A Touch of Zebra in the Old Workhorse: Correcting the Bias Striping in Post-SM4 ACS Images

Norman Grogin, nagrogin@stsci.edu

Hubble’s Advanced Camera for Surveys (ACS) includes a Wide Field Channel (WFC), the telescope’s workhorse for visible-light pictures from 2002 until its electrical failure in 2007. The WFC has resumed this role following Hubble’s Servicing Mission 4 (SM4) in 2009, thanks in part to replacement electronics for controlling its charge-coupled device (CCD) light-detectors. These replacement electronics have brought the WFC back to life as an even more sensitive instrument for detecting faint light from the cosmos. This is because the electrical noise contribution to each pixel in an image, the so-called “read noise,” is now 20% less than the WFC had with its original CCD electronics. However, the new electronics also contribute a low-level noise that appears as a fluctuating horizontal striping across all post-SM4 WFC images—a touch of zebra in the old workhorse (Figure 1, top).

Specifically, this striping noise arises from the application-specific integrated circuit (ASIC) that provides bias and reference voltages for the two WFC CCDs. M. Loose described the characteristics of this ASIC in the proceedings of the 2010 HST Calibration Workshop.  One of the ASIC reference voltages e xhibits a low-frequency flickering (1mHz to 1Hz) that is not nulled-out during the CCD read, and therefore it appears as an additional time-variable contribution to the image bias.

The restored WFC no longer has a completely steady bias. The time variation of the signal is consistent with electronic “1/f” noise, also known as “pink noise”—meaning the flickering is more pronounced over longer time-scales or lower frequencies. The implication for WFC images is that the variation is too slight to perceive along an image row, but the bias level does noticeably vary from row to row, and more strongly between more distantly separated rows. Moreover, the bias striping appears to be uniform among all four of the amplifiers used to read the two WFC CCDs: the same striping pattern appears across the amplifier boundary of a given WFC CCD, and is mirror-reflected across the WFC CCD gap. This bias variation amounts to less than one-quarter of an individual pixel’s read noise. Nevertheless, the striping appears prominently in low-background WFC images because human vision is so attuned to pattern recognition.