This afternoon we had 2 more locklosses due to the instability at 1.88 Hz which we have struggled with for a few weeks.  (31707) 

This time Keita and I looked at the filters for the soft loops, and saw that we have some kind of plant inversion from a long time ago that has a complex pole at 1.83 Hz, which was giving us 15bB of gain there unnecessarily. (reminder, the soft loops generally have bandwidths of a few hundred mHz, except for CSOFT P which needs to have gain at 0.5Hz to supress our dPdtheta instability).

We turned off these invY filters, and saw that the noise at 1.88Hz in several of our YAW ASC loops was reduced.  For the soft yaw loops, the peak at 1.88Hz was reduced by about 20dB (in the attachment, dashed lines are from last night's undisturbed lock stretch, the solid lines are from the last lock where the invY filters were off.)  There is still clearly extra noise at 1.88 Hz, so something else may be nearly unstable.