I have some data for how much motion of any PRMI-relevant optic affects the beam seen by the POP QPD. 

For this, I am using the QPD calibration from the micrometer (elog 8851) to get me from counts to mm of motion.  Note that the pitch calibration hasn't been redone (I tried locking the PRMI this afternoon, but ITMX kept drifting away from me**, so I didn't get any more data.) The pitch calibration is obviously very rough, since I only have 2 points defining my fit line. 

Anyhow, if we assume that's close enough to get us started, I now have a calibrated QPD spectrum:

As detailed in elog 8854, I took single frequency transfer functions, to determine the effect at the QPD from shaking any single PRMI optic.  These transfer functions gave me a conversion factor between the optics' oplev readings (in microradians) to the counts seen at the QPD.  I used this number, as well as the QPD calibration from the micrometer data, to convert each optics' oplev spectra to motion that one would expect to see at the QPD. 

I have not yet completely figured out how to make an estimate of the PR folding optics' affect on the POP QPD spot position, if I know their motion.  The current plan is to do as Den did in elog 8451, and infer the PR2/3 motion from the ITMX/BS motion measured by the oplevs.  My plan was to take the spectra of the oplev signals while the BS/ITMX are undamped, divide by the SOS pendulum transfer functions, then multiply by the TT transfer functions (which I finally wrote down in elog 8564).  I'm planning on using the undamped data, since the oplev signals are still within the linear range of the oplev QPDs, and I won't have to take the SUS damping into account.  Anyhow, after I do that, I'll have an idea of how much the tip tilts are moving, but not what that does to the cavity axis.

However, after looking at the plots below, it seems like the PRM is the main culprit causing the PRC axis motion, although the BS (and to a smaller extent the ITMs) are not innocent.  Since the plots get very busy very quickly, I have many plots, each plot comparing one of the above QPD spectra (either pitch or yaw) with a single optics' oplev inferred motion.

EDIT:  After talking with Koji, I realize that, since the ASC was engaged during the PRM oplev spectrum measurement, I cannot yet say whether the motion is due to PRM, or if it is from PR2 or PR3, and imprinted on the PRM via the ASC servo.  The lump where the PRM-caused motion is greater than the QPD spectra is entirely in the region where the ASC is active.  So, the QPD motion I expect without the ASC would be something like the green trace in the PRM comparison plots.  The blue trace is then the closed loop measurement.  Since the ITMs and BS are below the closed loop values, they aren't the ones causing the big lump.  I should retake all of these spectra at a time when the PRMI is locked, but the ASC is not engaged.  I'm not sure if I'll have a chance to do that tonight or not.  If I can find some GPS times when the PRMI was locked, before we had ASC, I can get the oplev data.

PRM:

BS:

ITMX:

ITMY:

 I think part of the reason PRM is dominating is that it's damped motion is ~10x greater than any other optics', most noticeably the BS'.  I'll write a quick separate elog about this.  Also, note that the ~3Hz resonant gain had been turned off in the PRM oplev loop, but not in any other loops.  This is why there isn't the sharp dip in the PRM's oplev motion.  Also, since the PRM ASC was engaged for this measurement, and the ASC pushes on the PRM to minimize the QPD motion, it isn't totally crazy that the PRM's motion is greater than what we actually see at the QPD, if it is compensating for the motion of other optics.

** Re: PRMI locking this afternoon, it was almost as if ITMX were bi-stable.  I aligned both arms, to set the ITM positions.  Then, I would lock and tweak up the michelson to get the AS port nice and dark (usually touching ITMX today, since it seemed like the drifter....ITMX at this point was usually between -7 and -15 microradians in pitch from the center of the oplev QPD).  When I then brought the PRM back into alignment, ITMX was starting to drift away.  As soon as I hit the LSC Enable switch, and looked back over to the OpLev screen, ITMX was misaligned, usually around -65 urad in pitch.  I did this circus probably 3 or so times before giving up.  Koji said that he had seen this bi-stability before, but he didn't remember what fixed it.  The drifting that Koji mentioned in elog 8801 seems to have been fixed by centering all the PRMI oplevs every day, but I had already done that, and was still seeing ITMX drift.