After the recovery of INJ and the realignment of B2 photodiodes we had a look at the correlation between parking position (intended misalignment) of the PR mirror and the slow variation of transmitted power from the RFC. An alignment check and tuning of the PR was performed by Paolo before the investigation.
With the nominal TY misaligment for the PR (-60urad) the backreflected beam is stopped on the SiC diaphragm embedded at the input of IPC3 on SIB2. With this position, we do not experience drifts of the RFC_TRA_DC. By changing the side of the parking (with the same angle) the outcome does not change and everything seems rather symmetrical.
With this position of the PR, if the ITMs are aligned (even with ETMs misaligned), then flashes can be seen on the SIB1_BD3 mount, very likely correlated to the fringes of the short Michelson. It is not excluded that these flashes can recombine with the light going to the RFC an perturb its lock. These flashes disappear once the PR is correctly aligned.
When PR_TY is misaligned by a larger amount (around 150urad) the backreflected beam hits clearly SIB1_BD6. A drift on the power of RFC_TRA_DC occurs some minutes afterward. By reducing the misalignment PR_TY (around 110urad, still hitting a bit SIB1_BD6) this drift slows down. It reverts when we go back to the nominal parking position.
A side observation: we were asked by Michal to check the angle of incidence of the waveplate of the IPC3. Not easy to correctly estimate due to the mechanics, but it looks rather close to zero.