001 (HRC) and Feb 27-28, 2001 (WFC) with the flight build detectors HRC#1 and WFC#4 at nominal temperatures and gains of 2 e-/DN and bias offsets of 3. The images were obtained with SMSs JRHW30B (F606W) and JRHH30B (F606W) and JRHH30C (F658N). A shorter version of JRHW30C (F658N) was executed manually because of the limited time available. The illumination was provided by a stable 100 W lamp in a Oriel housing (instead of the standard fiber light) feeding a fiber bundle that was inserted in the center plug of the RAS source plate. A Mylar diffuser was placed at the RAS pupil plane to simulate the OTA beam angles over the entire field, as for the standard RAS/HOMS flat field configuration. Further details on the experimental configuration can be found on the calibration plan for this campaign as well as a description of the SMSs (RAS/HOMS Calibration Plan : BATC (Feb-Mar 2001)). The light output was monitored with a photometer pointing at the fiber bundle but slightly off-axis from the direct RAS illumination path. Its output was read from green room 5 with a small telescopic ocular provided by J. Sullivan and noted in the calibration logs in uWatts or nWatts.
Median HRC and WFC bias frames were constructed from individual biases obtained during execution of the SMSs. These were subtracted from the flat fields and the residual bias levels adjusted with the leading physical overscan for the WFC and the virtual overscan for the HRC. We note that it is preferable to subtract a bias frame acquired during the shutter shading activities, with the proper external illumination, instead of a "superbias" acquired in a completely dark environment such as thermal vacuum, in order to minimize possible contamination from light leaks around the shutter (see WFC Bias and Overscan Analysis). No attempt was made to match the shutter blade sides of the biases and the flats (the shutter does not move during acquisition of a bias frame) although the filters (F658N or F606W) were matched when possible.
Our main goals are to measure deviations in count rates and evidence of shading effects at short exposure times. Median counts per pixel were calculated on each WFC quadrant and on a 200x200 subarray at the center of the HRC frames to avoid the Fastie finger. The count rates were then simply derived by dividing the median counts (DN/pix) by the exposure time. To determine the importance of shading from the finite time-of-flight of the shutter blades over the detectors, all exposures of a given exposure time and shutter side acquired in JRHW30B and JRHH30B were first co-added. The short exposures were then flat-fielded by the longest exposures (8 sec for WFC and 4 sec for HRC) with matching shutter side (Landsman 1996). Before flat-fielding, the 8 and 4 sec exposures were scaled appropriately using 200x200 pixel regions. The resultant flat-fielded images were then binned by a factor of 8 in both directions, as specified in Hack (1999). This binning offers a reasonable compromise in contrast between the striations and background - greater binnings can potentially wash out important features.
The binned, flat-fielded WFC frames are shown in Figs 1-2 for both shutter sides. Diagonal striations are clearly visible in each quadrant, oriented parallel to the Amps B-C axis. For side A, a bright, narrow band crosses through the Amps B and C quadrants in the 2 and 4 sec frames. It is not seen in the side B data. In fact, although the shading patterns between the two shutter sides generally appear similar, a closer examination shows that the details of the location and "depth" of the striations are actually different.
At short exposures, the centre of the frames is dominated by a smooth halo crossed by the dark striations. The exact origin of this halo is unknown. It is unlikely the result of the bias subtraction - an identical pattern is observed when a bias frame from the dark environment of thermal vacuum is removed or when *no* bias frame is subtracted i.e. the bias level is simply adjusted with the physical overscan. One possibility is that the halo is the result of internal reflections or scattering, for example on the front or back edges of the shutter blade or between the CCD window and backside of the shutter.
The peak magnitude of the halo/ripples was measured from normalized cuts between rows 265-290 of the binned images and is plotted in Fig. 3. Typically, at the shortest integration time of 0.5 sec, the halo magnitude is ~0.5% and quickly drops to ~0.08% at 4 sec for both shutter sides.
For simplicity, we only consider the count rates measured on the SMS JRHW30B (F606W) data and normalize them to the 8 sec count rate. From Fig. 4, the lamp output decreased by ~1% over the 10 hour period of the WFC shutter shading test. A linear fit was made to the portion of the curve spanning the W30B data and the count rates were corrected appropriately. This correction is very small and makes essentially no difference to the results. The percentage change of the count rates with respect to the 8 sec rate is plotted in Fig. 5. Typical uncertainties are ~0.2%. Below 2 sec, the count rates show significant scatter, roughly 1.5% peak-to-peak. For shutter side A, the fluxes peak at ~1 sec and then slowly level off, as seen in the Jun 2000 data. For the other shutter side, the curve is flatter.
Dark and bright regions near the noise level are seen in the flat-fielded HRC frames shown in in Figs 6-7. There are no well-defined striations except possibly for a broad band in the Amp D corner, oriented parallel to the Amps B-C axis of the 0.3 to 1 sec frames. The peak-to-peak magnitudes of the bright and dark regions are typically of the order of 0.2% or less.
The count rates from the SMS JRHH30C (F658N) images are considered. No correction was made for the small change in lamp output over the period of the SMS (see Fig. 4). The rates are normalized with respect to the 100 sec count rate. The deviations below 4 sec are relatively flat at 0.5-0.7% except for a sharp increase to 4-5.5% at the lowest permitted integration time (0.1 sec). Data between 4 and 100 sec are not available to completely track the behavior in count rates at these intermediate integration times.
CEI SPECIFICATIONS :
In paragraphs 18.104.22.168 and 22.214.171.124 of the Contract End Item Specification (Part II), Nov 1995, the exposure non-uniformities are specified as less than or equal to 5 and 10 msec for the HRC and WFC, respectively. For the shortest integration times (0.1 sec for HRC and 0.5 sec for WFC), the non-uniformities should therefore be <5% for the HRC and <2% for the WFC. For the WFC, non-uniformities are observed at levels of <0.5% and for the HRC, at <0.2%, so the observed shutter shading easily meets the CEI specifications.
Clear evidence of shading is seen in the WFC images at levels of ~0.5% at the shortest exposure time of 0.5 sec. The magnitude of the features decreases considerably at longer integration times. For the HRC, dark and bright regions, possibly from shutter shading, are detected near the noise level of the images. Inaccuracies in the shutter timing appear to be important at integration times of 2 sec or less for the WFC. The observed shutter shading meets the CEI specifications.