J. Kissel, R. Kumar, A. Pele

Team SUS is on the hunt for understanding what / why 3.4 Hz is such an issue in the IFO. Today, we began careful characterization the top mass control of as many QUAD R0 reaction chains as we could. Rahul's posted his preliminary aLOG showing what he managed to gather in LHO:68398. We gathered data in different configurations of control of the top mass. The sets of loops that control the reaction chain top mass are
    (1) R0 DAMP "top mass damping loops" from the R0 OSEM sensors to the Reaction Mass TOP (R0) actuators, in L, T, V, R, P ,Y
    (2) L2 WIT DAMP "R0 tracking" control from the PUM (L2) OSEM sensors to the Reaction mass TOP (R0) actuators in L, P, and R.
namely, 
    (a) All damping OFF
    (b) Only R0 damping ON (all DOFs on)
    (c) Both L2 WIT R0 tracking ON and R0 damping ON (all L, P, and R ON for the former, all DOFs on for the latter)

Once finished gathering these, I also moved on to tuning open loop gain transfer function templates. Unfortunately, I ran out of time before I could take open loop gain TFs with LIT WIT "R0 Tracking" ON vs. OFF, but even the R0 DAMP open loop gain TFs alone are informative.

Also of note, the ETMY reaction chaing has *no* position / alignment offsets in place.

I'll add the locations of these templates in the comments, but I'll bring some conclusions immediately forward, with accompanying attached plots. Note, in all the plots, I've tried to highlight where 3.4 Hz is with a dashed magenta vertical cursor. In open loop gain TF plots, I've used dashed black vertical cursor to indicate the lowest lower unity gain crossing and highest upper unity gain frequency crossing, to show that there's plenty phase margin when there's only R0 to R0 damping on.

    (i) The R0 DAMP YAW top-to-top damping loop DOF has had its "normY" = zpk([],[],43.478,"n") gain of 43.5 off for years, and thus has been *way* underdamped for a long time. One can see this in the R0 Y 2 R0 Y open loop gain TF, where there is no UGF crossings. Turning this filter ON would improve the open loop gain to match all 3 other QUAD reaction chains, which I believe are sufficiently and not aggressively damped.

    (ii) The R0 DAMP PITCH loop is *way* overdamped, which can be seen in the R0 P to R0 P closed-loop plant TF. This seems to be consistent across all QUAD reaction chains, as the same design and gain is used everywhere.

    (iii) The R0 DAMP VERT loop is underdamped. There's only a small tiny region where there is R0 V to R0 V open loop gain TF's magnitude above unity. This seems to be consistent across all QUAD reaction chains, as the same design and gain is used everywhere.

    (iv) While the L2 WIT "R0 tracking" drastically alters the R0 L to R0 L closed-loop plant transfer function, it really only changes the R0 P to R0 P between 3 and 3.5 Hz, in a frequency dependent way.

    (v) To R0 DAMP ROLL loop is reasonably damped as shown by the R0 R to R0 R plant TF, by it's getting all of it's damping from the transverse DOF (see R0 T to R0 T open loop gain TF), since there's very little "diagonal" R0 R to R0 R open loop gain.

From (i) turning on the R0 DAMP Y normY filter is an easy fix. We should just turn this filter back on. I've already confirmed that the loop is stable with with and without L2 WIT R0 tracking ON.

We should fix (ii), (iii), and (iv) with a proper frequency-dependent loop shape change (something like a Level 2 design) that works for all Reaction masses, since the issues seem to be systemic of this very old, poorly commissioned design. but if someone really felt like it, they could "just" adjust the gain of the loop to get more loop gain. 

(v) is not necessarily an issue, but the Roll and Transverse DOFs have resonances right on the 3.4 Hz frequency of interest, so it's worth further exploration and taking the next step to see how these loops are influenced by the L2 WIT R and L2 WIT P to R0 drive loops.

Next Tuesday!