Some observations: both loops seem to be roughly critically damped, while I was aiming for some slight overshoot, so the servo gain is a bit too low. For the NI, the temperature stabilizes slightly below setpoint, which causes the TCS power to increase with a ramp. This might be explained with a linear decrease of the environmental temperature, which a simple integrating servo cannot follow perfectly. The small difference between TCS power and set-point for the WI are due to the finite resolution of the power meter and a do-nothing-limit in the rotator servo. Finally, some small steps can be observed in the mirror temperatures, which are artifacts of the line-monitoring algorithm.
Fig 2 and 3 show the etalon plots for both mirrors. Also these plots show that we are not yet exactly on the top or bottom of the fringe. One thing to note is that for both mirrors, the curves do not overlap exactly with the historical data. One explanation for this might be misalignment of the end-benches, but in the past this was observed as a slow drift over several weeks. Another cause migh be the alignment of the TCS beams, which probably changed quite a bit during all the work of last week. This could have an effect, since the frequency of the drum mode is more or less dependent on the temperature of the mirror as a whole, while the etalon depends only on the temperature of the center of the mirror. A different temperature distribution due to TCS alignment could therefore introduce a phase difference in the etalon plots.
I changed the set-points to 3.02 K for the NI and 3.16 K for the WI. The servo gain has been increased a bit.
