MODCOMP II Revisited...

My excitement over getting my MODCOMP II running in June (reference post attached) - was somewhat "premature". The MODCOMP II performed well - and then started having intermittent failures. I observed that many of the failures seemed to be the result of badly seated chips. I thought that a bit strange, since I had checked them all as part of the restoration process. As time went on, the system became even more dysfunctional - until it wouldn't run even the most rudimentary programs. I checked with a former MODCOMP hardware support person - and he asked: "What brand were the failing chips you've found so far?". I hadn't even thought about that. He said "Let me know after you check - and I wouldn't be surprised if they were all TI (Texas Instruments)". I did check - and virtually 100% were TI chips. Factoid: Many TI chips produced during 1975-77 period had very thin "legs" and poor tinning. When they were wave soldered, the solder offered them protection. However, when they were socketed (as in the MODCOMP) they had very little protection from the environment. They would rust and eventually fail. He suggested that I just replace all the TI chips in the MODCOMP if I wanted a reliable system. I decided that was too "radical". So armed with a logic analyzer and scope, I began to debug the MODCOMP - one chip at a time. After finding 20 or 30 bugs - ALL of which were TI chips - I finally "relented". I have now replaced about one hundred and fifty TI chips in the Modcomp's CPU and memory planes. I'm about 3/4 of the way through the process - and the MODCOMP is becoming very reliable. It will run basic memory diagnostics by the hour without failure. Final thoughts: The "badly manufactured" TI chips turned out to be a lot more fragile than I had originally supposed. When I removed many of them from their socket, their legs would simply fall off. It wasn't unusual to have two or three legs fall off one chip! Some were incredibly rusty over their entire surface - which wasn't at all obvious looking at them from the top. The MODCOMP sockets were "good news - bad news". Bad, because the sockets exacerbated the TI chip manufacturing issue - and good, because they made the chips easier to replace! Regards, Lyle ---------- Forwarded Message ---------- Subject: MODCOMP II Rescue Revisited... Date: Monday 04 June 2007 21:06 From: Lyle Bickley <lbickley at bickleywest.com> To: "General Discussion: On-Topic Posts Only" <cctech at classiccmp.org> A number of you probably remember me "announcing" in Jan, 2007 that I had rescued a MODCOMP II/12 from a lab here in Silicon Valley. The link for that is below (sorry, it's slow - the pictures are large). http://bickleywest.com/modcomp.htm After a lot of work cleaning up both the CPU and I/O chassis I began checking out the system. It turned out that the I/O chassis was in pretty bad shape. I went back to the lab facility where I found the MODCOMP II - looked through several more buildings - and found another lone MODCOMP II I/O chassis. I went through the "salvage" process again - and finally picked it up about a month ago. It was in a lot better shape than the original I/O unit. And the really great news - it was an identical configuration to my original I/O chassis. After cleaning it up, doing all the usual capacitor, power supply checking - I powered it up - and everything "looked" good - and no bad "cooking" smells ;-) I cabled up the CPU and I/O unit, powered 'em up - and to my amazement the front panel controls seemed to work O.K. I then did the usual hand memory tests - and core memory worked O.K. every location I tested. I then wrote a bunch of small diagnostics (in machine code) - and found that I/O was not easy to code - but I did enough to check that the console in/out ports worked O.K. Today I loaded diagnostic monitors, and diags - and most of the system - including I/O is working! The really good news is that all 64K (words) of core memory passed the long manufacturing memory diagnostic - which even tests for "hot cores" failing. It's been months in the making this critter come alive - but it sure feels great when the diagnostics tell you that you are on the right track :-) Cheers, Lyle ------------------------------------------------------- -- Lyle Bickley Bickley Consulting West Inc. Mountain View, CA http://bickleywest.com "Black holes are where God is dividing by zero"