Kiwamu, Rana

As we discussed in today's meeting, we need to improve the DRMI acquisition so that its better than 1 lock per 30 minutes (the situation all week). Kiwamu spent much of today examining a bunch of lock acquisition attempts and trying to decipher what was happening by plotting the time series and making up good stories. He will put in some details in the morning.

Around midnight, after we gave up on BS OL tuning, we decided to adjust the DRMI loop gains and the loop turn on thresholds. We had been using high thresholds to turn on PRC, then MICH, then SRC. We noticed that LLO has been leaving PRC ON all the time and then triggering SRC first and then MICH second. So we put in something similar and adjusted the thresholds to approximate the levels they were using roughly. We then also lowered the MICH and SRCL UGFs by a factor of a few to also move more towards the LLO setup.

These few steps seem to have made a large improvement. Just after we made the changes, we got several locks within 10 minutes (from midnight until 00:10). At first there was some small oscillations at a few Hz, but we think this came from having a 5 Hz MICH UGF and a triggered 3 Hz boost. We adjusted the triggering to use zero history and no ramp for the integrators and the oscillations disappeared. After that, we just had some short locks with bad alignment.

Its only a single stretch of 10 minutes, but the improvement was so dramatic that I am willing to say we are now in a region of parameter space which allows fine tuning of loops shapes and thresholds. We'll need to try with a fresh alignment to be sure, but indications are positive. We are also going to import Jenne's 'AND' logic from the 40m LSC system to see if we can make a better SRC trigger.