CCD Fringing occurs at wavelengths longer than 7000 Å with both the G750L and G750M gratings. Peak-to-peak fringe amplitude can be up to 30% of the mean in raw data, dropping to less than 3% after calibration with contemporaneous flats.
Charge transfer efficiency (CTE) may be recast as charge transfer inefficiency (CTI = 1 - CTE), which is the fractional loss of charge per transfer to an adjacent pixel. Measured CTI is worse for weak sources, low background, and time since launch. CTI increases by ~15% per year. CTI mitigation is achieved with pseudo-apertures, which place targets closer to the readout amplifier.
Amplifier Ringing is visible in serial overscan pixels and leads to a small bias depression proportional to the signal in each row.
Extreme UV over-illumination elevates CCD dark current for times up to 40 minutes.
Interference noise produces a ripple pattern with a maximum amplitude of 1 ADU in the CCDGAIN=4 setting.
TIMETAG errors in raw STIS data used to cause spurious gaps, negative steps, or large positive jumps in time. On 2001-Sep-13, HST archive software was updated to fix all known TIMETAG problems on-the-fly, even for old datasets.
Image ghosts are caused by reflection off the CCD surface and then the CCD window. Out of focus annular ghosts are offset from the main image and contain about 1% of the signal.
Parallel ghost spectra (railroad tracks) are extremely rare, occuring for unknown reasons when observing bright red targets.
Short circuit in "Side 1" electronics on 2001-May-16 forced a switch to the backup "Side 2" electronics. Closed loop control of CCD temperature is not possible with Side 2.
Opto-isolaters in MAMA electronics are susceptible to spurious control signals generated by cosmic rays. To protect the MAMA detectors, they are only used for 5 consecutive SAA-free orbits each day.