I saw no sign of real work in BS M2 coil balancing, I had some time, no SUS person was around, so I started doing it manually.
Coil gains before/after the adjustment:
| Before | After | |
| H1:SUS-BS_M2_COILOUTF_UL_GAIN | 1.23 | 1.01 |
| H1:SUS-BS_M2_COILOUTF_LL_GAIN | -1.23 |
-0.985 |
| H1:SUS-BS_M2_COILOUTF_UR_GAIN | -1.23 | -1.015 |
| H1:SUS-BS_M2_COILOUTF_UR_GAIN | 1.23 | 0.99 |
Seems like the balance was not that bad anyway. I removed funny overall factor of 1.23 as I and Jeff Kissel agreed last week.
What was done:
H1:SUS-BS_LKIN settings and filters were done newly by me as the existing things didn't make sense. You need to excite pringles mode, S/N is poor, so you want to bring the frequency low-ish to make the motion larger, and the integration time should be very long.
100000 (somewhat smaller than 132k counts)
100 (doesn't matter as far as P and Y
SINGAIN and COSGAIN are the same).
30 (adjusted to
maximize I phase)
0.01Hz elliptic LPF
H1:SUS-BS_LKIN_Y
0.01Hz elliptic LPF
The above LKIN2OSEM matrix will excite the pringle mode, and the idea is to read the OPLEV signal after the lock in and minimize the output. Note that the optic moves in YAW when the're is a PIT imbalance because of the pringle excitation. When you see YAW signal you change PIT balance etc.
The first plot show you what S/N to expect. Left is before the adjustment, right is after.
The second plot is the time series of the lockin output during the adjustment. The first big jump at about t=-50min is the start of the excitation, and I left it for some time, made the first adjustment at t=-40, left it for some time and made the second adjustment at t=-35 min.
It's nice if somebody automates this.