When the default gain for the UVIS detector is used, photometric information well
beyond saturation can be recovered for relatively isolated sources in unbinned exposures (where the CCD full well is the limiting factor), but not in binned exposures (where the ADC is the limiting factor). This is discussed in detail in Section 5.4.5
and Section 5.4.6
The cosmic-ray fluxes for WFC3 UVIS are comparable to the levels seen in ACS,
STIS CCD, and WFPC2. As with these previous HST
instruments, typical WFC3 imaging observations need to be cr-split or dithered to obtain adequate cosmic-ray removal (see Section 5.4.10
). Dithering can also mitigate bad pixel effects, and can be used to sample the point spread function; it is recommended for the vast majority of observations.
No significant image-persistence effects following over-exposure were observed in
instrument-level ground test or on-orbit data using the UVIS CCDs, as expected for back-illuminated devices.
Shutter-induced vibration, or shutter jitter, affects only very short exposures.
Shutter jitter causes slight blurring in image data and is not to be confused with exposure time deviation. Exposure time deviation is discussed in Section 6.7.1
, and the UVIS shutter mechanism is described in Section 2.3.3
The image quality analysis carried out during the third thermal-vacuum campaign
revealed that vibrations of the UVIS shutter caused changes in the width and in the central pixel flux of point sources in short exposures (WFC3 ISR 2008-44)
. The shutter mechanism employs a rotary disc blade with 180 degree symmetry, including two cut-outs for the open (expose) positions. The blade is rotated 90 degrees from closed to open, for exposure, then another 90 degrees from open to closed, such that consecutive exposures use alternating cut-outs (sides) of the blade. Vibration occurs only when the servo-control electronics are enabled to rotate the disk. For the shortest exposures (0.5 sec), the shutter performs a continuous closed-open-closed, 180 degree sweep and the vibration is present for the full exposure time. For longer exposures, the shutter opens with a 90 degree turn, stops, then closes with another 90 degree rotation, with the control circuit enabled for ~0.7 sec for each move. Thus, all exposures (other than 0.5 sec) are subject to ~0.7 sec of jitter, with the remaining exposure obtained in a quiescent state. The amplitude of the vibration depends on which side of the blade is in use, so that the PSF width is alternately larger and smaller in a series of short exposures. The observed difference between PSFs in consecutive exposures diminishes with increasing exposure time as the brief period of vibration becomes less significant.
To test on-orbit performance of the shutter, observations of the calibration standard
star GD153 were made during SMOV (WFC3 ISR 2009-20)
. Series of exposures were made for a given exposure time, from 0.5 sec to 20 sec. The widths of the PSFs in the consecutive 0.5 sec exposures are shown in Figure 6.19
As anticipated, the fluctuations in the PSF width are greater in the 0-g environment.
The difference in the FWHM in alternate exposures is greater than 10% for the 0.5 sec exposures and diminishes to 1% for the 10 sec exposures, as can be seen in Figure 6.20
The shutter-dependent fluctuations in FWHM are superimposed on the gradual
change in FWHM that is generally observed over the course of an HST orbit due to “breathing”, a gradual periodic change in focus. In these SMOV observations, breathing had become the primary determinant of FWHM for exposures of 3 sec (see WFC3 ISR 2009-20
Photometric accuracy can be achieved in short exposures affected by vibration by
using a sufficiently large aperture (see WFC3 ISR 2009-20
). Fluctuating PSF sizes pose a greater problem for science programs with very bright targets when the results depend on PSF stability or on the highest possible spatial resolution. Such programs may now be able to make short exposures with less vibration using the exposure-level option BLADE=A in APT (introduced in version 21.2.2). Since this option causes additional movement of the shutter mechanism, using only the SHUTRPOS=A reported in the science header and bypassing B, its use will be allowed only as an available mode when sufficiently justified. The interested PI should send a scientific justification for using this mode for exposures with less than a specified exposure time to the Contact Scientist or the Program Coordinator, who will forward it to the appropriate instrument scientist for consideration.
The outer window is contaminated, seemingly by a mineral residue introduced
during acceptance testing of WFC3. These contamination features have been dubbed “droplets” due to their appearance at the time of discovery. In external flat-field images, these features have a strength of approximately ±0.5%. The droplets cause changes in PSF profile, such that flux in the core is redistributed to the near wings. In large-aperture (10 pixel radius) photometry of point sources stepped across a strong window feature, the feature does not significantly increase the photometric scatter. For small-aperture (3 pixel radius) photometry of point sources stepped across a strong window feature, the photometric scatter increases from ~0.5% to ~1%. Quadrant A has the lowest density of features. There are approximately 50, 129, 108, and 179 droplets in quadrants A, B, C, and D, respectively.
The best strategy for mitigating the flat-field features is an appropriate dither
pattern. Although there are positions within a flat-field feature that cause systematic errors at the level of a few percent in point source photometry, other positions separated by 20 to 40 pixels show much smaller errors, suggesting that dithers on this scale would be sufficient for most photometric programs. To ensure a point source does not hit a particular feature twice requires larger dithers of approximately 100 pixels, which is the typical diameter of these features.
WFC3 ISR 2008-10
describes the characterization of the droplets and their photometric effects based on ground testing, and WFC3 ISR 2009-27
reports that about 30% of droplet positions have shifted by about 1 pixel after launch, but have been stable since then.