We've been having difficulty locking this afternoon (microseism around 0.6 um/sec rms, winds consistently gusting up to 35-40mph at times up to 50mph), the locklosses seem fairly typical of the type that plauge us with high ground motion when we are locked on TR_CARM and ALS_DIFF, see alog 22211

These locklosses seem to be caused by some kind of glitch that happens only while ALS is on DIFF.  I had a look at the LLO guardian and talked to Adam, who doesn't think that LLO has this problem.  There seem to be three differences in the way that we do this step compared to LLO: they ramp the CARM offset much faster (3 seconds for the whole thing vs 25 seconds here), they don't run a servo to ALS diff from AS45Q durring this step, and I believe that they go to a lower CARM offset before they switch DARM from ALS to AS45Q if I'm interpreting the normalizations correctly.  

For now I just sped up the ramp times a good deal, reducing both the 15 second ramp in reduce CARM offset and the 10 second ramp in SWITCH_TO_QPDs with 3 second ramps.  This seems to be a bit better, I think that it has worked  3 out of 5 times that we have tried it in the last few hours, which was better than earlier in the day.  For now I will leave this in.