Yesterday I put both X and Y lasers into the DOWN guardian state, left them for a while to run unlocked.  I then set both guardian states to LASER_UP and left them to lock themselves and run over night.  Attached is the chart (guardian_locking_X_and_Y_lasers.png ) showing the output power, which is +/-0.015W for the Y arm laser and +/-0.008W for the X arm laser.

The PZT movement is +/-5V on both lasers, with the full range available being +/-35V, so this seems good.  The chiller movement is approximately 0.05C on both chillers, which again is very small compared to the chiller range.

There is a glitch seen periodically in the Y-arm laser power output with a 15minute interval.  We believe that we have tracked this down to being the interval that the FLIR camera uses to re-calibrate (this involves some internal mechanical motion) and looking at the attached chart (Y_Laser_FLIR_GLITCH.png) it seems that we have got rid of it by remotely turning off the FLIR camera from the MEDM screen.

The chiller setpoint for both lasers looks to have some small amount of oscillation.  The period is 31mins and it is seen at about the same level in both lasers.  Looking at the LVEA temperature data (temperature_fluctuations.png ) there is no correlation to actual lab temperature changes.  Also we can see that the oscillation appears in the PZT output channel, showing that the faster PZT actuator is subtracting out the length change caused by the oscillation in the chiller.  The chiller uses this voltage as the setpoint for its actuation and it looks like the chiller lags the PZT by 180 degrees.  In order to reduce the unit gain point of the servo I have reduced the gain by a factor of 3 on both X and Y chiller servo loops.  I then put the lasers into DOWN guardian state and then back into LASER_UP.  I will leave them to run and we can look to see if this fixes the problem.