I and Daniel measured the WFS sensing matrix by a very slow (0.1Hz) excitation. This is because POS signal is much smaller (it's naturally smaller for a given mirror angle, and WFS placement is unfortunately such that both WFSA and WFSB are more sensitive to ANG).
Originally before the beam path change, I think Bram and Arberto measured the sensing matrix by giving a large offset to either ANG or POS. I also briefly tried this, but in order to have anything for POS, the offset needed to be so huge that the beam almost comes inside one segment, so I decided that this is not a good method.
Anyway, at 0.1Hz I was able to have a very good coherence for both PIT and YAW, POS and ANG. (Data file is /ligo/home/controls/keita.kawabe/OAT_2012/WFS_sens_20120831.xml)
Sensing matrix for PIT was OK:
WFSA = -0.313*POS -5.056*ANG
WFSB = +0.620*POS - 3.930*ANG
Sensing matrix for YAW was hopeless:
WFSA = +0.139*POS + 2.754*ANG
WFSB=+0.091*POS + 3.751*ANG
Thing is, for YAW, the sign of POS signal for WFSA and WFSB are the same, and in order to subtract large ANG signal we also need to subtract a large portion of POS, making the POS signal even smaller.
We think that the problem is that the beam parameters on the table is different from what was assumed when designing the Gouy telescope. We might try something on Tuesday.
In the mean time, I inverted the sensing matrix and put them in the input matrix:
PIT = [-0.9004, 1.1580; -0.1420, -0.0717]
YAW = [14.5307, -11.2063; -0.3703, 0.5656]
We turned PIT on and it seems to be doing fine. Didn't bother to do YAW.