Enomoto-kun was playing with MC common gain and we saw some improvement on MICH sensitivity at high frequency above 500Hz.

I plotted MICH sensitivity  with PD power of MC transmitted light with a simple conversion factor of 1.2e-10. I also plotted a coherence between them.

It seems to be that the MICH sensitivity is limited by some amplitude noise at lease from MC or upstream, but many of peaks are added by something after MC, which is probably clipped light at view ports or the limit of size of mirror.