Sheila, Stefan

When we first locked the TCS was in the following state:
 State A:
  H1:TCS-ITMX_CO2_LASERPOWER_POWER_REQUEST = 0.88
  H1:TCS-ITMX_RH_SETUPPERPOWER = 0.0
  H1:TCS-ITMX_RH_SETLOWERPOWER = 0.0
  H1:TCS-ITMY_CO2_LASERPOWER_POWER_REQUEST = 1.1
  H1:TCS-ITMY_RH_SETUPPERPOWER = 0.5185
  H1:TCS-ITMY_RH_SETLOWERPOWER = 0.5185

With these setting the recycling gain was fluctuating by about 30% (plot 1 between the -100sec and -200sec marks). Aslo, the dark port looks quite bad (picture 3). Since we worried about the PRC optical mode, we then decided to drop 0.3W of the TCS CO2 common heating to:
 State B:
  H1:TCS-ITMX_CO2_LASERPOWER_POWER_REQUEST = 0.58
  H1:TCS-ITMY_CO2_LASERPOWER_POWER_REQUEST = 0.8
The ring heaters stayed the same.

Plot 1 shows that over the next hour the recycling gain ratty-ness fell to below 5%, with no effect on the DC buildup of either carrier or side-band (rest of plot 1). Also the dark port improved significantly. (Picture 4)

We thus tried to optimize the settings a bit more, and dropped the TCS CO2 heaters by an additional 0.15W down to
 State C:
  H1:TCS-ITMX_CO2_LASERPOWER_POWER_REQUEST = 0.43
  H1:TCS-ITMY_CO2_LASERPOWER_POWER_REQUEST = 0.65

Unfortunately we soon after lost lock and didn't reacquire for about 1hour, after the thermal state reached equilibrium. We had no problem re-acquiring lock in this state (plot 2 & dark port picture 5). The recycling gain ratty-ness was about the same (<%5), but the recycling gain seemed to have dropped a bit. So we decided to step back to State B, while watching the interferometer. This started in plot 2 at about -100sec. The recycling gain lookced quietest at 800sec (900sec or 15 min after the TCS COS step from state C to state B).
Since the COS central heating reaches roughly 50% of its final actuation range after 15min, at Sep 23 2018 03:50:00 UTC we set the TCS system to 
 State D:
  H1:TCS-ITMX_CO2_LASERPOWER_POWER_REQUEST = 0.505
  H1:TCS-ITMY_CO2_LASERPOWER_POWER_REQUEST = 0.725





(Note: the short period between 1400sec and 1500sec on plot 1, and -800sec to -100sec on plot 2 had an additional low-pass filter on the PR_GAIN, so disregard those sections.)