At LLO there are problems with engaging the PRC2 ASC loop when the recycling gain is low, but we don't seem to have this problem at LHO.  One theorey about the difference could be that we just arrive in lock with a higher recycling gain.  Although I believed this myself, the data I downloaded from the first 13 days of ER8 seem to indicate this is not the difference.  I looked at times when we first arrived on resonance (112 examples), when we transition to DC readout (after the ASC including soft loops has been on for about 1 minute, 85 examples), and after we power up to the maximum available power (60 examples).

• We have a 5% discrepancy between the recycling gains measured in transmission of the 2 arms, so that gives us an idea of the accuracy of this calibration into recycling gain.
• We seem to have a bimodal distrution of recycling gains when we first lock, centered around 34 (35 according to the Y arm) and 36 (38 according to the Y arm). 
• The ASC brings us to a more consistent high recycling gain.
• According to the X arm, we keep the high recycling gain when we power up, while the Y arm indicates that the recycling gain drops a bit as we power up.  This is interesting, it might be nice make the same plot for POP_DC. 
• I was curious if the locks that start with a lower recying galn survive to low noise, so I looked at the same plot made for only locks which reached the nominal low noise state (44 examples). It looks the same, meaning that we make it to low noise even when we arrive on resonance with a low recycling gain.