There is a fix for early design's that is missing a diode, I think it was.
I have the exact same problem with one of my P-Code cards. One has the diode, the other does not. The one with, works, and the one without does not.
I have both an ATA5282 (non-working) and an ATA2383. I was going to provide pictures, but I don't feel like disassembling both again.
ATA5282 had a 9-pin SIP resistor network soldered to the B side of the 74LS245 (pulling each line about 480 ohms to + 5 volts). The ATA2383 had an 74HC245 in place of the 74LS245 and no SIP.
Now it is impossible to tell if the late 1982 one was repaired or if it was produced this way, but it struck me as odd. There is not a single diode any place on either PCB, so I guess I'm wondering "where" this missing one is supposed to be.
I had no tools beyond visual inspection and basic continuity in trying to diagnose the ATA5282 card until you provided the diagnostic disks in a later posting.
Before unsoldering the SIP/74LS245 cluster, the Mini-Memory, Test #1 would fail both ROMs and all GROMs. Once I had a socket in place, I put in an 74HC245 to match the ATA2383 PCB and re-ran the diagnostics. It now passes ROM 1 (4732), but still fails on ROM 2 and all GROMs.
I'm ordering an 8Kx8 prom (no 4764 direct equivalent) and when it arrives will depopulate ROM 2, re-socket and try again. Looking at http://www.mainbyte....p_code_card.jpg and all the common data lines affected, I'm really hoping it's ROM 2 and not any/all the GROMs.
I will keep the forum posted.