DEC PDP 8/e MM8Ej RAM (8K) Repairs

SHARE	Return to Threads
DEC PDP 8/e MM8Ej RAM (8K) Repairs	by Bill Degnan - 11/21/2022 10:27
PDP 8e with H-19 terminal on the ground, displaying a directory from RK05 diskpack. Click image for larger view. This thread will document the current progress with the PDP 8/e located at Kennett Classic. It's the same system written about earlier here. The current issue is that the system boots from RK05 diskpack 0, but the MEMORY command says only 8K RAM is available, even when more RAM is installed. 8K is not enough for any useful operations on this system. The proper next step is to run the MAINDEC DEC8X diagnostics, which naturally bombs without more RAM in the system. Love the classic chicken and egg DEC troubleshooting problems. There does not seem to be any manual online for this specific board, one has to reference an earlier 4K version (as of this writing). Closest thing I could find is here: https://homepage.cs.uiowa.edu/~dwjones/core/mm8ej/ So, we have to experiment. First, I confirmed that power to the backplane was still OK. Good. Each 8K core memory stack consists of three controller cards. I set out next to confirm that I have the correct EMA jumpers on each core stack G111 controller. I have enough core stacks for 32K. Ethan Dicks suggested that the jumper configs for a 32K set of MM8Ej RAM would probably be: 0-8 IN IN IN; 9-16 IN OUT IN; 17-24 OUT IN IN; 25-32 OUT OUT IN. (testing revealed that he is correct). Next, I verified at least superficially that I had a good first 8K stack by manually testing from the front panel and loading what programs I could from RK05. The OS/8 MEMORY command returned "8K" with just the one core stack. I ruled out (for now) that both my EMA boards were faulty. I have two identical EMA boards and they perform as expected. I next put my attention to the 2nd 8K core stack, thinking it may be faulty someplace. WHile installed in the system, the MEMORY command still returned 8K. I decided to switch cards from other stacks to see if I could identify the faulty component(s). More to come. Reply
Getting 32K Again	by Bill Degnan - 12/09/2022 10:38
Click image for larger view. Armed with the likely EMA jumper values for an 8K core stack, I set to test every board related to memory so I could weed out the bad board(s). Fortunately I have some spare boards, enough to make 40K. 4 sets are needed for a full 32K. Two bad boards found. For future reference, the good boards are installed in the PDP 8e and the two bad are listed here, for future repairs: G111 -electrical fault or some kind. When installed the board does not communicate with the CPU. HYPOTHESIS - The board seems to be jumpered for the 8-16K space but maybe the jumpers' solder or a related component have a short and thus not able to report as being for the 2nd 8K memory space? A future project. One of the core boards (H212) is dropping bits. If I toggle in from 0000 0000/7001 6046 6041 5002 5000 when I examine address it's ok until you get to 0008, examine 0008 and we don't see 5000. Instead it's: 000 000 000 000 (0000) rather than +0+ 000 000 000 (5000) (where + = light 0 equals light off) So future me, if you're reading this now, I'd guess the G111 is fixable, more so than the actual H212 core board. Reply