Be wary of the puny little logic power supplies in these drives. I had one nicknamed the Lazarus drive. It was a spare that would go dead and then return to life sometimes after a power cycle. At one time, the +24v connections for all of this group of PF4 drives were all tied together and ran through a SLC DC output card. All the 0vdc connections on the drives were together too. So with all the drives logic power supplies in parallel (I think we had 7 of them) supplying DC power for that transistor output card, and the signals coming from the card back to the inputs on the drive, it just seemed like a bad idea... That output card also signaled an indicator lamp or two if I recall correctly.
We had all sorts of weirdness going on similar to what you report. They all had 22-comm-e cards and problems with MSGs failing too. I got out the prints, noted the abnormal design. After seeing the crew fight this a couple of more times, I changed the wiring, leaving all the drives' 0VDC terminals connected together and connected them to a separate 10amp power supply common, and used it as the output card power source for the hardwired signals to the drives. This stopped the random and multiple failures they were having, but that one Lazarus drive we kept around for entertainment and as an unreliable emergency backup.