The issue seems to be at the level of the ADC or electronic rack reading the optical lever. The first plot is showing the time series for a few channels of this optical lever. At the time of the power change, there is a gap in the data, meaning that it’s an issue at the level of the readout electronic. The PSDF_X1/X2 channels have also short glitches before these data gap, glitches looking like digital noise. The second figure with the trend data shows that the issue on the redout of the PSDF_X1/X2 channels start half an hour before the power drop, with also a few data gap. I might be interesting to check the corresponding digital electronic.