Wide Field and Planetary Camera 2 Instrument Handbook Update for Cycle 16

1.7 WF4 CCD Anomaly

---

A serious electronic anomaly has appeared in the WF4 CCD electronics of WFPC2 wherein sporadic images have a low bias level and low photometric counts. In most cases the bias level itself is automatically corrected in the calibration pipeline, but the photometry remains low. In more severe cases the bias level can drop to zero (i.e. below the A-to-D converter zero level) and the resulting image appears blank (though bright objects and cosmic rays are sometimes visible).

The very first hint of the problem occurs after Service Mission 3B in March 2002 when a few images are seen with a bias level one or two DN below the normal 311 DN in WF4. The frequency and severity of the problem gradually increased with time. By 2003 the lower envelope of WF4 bias levels is clearly drifting downwards, and by 2004 there are many images with bias below 300 DN. By early 2005 a few blank images with zero bias level began to appear. By late 2005 nearly all images have significantly low bias levels, and 10 to 20 percent have zero bias and are blank. The WF4 bias levels at gain 7 are plotted in Figure 1.2 . Bias levels at gain 15 show similar behavior.

The problem appears to be closely correlated with the temperature of the warm electronics board inside the WF4 CCD camera head. Specifically, images with low or zero bias are associated with peaks in the temperature, which occurs as heaters within WFPC2 cycle on and off. Apparently some failing component in the CCD signal chain has developed a hyper-sensitivity to temperature. While we do not have direct control of the temperature of this circuit board, we do have some ability to adjust the overall operating temperature of WFPC2. In January 2006 an experiment was performed where we lowered the operating temperature of WFPC2, and the WF4 anomaly was greatly reduced, as hoped. Base on this success a second downward temperature adjustment was made on February 20, 2006, and this was adopted as a new permanent operating temperature for WFPC2. At this temperature the WF4 bias level is still slightly low, but the photometry is thought to be correctable, and most importantly, there are no blank images.

The downwards temperature adjustment was achieved by modifying the replacement heater set points. These heaters are located throughout WFPC2 and have a single control. The heaters are turned on when the WFCP2 temperature drops to a lower limit, and are turned off when an upper temperature limit is reached. These limits are set in the WFPC2 software. Previous to January 2006 the limits were 10.9 and 14.9 degrees C, and after the second adjustment in February 2006 they are 10.0 and 11.3 degrees C.

The temperature reduction appears to have had no adverse impact on WFPC2 performance. The PSF size is the similar to that before the adjustment. Some small motion of the CCDs in the focal plane did occur, but these are small (0.01") and similar in size to long-term drifts in the CCD positions.

The photometric impacts of the anomaly are illustrated in Figure 1.3 . Near the normal bias level of 311 DN the photometric scale is normal. However, as the bias level drops, the observed counts decrease. The amount of decrease is somewhat dependent on the pixel brightness. For bright pixels, say near 1000 DN the photometric loss at bias 150 DN is about 13%, while near zero bias it is about 20%. Faint pixels suffer larger losses. A pixel with 15 DN suffers about 25% loss at bias 150 DN, and about 40% loss near zero bias. This figure is for gain 7; at gain 15 the losses are roughly double those at gain 7 (for a given bias level in DN). The figure is generated by comparing internal flats which are impacted by the anomaly against "normal" internal flats. To the extent the curves in Figure 1.3 are well-defined and repeatable over time, we believe the photometry is correctable. Preliminary corrections are given in Instrument Science Report WFPC2 2005-02, which contains additional description of the anomaly.

Besides the photometric effects, WF4 images with low bias levels will also suffer increased background noise in the form of faint horizontal streaks and stripes. These can have an amplitude up to ~1 DN RMS at low bias levels (bias <200 DN), and are generally weaker at higher bias levels. Usually these artifacts can be removed with suitable spatial filtering of the image (see above mentioned Instrument Science Report for details).

As of this writing we expect WF4 to continue to produce good data for several more years, but there is some possibility the hardware failure might accelerate with WF4 being completely lost. The other three CCDs appear unaffected, and in fact small targets are usually placed on the PC1 or WF3 CCDs, so the WF4 anomaly has much less impact than it otherwise might. Large targets and surveys do benefit from the added sky area contributed by WF4, and would be more impacted by its failure. In addition, some specialized filters (polarizers, ramps, quad filters) rely on WF4 and would be severely impacted, were it to fail completely.

Figure 1.2: WF4 CCD Bias Levels as Function of Time.

Figure 1.3: Photometric Impact of the WF4 Anomaly (Gain 7).