Since we replaced REFL PDA2 (klog32601), I performed phasing of RF135 and tune the OLTF of MICH and PRCL.
After the installation, I measure the dark offset and subtracted them.

Then, I tried to lock PRMI with 3F signals by just changing the gain, and it was succeeded.
So, I performed the phasing by injecting 94.125 Hz signals from PRCL1_EXC (fig1).
Red/blue lines show the excited signals after tuning and pink/cyan lines shows those before tuning.
Since the peak can be seen only in I-phase signals after tuning, the new phase (75.52 degs) seems good (fig2).

After that, I measured OLTF of MICH (fig3) and PRCL (fig4).
To obtain the similar OLTF gain, I set a gain of 4 at INMTRX for MICH, which is implemented into VERTEX guardian.
UGFs are 15 Hz and 60 Hz for MICH and PRCL, respectively.

After that, I repearted similar test with klog32587.
The result is 15 success out of 20 trials, which is 75%.
So, PRMI lock probablity seems significantly improved thanks to the REFL PD replacement.

Note:

I measured REFL51 signals by injecting 94.125Hz signals from MICH1_EXC to check the phasing of RF51 (fig5).
Though phasing is not good for MICH signals but I left it as it is because we have an experience that RF51 phasing should be done to minimize glitches when ARM is flashing.