The STIS team routinely monitors the readnoise of most GAIN, binning, and amplifier modes available with the STIS CCD by measuring the rms value of bias images. Monthly monitors are posted at the STIS monitoring website. Recently, it was discovered that the readnoise suddenly increased for GAIN=1 from 5.7 e- to 6.25 e- on the STIS CCD between 5/14/2014 02:57 UT and 5/15/2014 00:54 UT (See this earlier STAN). We now also report that an increase of 3% from 8.23 e- (2.05 DN) to 8.47 e- (2.11 DN) for observations taken with GAIN=4 occurred within the same timeframe.
Other increases in STIS/CCD readnoise for GAIN=4 have occurred in the past. After SM3A in 1999, the GAIN=4 read noise increased slightly in concert with a jump observed for GAIN=1 (see STIS ISR 2009-02). After SM4, the read-noise had also increased by 0.2 – 0.3 e- from its 2004 levels. There was also a larger increase of about 1 DN when the switch from side-1 to side-2 for STIS was made in 2001, but that was due to semi-coherent electronic noise that can in principle be filtered out, and that case appears different from the other read-noise jumps.
The biggest potential impact is for pending STIS/CCD programs that use GAIN=4 where there are multiple reads of point sources (or spectral regions) with low S/N per individual read. Most users may see a degradation in their S/N of a few percent, within the quoted 10% accuracy of the STIS CCD ETC. PIs concerned about the impact of the read noise increase in their programs, are encouraged to get in contact with their program point-of-contact or the STScI help desk at email@example.com. The relevant reference files used by the STIS calibration pipeline, CalSTIS, will be updated in the near future.