7.6 CCD Gain Selection

---

Please check for updates on the ACS Web site.

As quantified in Table 4.1 both the WFC and HRC CCDs have selectable gain values near 1, 2, 4, and 8 electrons per digital number. Various factors should influence the gain selected in Phase II for your science program: level of support and calibrations provided, influence of associated readout noise on data quality, dynamic range on the bright end, and data compressibility for WFC in limited applications.

7.6.1 WFC Gain

GAINs 1 and 2 are fully supported for the WFC, since GAIN=1 provides the smallest readout noise, while GAIN=2 (or higher) is needed to sample the available full well depth. It is the goal, now closely achieved, to provide equal calibration support for data taken in these two supported gains, although more calibration data will be taken in the default GAIN=2 setting. Calibration support will not be provided for the “available-but-unsupported” GAIN=4 and 8 settings; users proposing their use should provide special justification and discussion of calibrations to be used. Note that WFC auto-parallel data is taken with GAIN=2.

While the readout noise is lower at GAIN=1, the advantage over GAIN=2 (< 0.3 e- extra rms) is modest. GAIN=2 has the offsetting advantage of completely sampling the full well depth of nearly 85,000 e^–, thus providing a > 0.3 magnitudes dynamic range extension before saturation is reached. The latter could be advantageous even for programs in which the prime targets are very faint, if serendipitous objects in the field of view can be used to support image-to-image registration solutions as needed for optimal dithered image combinations. Furthermore, charge is conserved even beyond filling the full well depth; for point sources at GAIN=2 it is possible to obtain valid aperture photometry several magnitudes beyond saturation by summing over all pixels bled into. Both GAINs 1 and 2 provide better than critical sampling of the readout noise supporting robust background sky-level determination even at low values. Evidence also suggests that a minor electronic cross-talk feature is relatively less pronounced with GAIN=2 than GAIN=1 (see ACS ISR 2004-13).

The large pixel count for WFC can create data rate problems if images are acquired as quickly as possible over multiple orbits. The available-but-unsupported mode COMPRESSION is more effective when the noise is undersampled which could result in special circumstances for which the GAIN values of 4 or 8 are preferred.

7.6.2 HRC Gain

GAINs 2 and 4 are fully supported for the HRC, and analogous to the supported WFC values provide a low readout noise case and a GAIN that provides sampling of the physical full well depth.

GAIN=4 on the HRC, which is needed if high dynamic range on the bright end is desired, does not provide critical sampling of the readout noise. Not only is the readout noise penalty in going from GAIN=2 to 4 non-trivial, but background estimation will be less robust without critical noise sampling. As with WFC, when the full well depth is sampled with GAIN=4 the detector response remains accurately linear up to and even well beyond saturation. Compression is not an issue for the small HRC images, therefore use of the unsupported GAIN=8 is not anticipated. GAIN=1 is available-but-unsupported, but the very modest improvement of readout noise in comparison to GAIN=2 (< 0.2 e^– higher rms) seems unlikely to present compelling need for its use.