Today Sheila and Thomas found that SRM top coils (probably vertical damping) rang up on their own when SRM was misaligned (first attachment, circled in red), but after they flipped the (mis)alignment offset it was OK (circled in green).

While SRM was misaligned with a flipped offset and was damped OK, I sent +-20000 counts signal into T1, T2 and T3 coil one by one and measured the response of each OSEM (second attachment).  I could have tested with a larger signal but I didn't want to disturb PRMI locking.

Clearly T1 and T3 responded mostly linearly but T2 coil seems to be too close to the edge. Note that in an ideal case everything should be linear and that T2 and T3 should show the exact same response in opposite sign.

For T3, the inclination of the plot (i.e. OSEM sensitivity) is about 2.6e-3um/ct for the entire range tested.

For T2, this is about -2.4e-3um/ct at  OSEMINF~63um, which sounds OK, but at the opposite end (OSEMINF~150um) this is -5.2e-4um/ct, that's a factor of 5 smaller than T3. If I extrapolate using quadratic fit, it seems as if the sensitivity becomes 0 at OSEMINF~155um or so.

As T2 goes closer to open it loses sensitivity and thus V, P and R damping all start to couple with each other strongly. My guess is that under that condition, when there is a large transient the damping loops go crazy because of the strong coupling and a large nonlinearity of T2 OSEM. Unfortunately as we "align" SRM it goes closer to the bad side.

From the first plot, when SRM is misaligned with flipped offset (i.e. now) T2 OSEMINF ~ 120um, when "aligned"  it should be about ~132um, when SRM started oscillating it was ~138um, though there was a time when it didn't oscillate at ~145um.