Keita made the point that we should phase AS36 where it needs to be for putting the BS signal all in one quadrant, and then use the MICH ASC offset if we need to adjust the BS pointing at 35W.  Keita, looking at the BS resonances while we're in lock, guesses that we're using a phase about 70 degrees away from what we should be.

When we had ~just gotten to 35W IFO lock I put in a BS pitch dither line at 8 Hz with 600 counts, and phased AS36 to minimize the peak in AS_A_RF36_I_PIT.  So far, the ASC seems okay, although we're only about an hour into the lock.  This doesn't put the SUM of Q to zero, which leaves us vulnerable to DC centering-like effects.

As Daniel has already noticed, to do this I have moved the AS_A_RF36 phase 34 degrees opposite of what I had been asking operators to do, which puts us 74 degrees different from what we had been using for the later parts of lock (after about 90 mins).  This is nearly exactly what Keita had guessed by roughly eyeballing some spectra.  The next question becomes: how was it working at all?? Have we basically not been controlling our BS motion for the last few days?  I'm hopeful, although haven't confirmed, that we'll be able to acquire lock with this phase and not have to change things part way through the lock, with the idea that we had been acquiring with a phase about 40 degrees wrong, but then moving it a further 30 degrees in the wrong direction made the gain in MICH ASC so low that we weren't controlling the BS, which is why we struggled to lock.