The interferometer unlocked during the night, and did not try relocking. There might be an issue with the automation of the autorelock.

After the recovery the power in the arms has been lower by about 10% and the DARM optical gain worse by 20% than before the WE mirror swap. This made the sensitivity much worse.

Figure 1 shows an exporation of the beam spot position on WE done in CARM NULL 1F between 6:30 UTC and 8:30 UTC, using instruction listed in https://git.ligo.org/virgo/commissioning/commissioning-tasks/-/issues/68 applied to WE. In horizontal direction there was not a significant change, but in the vertical one there was a clear improvement in the carrier recycling power (higher power on B4). A 6mm vertical displacement increased the power by ~10%, back to similar values that we had before the WE replacement.

With the old WE mirror, the issue was mainly in the horizontal direction, with the losses depending on the horizontal position on the WE. So this has clearly changed. The dependence looks higher, meaning the point absorber on the new WE mirror is either bigger, or closer to the center of the mirror.

Figure 2. Once in LN3 the change in spot position has also improved the DARM optical gain by ~20%, restoring it to a value similar to what it was at the beginning of April.

Figure 3. However the offset on WE mirror position creates a very clear coupling with DIFFp_TX, with coherence up to 80%

Figure 4 shows the WE mirror before moving the beam and Figure 5 after moving the beam. Displacing the beam reduces the power on the two markes on the baffle, but it also reveals more a ring like structure at the top of the mirror, which is a few cm from the edge of the mirror. Looking at pictures from the installation that ring like feature corresponds roughly to were the edge of the first contact coating was on the front side of the mirror  https://logbook.virgo-gw.eu/virgo/?r=66548 It is noticeable, but not brighter than the point defects on that part of the mirror.

Figure 6. I have then made a couple of attempts of adding also a vertical offset on NE to reduce the DIFFp coupling. This was not successful. In retrospect, it is  because the DIFFp correction is applied to only one mirror: WE. And the COMMp correction are applied to NE, instead of having a real differential and common angular driving. Hence adding an offset to NE doesn't change the DIFFp coupling, it just adds some COMMp coupling. The COMMp correction are about a factor 10 smaller than the DIFFp correction above 10Hz, hence if an offset on WE is needed, swapping which end mirror is used for which degree of freedom would reduce the impact of a WE mirror offset by a factor ~10.

Figure 7 compares data with the WE vertical offset (purple) and data during the night without (blue). The offset improves the sensitivity above 100Hz due to the better sensing noise (higher DARM gain), but spoils it below 50Hz due to the additional DIFFp TX coupling.