After the whitening gain tune up for the BS OL QPD this week, we tried using the OL damping, but found that the performance was underwhelming.

Using the HSTS BS SS model from the SUS group in the noise budget directory, I made some filters to give us gain in the 0.1-3 Hz band and some low pass filtering above 10 Hz. Also some notches for the bounce/roll modes, but since I couldn't find the measured frequencies, I just copied from LLO.

The attached Bode plot shows the modeled pitch and yaw loop shapes. The yaw one is a little less stable due to the different yaw eigenfrequencies - I just used simple poles/zeros rather than try plant inversion. Rather than fine tune the OL loops we'd be better off tweaking the BOSEM damping loop for yaw to reduce the 2.1 Hz Q. I also chose to have the lower UGF be above 0.1 Hz to prevent too much interaction with the slow WFS loops. The 0.2 Hz bump in the yaw loop is an optional 'MSboost' filter that I've added to squash a 0.15-0.25 Hz microseismic peak. I would leave it on all the time, but we could have a few of these for different microseismic states. The 4th attachment shows the step and impulses of the loops with MSboost ON. The top left plot shows the step response of the pitch loop. The bottom left the step response of the yaw loop. The plots on the right show the responses of the pitch and yaw loops to step and impulse applied to the M2_L DoF. The impulse amplitude is sized to be close to what the lock acquisition transient is in force, but probably too long in duration.

I will show the before and after spectra, but the winds started audibly shaking the control room an hour ago, making it kind of an unfair comparison.  The 0.1-1 Hz ground noise has not changed much, but the 2-30 Hz noise is up by a factor of ~10.

A piece of good news: the broadband noise seen earlier this week in the yaw error signal is gone. The broadband noise in both pitch and yaw is now 1e-10 rad/rHz above 2 Hz. The huge peak at 20 Hz in the pitch spectrum is probably the flagpole mode of the cerulean pillar.

For all OL, we should add the OL feedback signals (OLDAMP_{P/Y}_OUT)to the DAQ and remove the OLDAMP_IN2 channels (not needed since we already record M3_OPLEV_OUT). 

These will be 256 Hz channels, so impact on frame size is minimal.