With Terrence.

We moved both oplev QPDs. In the aligned position of the optic the oplev beams were not close to the centre of the QPDs with even one output very close to the saturation (92 out of 100 in normalized displacement units).

The oplev readout is used as a reference for the position of the optic for the main interferometer alignment. In order to reproduce the appropriate orientation we took note of the amont of DC actuation the marionette requires to keep the optic aligned. The procedure was the following:

1. In the ALIGNED state of the suspension we took note of the amount of DC control actuation applied on the IM. It was 344 counts for pitch and -175 counts for yaw.
2. We set the actuation to exactly those values at the TEST channels. This automatically off-loaded the DC control actuation signal. We used a 10 seconds ramp.
3. We set the suspension to IM_DAMP_ENGAGED state.
4. In order to avoind tripping we modified the oplev watchdog trigger values: we increased upper RMS threshold to 10,000 counts (later changed to an even higher value) and lowered the lower SUM threshold to 0 counts. Later we set the RMS threshold.
5. We removed the aluminium wind shield.
6. We moved the QPDs.
7. We put back the wind shield.
8. We defined new oplev set points using the ALL_OPLEV medm screen.
9. We set the suspension to ALIGNED state.
10. We off-loaded the actuation in TEST blocks (10 seconds ramp), which automatically increased the DC actuation signal.
11. Doing the diagonalization again was not needed.