curiousmarc
patreon
HP 9825 update: still breaking faster than we can repair it!
🕑 Added 2021-04-27 20:51:27 +0000 UTCMy fried HP 9825 restoration is proving to be quite challenging. We know that we lost a bit in our boot ROM while we were debugging the RAM refresh. But since the data bit line is shared between half of the ROM chips, a RAM chip, and a buffer chip, we don't know if the ROM actually lost its bit, or if the bit line is shorted high by any of the other chips.
Carl brought his HP 547A Current Tracer and 546A Current Pulser probes. The combo of the two should allow us to pinpoint the location of a short, even on a complicated net were everything is interconnected. The idea is to inject a pulse into the shorted net, and the current probe should light up near the chip pin that's sinking most of the current. Or so goes the theory.
But the results were inconclusive. We did find a ROM that did sink marginally more current. But the results were identical to the same testing performed on the good board. So it does not look like we have a short strong enough to be detected with this method.
However, while we did the testing, we found a ROM chip that became toasty hot, definitely faulty. That had to be the culprit. So I went about to desolder it. And this is when this happened:
Now my desoldering iron failed! It's a pandemic! We traced it down to the heating element. The connecting pads to the ceramic heater had delaminated. Unfixable. Apparently that's the major change between the Hakko FR-300 I have and the newer FR-301. They completely changed how the connections to the heating element are done in the newer one, so this must have been a known weakness. Anyhow, I ordered a new FR-300 heating element, and painfully desoldered the toasty ROM using desoldering braid. But I want to avoid doing anymore desoldering with the braid for fear of damaging the PCB.
Once the bad ROM was out, fortunately an option ROM not needed to boot, we re-tested the board in the machine using the logic analyzer, confident that we would recover our bit. But we did not! There is still another fault! Aaargh!
And we can't really debug the RAM refresh either until we repair the ROM and our processor issues regular ROM and RAM accesses (right now it just hangs, which also stops normal refresh cycles).
Once I get my heater element we'll be able to do a cleaner investigation by removing all the interfering chips and move forward hopefully.
But for now things keep breaking faster than we can repair them...
Marc
Comments
CuriousMarc
See? Just a small investment in a IR camera, and fortune awaits me as the world's best fried HP 9825 repairman. Just use the camera - and change all the electrolytics and paper caps. Also replace all the tubes. Just kidding. In this particular case I can see how the IR camera could be quite useful.
Someone I knew in the early 80s made just stupid money repairing DEC PDP11 boards using a medical imaging IR camera - he'd take a shot of a known-good board under power, then image the 'suspect' board and compare - chips would either be extra bright = shorted or very dark = open. The board would then go to rework to replace these devices. He had a 95% success rate which required little actual 'technical' knowledge on his company's part.
Paul Schuur
Thanks for the update. Bon voyage with the rest of the journey.
