Repairing A Vintage HP 9825 The Hard Way

[CuriousMarc] is at it again, this time trying to undo the damage from a poorly designed power circuit, that fried the internals of his HP 9825 computer. (Video, embedded below.)

The power supply on this particular model has a failure mode where a dying transistor can lead to 13 V on the 5 V line. This causes all the havoc one would expect on the internals of a 1970s era portable computer. This particular computer is rather rare, so instead of calling it a lost cause, our protagonist decides to replace the faulty transistor, install a proper overvoltage protection circuit, and then start the tedious hunt for which chips actually let their magic smoke out.

It helps that [Marc] has a second, known-working machine, so he can isolate individual parts and identify problems. There were plenty of problems, and a handful of false positives, but after replacing four blown chips, the machine finally boots again. There are eight episodes of troubleshooting, and they’re full of information and troubleshooting tips for this era of hardware. Set a couple hours aside, and enjoy the trip down the retro repair rabithole.


16 thoughts on “Repairing A Vintage HP 9825 The Hard Way

  1. Sad to hear. HP was known for quality engineering, hopefully this power supply was an anomaly. On the upside, CuriousMarc is an excellent choice to fix it. And HP manuals are complete with circuit diagrams *and* descriptions of how the circuit is supposed to work. So he’s got that going for him.

    Hope the damage is slight and repairable. Wishing him success!

    1. Been watching this as they were posted on YouTube. They have Board #1 diagnosed 3 more boards to go, AND the keyboard too. So based on the number of chips on board #1 that failed, there’s likely to be 4-5 chips on each of the boards going forward. But board #1 was the hardest due to the “homemade” state machines to separate CPU reads of DRAM from refresh signals and reads from ROM. He titled those sections “Noodle Logic”. Definitely worth watching all the way up to episode 8 where the big breakthrough happens (tracking down the the Intel Memory controller and mapping ALL the noodle logic).

  2. I’m always looking forward to Marc’s next video. Mostly fixing broken retro stuff with great explanations of how things work, why they don’t, and also plenty of bad jokes.

  3. Well since the failure mode is the first thing mentioned, I would guess this as been a common issue with these old machines which would make them ‘rarer’, which they are, and would become even more so. I am all for keeping certain vintage machines functioning, so as I’m not familiar with this one, I do not know its capabilities and that said, I say it is a matter of preference and a task of dedication along with some peoples desire to just make something better as in how a person undertakes redesigning a car simply to make it better in their opinion. Old vintage machines. Better living through better technology not chemistry, we all see how that has affected us. My next question is would the chips used still be available, some obsolete chips would be expected I think . I just think this may be an exercise in futility. I’m going to take a guess this machine used a fair amount of TTL and 13 V on 5V components would certainly spell wide spread death all the way through the circuits.

  4. People keep saying that HP makes “great” stuff, my first HP experience was with a room full of HP 9000 workstations and I got the distinct impression that they were awful. The keyboard and mouse connectors broke all the time and the enclosures seemed to be specifically designed to inflict torture on whoever had to open them. The cable routing was badly planned, it took stupid amounts of disassembly to get to key DIP switches.

    1. I would say that, generally, HP (now Keysight, nee Agilent) made very high quality gear. There were a few clunkers, but not many major design flaws got to production, and most were addressed in revision, often with recall or free repair for flawed units. Not everything was as servicable as one would like, but when I dig into a 50 year old piece of kit with sixteen cards, joined by backplane and cables, I know I am dealing with the limitations of the tech at the time, and I know that there may be design flaws or materials issues that were there when built, and I know that time takes a toll on many types of parts. Then there is abuse (mechanical, environmental, and so on) I can point to devices from many other manufacturers that had bigger flaws that the manufacturer punted on.

      Did you ever deal with a room full of VT100’s? The keyboard switches were less than stellar, the cables (1/4″ TRS type connector at the rear of the main case) pulled around the case tended to have a variety of failures.

      Or a 10base2 network (talk about damage waiting to happen)?

      Or an IDS line printer (PNP driver for the print head hammers, no flyback diode. And no strain relief on the ribbon to the printhead)

      OTOH, I have several pieces of HP gear from the 1960’s through the early 2000’s (Agilent) with very few problems. Aged VFD displays, the occasional bad capacitor or connector, and rarely anything else that isn’t due to abuse.

      1. Please indicate in my comment where I mentioned anything about DEC. Their gear was crap too but they never bragged about it.

        By the way, your anecdotes do not disprove mine. You are only demonstrating HPs uneven approach, great test gear and crap workstations.

        1. You started with “People keep saying that HP makes “great” stuff”. I gave some information why, as your experience did not match.

          In this context, “great” would be a relative term, and my “anecdotes” support the common claim that HP products were generally better (better in the sense of reliability, fitness for service, support, and the like) than most of the industry.

          As for the HP9000 workstation, you might want to be a bit more specific which models, as these spanned several decades (1980’s until the last models produced in 2007 or 2008, IIRC) and several lines. I am sure some were better designed than others. The ones I have experience with- a limited set, to be sure- were quite reliable and, by the standards of their day, not bad to work on. I don’t recall which variants at this point, as I left the university position just before Y2K.

          1. In particular the keyboard and mouse connectors had thin plastic retention clips that would snap off at the slightest provocation. They became useless because friction was insufficient to keep them in place. Any QA system that allows such flaws to make production is seriously broken.

        2. Maybe you remember the HP ad where they cleaned up an old piece of test gear that someone found at the bottom of a lake. HP used that story extensively to hype their “quality” but a bad plastic clip can upend a million dollar software demo.

  5. My HP-25 calculator is still in use on my desk on a near daily basis after 45 years (and i guess 15 changes of NiCd or NiMH cells. Never been disappointed by the HP test gear either.

  6. Worked on the assembly line for the HP 3702B Microwave Link Analyzer in the early to mid 1980’s, those were the days. They were built to last and not very light for carrying.

  7. I have a nice HP9826… Does the PSU have the same issue? I haven’t used it much because I lack boot disks but I could at least power it on and verify it worked. Now I am scared to do that.
    So is the PSU on the 9826 the same or similar design as the 9825? Do I need to do preventative replacements?

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.