As part of the locking sequence, we switch the sensor for the SRM yaw loop from AS B 36I to a combination of AS B 36I and AS A 36I. This is necessary because pure AS B 36I is not a good sensor for SRM alignment at 20+ W of PSL power. Conversely, the combination of A and B is a slightly inferior sensor for SRM alignment at 2 W; one can see that POP90 increases slightly after switching.

So far, we've been doing this switch as part of the 2 W ASC engagement sequence, which (under normal conditions) occurs five minutes or so before increasing the power to 20+ W. However, over the past week, this switch seems to make the interferometer less stable, and tonight alone has caused two lock losses in a row.

Therefore, I moved this sensor switch later, so that it occurs just before increasing the PSL power. So far it's worked twice in a row.