We're still working to understand why we've got problems, but we've at least found *an* alignment of SR3 and SR2 that seems to prevent clipping at the AS port when we're centered on AS_C.   To get here, I had to move SR3 and SR2 by very large amounts, much more than they ever drift.  In order to more accurately see changes in power levels on the AS WFS, I had the DC centering loops engaged in this single bounce off of ITMY configuration.

In the attached plot, the bottom row is the slider values for SR2 yaw and SR3 yaw (so, in microradians).  As a reminder, folks have been checking all day and there is no indication from suspension sliders, OSEMs, or where applicable oplevs, that any of our optics have moved nearly this much, so this should not have been necessary.  That said, we've often found that (normally at least) we have lots of leeway clipping-wise when chosing SR3 and SR2 positions.  There's no reason a priori that I can think of that would prevent us from locking with these SR2 and SR3 positions, but TJ and Camilla point out that having moved SR3 this much may make it challenging to also have the HWS paths aligned. These moves are basically my moving SR3, then moving SR2 to re-center on AS_C.

The top row of the attached plot is the centering on AS_C.  So, we would like to only evaluate the amount of power on AS_A or AS_C (middle row) when the beam is centered on AS_C.  It turns out that it's the low-ish part of the power curves that are the values when AS_C is centered. 

I have highlighted using a blue line on the AS_A curve (middle row right side) roughly the trend.  When we started (at about -42 mins) the power on AS_A when AS_C is centered is at about 1.64 units on this y-axis, and as we move SR3 and SR2 in yaw, I can increase the power on AS_A to about 1.85 units.  Jennie found that a time of equivalent single bounce configuration from a week ago, we had 1.87 units on this y-axis.  (This y-axis is just AS_A_NSUM * 0.001).   I didn't take the time to go to the 'other side', but the trend of power on AS_A seems to show that we're into the plateau region, and at the same time the AS AIR camera looked much more normal and unclipped.

EDIT to note that part of the reason it's helpful to wait to evaluate AS_A power until AS_C was centered, is that that also gave time for the DC centering loops to catch up to my big SR3 moves.  I think that the reason AS_A sometimes is higher than the blue marker curve, is that the beam was clipping on AS_A while the DC centering loops were catching up.