The last time we tried going to 80W, we struggled with a strange lockloss that appeared to be the PRM diverging in yaw and pulling all the yaw ASC signals off. We tried going back to 80W, this time sensing PRC2 on POP X again, instead of on REFL. We first tested going to 65W on POP, this was successful but a PI caused a lockloss. We decided to go for broke and try 80W.
Moving back to POP X sensing works in terms of avoiding the diverging lockloss. The PRM ADS signal, and DC centering loop signals stayed stable. In the previous 80 W tests, lockloss would occur very regularly around the 25 minute mark. We surpassed that timing mark by about 10 minutes. Instead, the lockloss occurred due to a CHARD Y ringup around 0.5 Hz. We think that the alignment changed slightly because there were different offsets present in the ASC REFL WFS 9 and 45 yaw (usualy just 45 has an offset).
This type of CHARD Y lockloss is very familiar, and likely because of optical gain changes at the higher operating power and slightly different IFO alignment. The CO2s have not been reoptimized at any operating power since the new CO2 alignment. It is likely that with some CO2 optimization and a gain or sensing adjustment, we could hold lock at higher power.
We also noticed that as we thermalized the power increased up to about 440 kW, and started decreasing again. We would probably find the best operating power at a slightly lower power such as 75 or 70 W input.
Powering up past 60 W is a bit hairy, because we are still using RPC which steals some phase from the loops around 0.5 Hz. This results in 0.5 Hz buzzing that can be supressed by increasing the CSOFT P gain above 20. I had the CSOFT P gain up to 38 by the last 5W step to 80 W. This could be guardianized, or we could go to 60, move to non-RPC HBW ASC, and power up the rest of the way. I also increased the time between each power up step from 45 to 60 seconds, to allow the ADS time to catch up. These are just some proposals for the future. We are not yet planning to operate at a higher power.
We are choosing to stay at 60W for now, so we can tune CO2 and give some time for PEM work. It's also worth checking how necessary closing the POP beam diverter is. If there is not much scatter present, we can continue to use POP X, since POP tends to have a much cleaner signal than REFL, especially in yaw.
Just wanted to mention that during this lock, the LSC REFL power ran away exponentially up to around 12 mW until lockloss. This is almost twice the nominal. This sort of supports the slow drift in alignment hypothesis. Probably also why the PRG and arm power started falling away. This kind of change would have a major effect on e.g. CHARD Y. Unclear why the alignment would start to drift so badly, especially with the ADS operating. Do the camera servos agree that there was significant drift?
We have seen this same drift in yaw when using ADS or the camera servos
