J. Kissel

With the TMS in it's current rubbing state, it's difficult to properly model any damping loop design based on the measured transfer functions -- and as you know, we don't yet have a dynamical model that we trust to design on instead. But, I did so anyways, and discovered that the loops really just needed their gains adjusted. Some degrees of freedom had no open loop gain above unity, others had way way way to much open loop gain, and were just down-right unstable.

Here're the changes to the damping loops:
(1) Included a "from_um" or "from_urad," which takes out the "to_um" calibration of the sensors, so we can stick with the otherwise-normalized filter design. These are in FM5, like every other type of the suspension (which is usually poorly named "norm").
(2) Turned the ELF10 elliptic filter back on, as was the original design intent. 
(3) Changed the gains from [L,T,V,R,P,Y] = [-20.0; -7.0; -10.0; -20.0; -5.0; -5.0] to [-20.0; -10.0; -20.0; -0.05; -0.5; -0.1]. (These new gains are also now roughly consistent with the ratios between DC magnitude (i.e. the stiffness) what's seen in the transfer functions).

I attach the modeled open loop gain and suppression for each degree of freedom with the new gains I've installed. Note, I intentionally removed the phase subplot because the data is so noisy / incoherent, I couldn't trust anything regarding phase margins, and flew by the seat of the modeled closed loop gain and suppression magnitude to help me determine where the stability regions lie.

If you're interested in using the scripts to make these plots in the future, they live in
${SusSVN}/sus/trunk/TMTS/Common/FilterDesign/
design_damping_L_TMTS_20130313.m
design_damping_P_TMTS_20130313.m
design_damping_R_TMTS_20130313.m
design_damping_T_TMTS_20130313.m
design_damping_V_TMTS_20130313.m
design_damping_Y_TMTS_20130313.m