We found IMC could not restore the lock automatically with IO guardian after the lock was lost at once. We investigated and found the reason because the amplitude of the error signal (K1:IMC-SERVO_OUT2_OUT_DQ) (Fig.1) got about 1/4 smaller than the one of 1.5 days ago (Fig.2). 

We assumed that the EO coefficient (r13) of the EOM crystal (MgSLT) in the perpendicular direction against the direction of applying electric field is about 1/4 smaller than the one (r33) in the direction of electic field according to the link (the EO coeffcient seems to be measured with 633nm beam. I'm not sure of the same ratio as the case of 1064 nm beam).  So we rotated the polarization of input beam into EOM to almost 90 deg., that is we changed  from near s-pol to near p-pol during yesterday work (klog30531) and the modulation index got smaller in current polarization of input beam of EOM.

Therefore, s-pol may be suitable with the polarization for EOMs.

Tomorrow, we will continue the RFAM work. In that time we will fix the issue.