Terra asked detchar to investigate whether changing PI damping settings might cause glitches. Conclusion - they do not, at least for the examples I've seen.
I looked at 22 examples from alogs 31892, 31824, 31560, 31516, 31485, 31461, 31444, 31440, 31191 where operators reported changing the settings on modes 3, 26, 27 and 28 in either/both phase and gain. I plotted a 10 minute timeseries around the times reported in the alogs of channels SUS-PI_PROC_COMPUTE_MODE#_DAMP_GAIN and SUS-PI_PROC_COMPUTE_MODE#_PLL_PHASE and a spectrogram of GDS-CALIB_STRAIN of the same time. Two of the times I could not find any change in either the phase/gain so I really only looked at 20 examples.
Attached is an example where both the phase and gain of mode 28 were changed by the operator in alog 31485. The change in gain and phase makes no obvious glitches/change to the spectrogram. All other exmaples I have seen show no obvious change either. When pycbc live triggers have been available (because they only get generated when we're in oberserving) I can not find any noticeable change.
If you really want to look through all the plots yourself they are here.
Thanks Laura!
PI phase and gain settings are currently unmonitored in observe SDF, so we can keep them that way. Note that saturating the ESD drive does couple into DARM, so if you ever try significantly increasing PI gains (say, above 15,000), it would be best to drop out of Observe.