At 18:41 UTC I stopped the IR BPC for the remaining part of the long measurement with AOM loop on, in order to understand if the long term fluctuations of the AOM correction may be related to the angular control.
With the IR BPC disabled, the AOM correction drifted by about 4 Hz over 30 hours. When the IR BPC was enabled, the AOM correction drifted by about 350 Hz over 70 hours. The AOM correction seems correlated with the IR BPC correction signals.
There was a typo: with the IR BPC disabled, the AOM correction drifted by about 40 Hz over 30 hours (not 4 Hz).