PRFPMI lock acquisition is not stable today even though PRFPMI alignment seems good.
The reason is PR3 alignment, which reduces fringes of POP PNC and makes it unstable.
After initial alignment of GRX and IRX, the fringes were recovered and PNC became stable, so it seems necessary to perform initial alignment when POP PNC becomes unstable.
In this aternoon, PRFPMI lock acquisition was not stable.
I thought it was due to just large seismic motion but it was not true: the reason is that the fringe of POP PNC was too small (fig1: peak2peak value is about 100 cnts) even though X/Y transmission with PRFPMI is large (about 130).
So, I performed initial alignment of GRX and IRX.
After the initial alignment, POP PNC fringes recovered (fig2: peak2peak value is about 300 cnts) and PNC can be locked stably.
Then, PRFPMI lock acquisition becomes much stable (it is still failed sometimes maybe due to large seismic motion, though)
So, it seems better to perform initial alignment if POP PNC fringes become small and the PNC lock becomes unstable.