> Ratio between PRM and BS with PRCL is more than 10% different from the past. I'm suspecting that the modeling of feed-forward effect may be inadequate,
Difference in the actuator efficiency ratio between PRM and BS as ~10% cannot be explain the effect of FF.
So there seems to be an another reason.
Re-measurement with PRX might be necessary though I have no good idea about a difference between PRX and (PR)PRMI.

-----
In PRMI (and PRFPMI), MICH and PRCL loops have correlation via BS longitudinal motion as shown in Fig.5 of klog#24980. Actuator efficiency of PRM is estimated based on this fact by using an efficiency ration between BS and PRM. In the case with MICH2PRCL FF, control loops become slightly complicated as shown in Fig.1. We measured actuator efficiency of PRM with FF in klog#33609 and actuator efficiency ratio between PRM and BS was ~27.2. This value is ~11% different from past value (~24.5) in klog#29974 and klog#30091 which is measured with PRX lock. (Because BS and PRM are always in room temperature, I have no good idea about a ~10% change.)

I wondered that this difference comes from the effect of FF and compared the actuator efficiency ratio in PRFPMI with and without FF (measurement files are saved in /users/CAL/current/measurements/2025/0503/. With FF, estimated actuator efficiency ratio is reproduced as ~27.2 (see Fig.2-3). Without FF, it shows ~27.5 which is almost same as a result with FF (see Fig.4-5).​​​​​ So our measurement doesn't seem to be affected by FF.

By the way, as I checked latest foton file, a servo gain is set as 0.03875 in LSC-MICHFF2 which corresponds to ~25.8 as an actuator efficiency ratio between PRM and BS (see also Fig.6). According to the filter name (FFgain231226), it's decided by a measurement in klog#28094. On this klog, a measurement was done as follows.
> 6. Calculate the ratio between MICH2PRCL to PRCLout2PRCL (optimized MICH2PRCL FF).
If "PRCLout" means LSC-PRCL2_OUT, only MICH2PRCL is suppressed by PRCL OLTF though PRCLout2PRCL isn't so because LSC-PRCL2_OUT is in-loop TP. Was this effect corrected? (or negligible due to use only much higher frequency than PRCL UGF?) I couldn't find measurement file so I couldn't check.