Sheila, Haocun We tried with the POP X path with 36MHz to control PR3, and phased each quadrant it to minimize the Q signal, as shown below. We used -1 in the matrix for Pitch. This is fine for Pitch, but in Yaw, there is an offset between the new signal and the REFL signal. So in the afternoon, we changed to use 45MHz demodulation and changed the dark offsets. Then the laser broke down.
Since May 18 4:49 UTC (Tuesday 9:49 PM local time), POPX whitening gain has been "even" (i.e. gain step = dB/3 = even), which is known not to work (alog 26307). You need to use "odd" gain for the moment.
Was it before or after Tuesday 9:49 local time when you observed an offset in 36MHz YAW signal?
Update: According to Haocun the offset was observed on Wednesday local time. The measurement needs to be repeated with odd gain step.