IM4 was scanned in YAW while the DC SUM of various sensors were recorded, and it's apparent that there's a clipping.
In the attached, red, blue and green represent normalized DC SUM of the POP_A, POP_B and AS_C respectively plotted against IM4 YAW slider offset. Blue vertical line at -8520.7 shows the alignment we were at this morning.
There appears to be a hard edge on the left side on the plot that is common to the sled and the AS_C, and I guess that's the baffle in front of the PR2.
We're already very close to the edge. From the red and blue data at the blue vertical line, the single trip loss is either 0.3% (red) or 0.7% (blue), and you need to double the number for a round trip loss.
If we go somewhat to the left on the plot, you're already dead. I propose to move somewhat to the right.
The fact that AS_C shows somewhat different pattern means that there's some other clipping going on for the SRC path, on top of the PR2 baffle.
In the plot, Relative power=1 simply means that the measured power was the highest.
What was done:
Using straight shot while PRM, SRM, ITMY and ETMs are misaligned, center the POP sled and AS_C using picos.
Record QPD sums.
Move IM4.
Recenter sled QPDs using picos.
Recenter AS_C using ITMX.
Repeat.
It turns out that centering is important. For some reason, some quadrants put out bigger numbers than the others at 2%-ish level, so I centered them very well (both PIT and YAW less than 0.05 for the sled, and best effort for AS_C) to mitigate the effect.
Did somebody buy cheap 1% resistors at local Radio Shack?
I don't have any stamina left to do picomotor centering for PIT scan.