Rana showed up and diagnosed the problem as a railed FSS SLOW output. The SLOW Monitor about was showing ~6V, which is apparently a bad mode-hoppy place for the NPRO. Reducing the SLOW output brought things back into a good range which allowed the PMC to lock again.
In attempting to diagnose the problem I noticed that there is -100 mV DC coming out of the PMC RFPD RF output. This is not good, probably indicating a problem, and was what I thought was the PMC lock issue for a while. Need to look at the PMC RFPD RF output.