In the last few days we always had a pitch motion very well visible in the AS port, when locked. It turns out that this is due to PRM: when locked, the PRM witness sensor see a 20-30 urad pitch motion at about 150 mHz.
We tried to switch off the top stage offloading, but this did not improve the situation. So we need to rework the M3 stage length to pitch decoupling. What we have right now dates back to 2014...
It looks like this might be a bad diagonalization of the witness sensors.
In full lock, I looked at the PRCL longitudinal drive, and at the ASC error signal for PRM pitch (POP_A QPD).
Although the PRM pitch witness sensor is highly coherent with the longitudinal drive and with the longitudinal witness sensor, there is almost no coherence of the PRM ASC pitch signal with either PRM pitch witness nor PRM longitudinal drive.
