Figure 1. Lock acquisition were failing, with an image on the B1s camera looking several orders of magnitude too bright, and the power on B1 PD3 being a factor 10 higher than usual. After some thought this could be explained by the beam polarization at the OMC input being turned from S polarization to P-polarization. Which would increase the transmission of the beam towards both these sensors. A check in single bounce confirmed that most of the B1s power is on the B1sP (p-polarization) instead of the B1s photodiode (s-polarization).
The issues has appeared between the morning lock, and the first relock attempt at ~11:30 UTC. So is most likely related to the squeezing work that happened during that time. Checking the logfile there was no command sent to rotate the OMC waveplate. So this is very odd, unless the SDB1_rot process was bypassed and the waveplate driver was used directly by adding a cable with physical buttons to the driver. At the moment there is no rational explanation why the north waveplate on SDB1 has turned, rotating the beam polarization by almost 90 degree.
We have checked that action on the SDB1 rotator that form the shutter on SDB1 for the beam towards squeezing is not causing a cross-talk and rotating the waveplate.
Given that the polarization was turned by almost 90 degrees, it meant that a half wave plate, not a quarter wave plate needs to be turned. From the logbook entry of the waveplate installation it wasn't clear if the north or south waveplate is a half wave plate. But the sentence was with half wave plate first, so I assumed that the first waveplate seen by the beam is a half waveplate (north waveplate), this turned out to be correct as rotating I could bring B1sP close to zero. Then I used scans of the OMC around the TEM00 to reduce the size of the downward peak on B1sP when going through the p-polarized TEM00 mode. This polrization tuning in single bounce was done around 18:00 UTC. Probably there is still some fine tuning to be done. To be check more carefully during the next maintenance.
Figure 2. The B1s camera might have been damaged by the steady high power and a couple of unlocks. But the image still looks reasonable, although it is different from what I remember.
If in the future we want to send more light to the EDB OMC the infortunate experiment above can show us a way of how to do it without opening SDB1. We can just turn the polarization to P-polarization, and then on SDB2 add half waveplate on the B1 and B1s beam paths to turn it back to S-polarization. In this way the beam splitter sending the beam to the B1s beam dump on SDB1 will have much lower reflectivity and send a much larger fraction than 10% of the beam to SDB2, which can then be routed to EDB.
As a side comment we had an issue with the rotator driver two months ago. But at the time it was not receving commands, and that was solve by disconnecting, reconnecting, and adding back the rights for the USB port. https://logbook.virgo-gw.eu/virgo/?r=66613
