Figure 1 and 2 show the TFs from top right actuator (seen from AR side, H1 for MCI and MCO, H4 for MCE) to pitch and yaw OpLev signals, respectively.
Figure 3 and 4 show the TFs from bottom right actuator (seen from AR side, H2 for MCI and MCO, H3 for MCE) to pitch and yaw OpLev signals, respectively.
Figure 5 and 6 show the TFs from bottom left actuator (seen from AR side, H3 for MCI and MCO, H2 for MCE) to pitch and yaw OpLev signals, respectively.
Figure 7 and 8 show the TFs from top left actuator (seen from AR side, H4 for MCI and MCO, H1 for MCE) to pitch and yaw OpLev signals, respectively.
Figure 9 and 10 show the TFs from pitch and yaw excitation to  pitch and yaw OpLev signals, respectively.

I'm not sure which TFs are good ones but at least TFs of MCO has several times smaller DC gains and larger coupling.
Also, MCO pitch to pitch TFs have very small Q resonance compared to the other suspensions'.

I and Takahashi-san discussed these results and conclded that MCO might touch somewhere.
So, we investigated MCO today (klog20385t).

Note:
The automesurement data are stored at /kagra/Dropbpx/Measurement/VIS/PLANT/{MCI,MCE,MCO}/2022/04/PLANT_{MCI,MCE,MCO}_STANDBY_TM_{TEST,COILOUTF}_{L,P.Y,H1,H2,H3,H4}_20220405****.xml (**** is the starting time of the measurement in JST).

I checked the measured transfer functions. For MCI and MCE, there are no differences from the reference (Acceptance check document for input suspension). For MCO,  the transfer functions should be measured again after the recovery yesterday.

According to the values of OpLev calibration factors (fig1), the gains of each suspension TF should be changed from the gains at acceptance check as follows.
MCI yaw:  ~ 1
MCI pitch: ~ 18
MCE yaw: ~ 0.97
MCE pitch: ~ 4.2
MCO yaw: ~ 1
MCO pitch ~ 1
Note that MCO calibration factor was changed not only at OPLEV_TILT_{HOR,VER}_GAIN but also OPLEV2EUL matrix as like fig2.
These changes can explan the gain difference between new measurements and the old ones.