While trying to phase AS_B 90MHz signals (hooked up to AS_B_FR45), we noticed that the phase to the signal changes dramatically with the amopunt of power on the quadrant. Attached is a time series of the DC power (Blue. 0 is on top, more power goes negative), as well as I and Q phase. On this plot we first maximized the light on Seg1, then phased all signal into I, then noticed that the phasing changes with power on the segment, then reduced the power into the DRMI in 4 steps. As you can see the phasing dramatically changes with the power on the diode...
Here are the 36 signals at the same time. As the power drops by roughly a factor of 2, the signals also drop by roughly a factor of 2, without any apparent change in the phase.
This confusion was due to dark offsets - the for some reasone changed significantly.