I did a quick exploration to check if it is worth increasing the damping gain of the SR3 controllers to get better RMS performance now that we have installed LPY OSEM estimators. The exploration, math and results are summarized in T2500367. 

The main takeaway is that we might want to increase the Longitudinal and Pitch gains inside the EST blocks in order to try to reduce the pitch RMS for SR3. The pitch RMS is dominated by the 0.6-0.7 Hz resonances cross-coupling from longitudinal.

The suggested change is to increase the estimator damping gain from -0.8 up to -3 for both longitudinal and pitch. This change will retain 50% of the old OSEM-only damping OSEM noise contribution. We don't recommend going beyond -3 for fear of making any controllers unstable.

If we wanted to implement a change this change, I think it would be wise to monitor witness signals to see how they change as we increase the gain, and ensure everything is working OK. 

I attach a few plots to show the calculated tradeoffs and noise budgets for the pitch degree of freedom of the bottom mass of SR3. The plots correspond to OSEM-only damping, so they don't reflect the reality of the estimator. Nevertheless, the important thing is that they show that we stand to benefit from increasing the pitch damping gain and squashing the 0.6-0.7 Hz resonances [1st figure] to reduce the M3 pitch RMS [second figure]. We couldn't do that with OSEM-only damping because the OSEM noise would eat us alive [3rd figure]. Now that we have estimators, we should increase the estimator-path gain and see what we can get away with. The details are in T2500367.