We started centering OSEMs. We've retracted each OSEM as much as we could until the OSEM reading didn't change any more (then pressed the OPO table against the EQ stops in one direction to make the magnet flag go further out and made sure that OSEM reading didn't change), and recorded the "open" reading. These were:

We weren't able to do this for H1 and H3 as it's almost impossible to do without removing counterweights from the suspension.

Compared with previous measurements in Feb 2018 (alog 40727) where the ADC was basically railing when fully open, it seems that something changed and we have much smaller numbers. LED degradation?

After 4 of 6 OSEMs were centered I took TFs, and they look fine.

All translational DOFs look pretty good and similar to O3 OPO (1st attachment).

Roll and PIT frequencies shifted, which is not a surprise given the change in the moment of inertia due to different mass distribution on the table (2nd attachment). Roll is now 2.16Hz (was 2.42Hz), P 1.69 (was 1.63). Yaw didn't change much if at all (1.41, was 1.43).

Roll TF looked qustionable at low-coherence frequency especially at around 1.73 Hz, so I took a finer resolution TF at narrower frequency range with larger excitation w/o railing DAC and it looks fine (3rd attachment).

Measurement files are saved as

/ligo/svncommon/SusSVN/sus/trunk/OPOS/H1/OPO/SAGM1/Data/2021-09-18_0950_H1SUSOPO_M1_WhiteNoise_[LPRTVY]_0p2to50Hz.xml

/ligo/svncommon/SusSVN/sus/trunk/OPOS/H1/OPO/SAGM1/Data/2021-09-18_0950_H1SUSOPO_M1_WhiteNoise_R_0p2to50Hz_finer.xml .

I trended OSEMINF INMON values from the start of O3 until the HAM6 vent on Nov 2020 to see if there have been any signs of degradation and it didn't help.

For all V OSEMs, OSEMINF INMON increased significantly, which either means that the OPO floated more (unlikely) or LED drive increased over the course of 2 years and 7 months. Though I cannot decouple OSEMINF from the position/alignment of the OPOS, the behavior of V OSEMs cannot be explained by the mechanical position/alignment change. There was no alignment offset nor alignment feedback to OPOS as far as I can tell.

H1 and H2 decreased from ~16000 cts to ~11000 (H1) and ~8300 (H3) while H2 went down and then back up to ~14000 or so.

FYI, below is the comparison of Now VS right before the Vent.

Via discussions and email (J.Kissel, K.Kawabe, S.Dwyer, R. Kumar, T.Shaffer, S.Aston, B.Weaver), a decision was made that we will leave the OSEMs and their calibration as is and not take open light readings on remaining 2 OSEMSs. We should be aware that if this decay is real it could be present on other AOSEMS.  Points made are below - paraphrased from Jeff:

• It is weird that the OSEM’s open light current had decayed so much. It’s a new SatAmp to Chamber cable, and new (longer) Rack to SatAmp cable, but the chassis themselves are the same
• The tests of “open light” being should be good enough, we question that they weren’t totally "open light,” as that maybe more likely than the AOSEMs LED’s decaying by up to *half.*
• If we accept that if the LEDs *have* decayed this much, it will “only” change the calibration of the OSEMs. This will cause systematic error in the readout of how much displacement is there for these few OSEMs, which the damping loops to errantly “think” that there's cross-coupling between degrees of freedom when there’s actually not. 
• As this is the OPO — the requirements on perfectly diagonal damping loops are relatively non-existent, the transfer functions look fine (the magnitude of each DOF has not dropped significantly, and the control filters are simple enough that the changes we see won’t impact the stability or the amount of Q reduction — i.e. the loop design is robust against the plant changes we’ve seen), and the damping loop Qs still get reduced “sufficiently.”
• Remaining two OSEMs are tricky to measure open light. We not not want to change the balancing or struggle to move them and knock something else.