Here is a comparison between the signals at the LSC-REFL_A servo now and 3 days ago. The filter for the ASC-POP quads made huge difference in the frequency range from 10Hz to 100Hz. The rms is about a factor of 2 lower due to the microseismic. This might explain the drop in RF power at this detector reported in alog 45100.
Here is the comparison between the readbacks of the REFL common mode board before and after the gain reallocation in the AO path.
The error signal shows a flat noise of unknown origin above 10Hz. The slow and ctrl signals should be identical above the 5Hz high pass in the ctrl path, but clearly are not. The gain reallocation actually made the coherence worse!
Turns out that the measurement in alog 45100 was done when RF level was particularly small but it got worse over time, just not as bad as before.
Microseism could explain the difference between somewhat smaller RF level after it got worse (right blue arrow) VS old high RF level (left blue arrow) but not the difference between particularly good (pink arrow) and old high RF level (left blue arrow).
Just by looking at the time scale, it looks as if the heating was passing through the optimal point at the pink arrow and kept going in the other direction.
Keita asked us to check the ITM HWS sensors to see if the LSC-REFL_A levels correlated with the IFO heating during the increase in power. It looks like ITMX doesn't change very much while going through the nominal point but ITMY is still trending. Recall that the ITMY CO2 level during lock acquisition is 0.55 Watts while ITMX CO2 is 0.19 Watts, not to mention that there are point absorbers on ITMY which definitely affect the overall heating spherical power fitting.
Overall, it looks as if ITMY CO2 requires about +5-10 udiopters central heating to go from the steady state to the optimal point, which corresponds to about 50-100mW of extra CO2 heating.
The numbers I used are on the TCS simulation MEDM with the central mask actuation gain for a single pass being 2.5e-5 diopters/watt.
We increased the ITMY CO2 Guardian request hot state power to 0.02W from 0W. We lost lock before we could see if this made any improvement though.
