This is a report on the continued work trying to resolve the issue with the MCo actuators (3295, 3600).
Conclusions:
- The electronics seem fine.
- The mechanical response, in particular the restoring coefficients, are messed up for unknown reason.
Therefore visual inspection must be performed.
[Background]
As I reported on 3295 and 3600, the DC responses vary a lot between the coils as seen by the oplev. In particular, the pitch responses varied by a factor of four at most. This strangeness had been known for a while.
A goal of the day is to localize the issue. Especially, I wanted to know whether this issue is from the electronics or the suspension mechanics. So I made a set of different measurements today to answer the question.
[The measurement]
The idea of the measurement is that we simply drive each coil at a time, but at AC frequencies, instead of DC. This should simplify the equation of motion for the angular degrees of freedom as
-I ω2 θ = τ
where tau is the torque that we apply to each coil. The angle(s) of the optic, theta, is monitored by the oplev.
I reset the compensation gains (3295) at the very output of the digital system to the default combination of 1 and -1 to reduce possible confusions during all the measurements.
Assuming that the placement of the magnets were done with a good precision, the angle theta should give us an idea of whether the applied forces are the same amount or not between the coil-magnet actuators. In this way, one can check difference in the applied force or equivalently the whole actuator electronics.
[The results]
The results are shown above.
Even though I don't understand the parasitic resonance at 13-ish Hz in the pitch response, the responses among different coils are identical within tens of %.
This directly means that the applied torques are the same among different coils within tens of %, which is good. As described in (3600), this indicates that the restoring coefficients were somehow messed up. The restoring coefficients must be affected presumably by some mechanical parts. The most straightforward way to check this hypothesis is, of course, visual inspection.
