Elli, Evan
For the past week or so I've noticed that initial alignment of the corner optics has become much more painful:
- With PRX locked in order to align PRM, the WFS loops seem to do nothing to improve the POP buildup.
- When locking MICH on a dark fringe in order to align BS, the two spots on the AS camera appear to escape from each other, as if there is some integrator feeding back to the BS pitch. This prevents MICH from locking.
The second of these problems was traced to the LSC_CONFIGS guardian: the threshold for LSC-ASAIR_A_LF_NORM_MON was too high (300 ct), so that even if the lock was broken, the guardian would not register it. This would leave the MICH loop trying to acquire lock with two integrators on and no limiter on BS-M3_ISCINF_L. I've turned the threshold down to 30 ct. The MICH locking seems more sluggish than it did a few weeks ago, but now it locks instead of wandering off.
Elli and I tried for a while to fix the first problem, but nothing we tried seem to work. Turning up the gain only made the PRX lock oscillate. Turning off the slow pitch/yaw bleed to the M1 stage seemed to have no effect. Different combinations of REFL_A and REFL_B WFS seemed to have no effect. So this will require some more attention.
Today I was able to get good buildup in PRX (ASAIR_LF of about 4500 ct). In order to do so I had to adjust both PRM (by tens of counts) and PR3 (by a few counts). Since the WFS loop only touches PRM, it is not so surprising that it cannot always achieve good PRX buildup. In the past we usually haven't had to touch PR3 for day-to-day alignment, but since it's now used for DRMI/PRMI ASC loops, its pointing is now more variable.
Once PR3 has been touched up by hand, the WFS loop can take care of the rest. However, since it involves a very slow bleed-off from PRM M2 to PRM M1, we may want to find something faster for initial alignment.
One point I am not clear on is why PRM M1 P/Y and SRM M1 P/Y now have different filter settings. PRM M1 has a 1/f shape everywhere; SRM M1 has a flat gain of order unity. This makes SRM M1 more prone to getting kicked when M1 feedback is turned on or receives a sudden change in input; at one point it caused the HAM5 ISI to trip.