Fail Of The Week: Electromigration Nearly Killed This Xerox Alto

The Living Computers museum in Seattle has a Xerox Alto, the machine famous for being the first to sport a mouse-based windowing graphical user interface. They received it in working condition and put it in their exhibit, but were dismayed when a year later it ceased to operate. Some detective work revealed that the power supply was failing to reach parts of the machine, and further investigation revealed an unlikely culprit. Electromigration had degraded the contacts between the supply pins and the backplane traces.

If electromigration is new to you, don’t feel ashamed, it was a new one to us too. It’s “the transport of material caused by the gradual movement of the ions in a conductor due to the momentum transfer between conducting electrons and diffusing metal atoms“, got it? Okay, that’s just a long way to say that passing a sufficiently high current through a conductor for a long time can physically move the metal of that conductor.

This one just doesn’t pop up very often. But in the case of the Alto, an under-specified power distribution system caused a lot of current to flow through too few solder joints. Those joints were left without enough metal to make a decent connection, so they failed.

The fix came with a set of sturdy busbars freshly soldered to the pins, but the interest in this piece comes more from the unusual phenomenon that caused it. That soldered joints can seemingly flow away defies belief. It’s still something most of us will never encounter, but like tales of ball lightning it’s one for the “Fancy that!” collection.

We’ve covered the Alto before, most notably [Ken Shirriff]’s work in restoring the Computer History Museum’s example.

51 thoughts on “Fail Of The Week: Electromigration Nearly Killed This Xerox Alto

          1. So, the subject of non-web-based election machines is too political? Dude, often the most political action one can make is to “refuse to get political” -also one of the most reactionary stances one can make by fully supporting the status-quo, refusing to even question it. The Internet was started by the US DoD, and the decision to open it up to commercial markets was political. Everything is political. The freedom, or lack thereof, to hack one’s possessions, and FOSS itself is political. A tech trade war with China will have profound effects on Hackaday readers. Just because political parties or individual politicians are rarely named doesn’t mean Hackaday isn’t political.
            Actually I find your statement offensive. Who are you to decide for us what’s “political”?

    1. OMG! If only a person could actually live off of troll food. Look at how many have fed this one! Come on guys! Now “Mike” is going to come back time and time again just looking to get you all riled up ’cause he knows he can. LMAO

  1. That’s an interesting thing. I would thought after enough material had degraded the resistance would have increased causing the joint to heat up melt the solder perhaps, start arcing and burn what was left of the connection away.
    The photo seems to show the broken connection to be quiet even around the pin.

      1. My notes always say “cracked solder point on the header.” I used to get into the whole thermal/stress fatigue explanation, but I’ve only met like two people that understand it. Now I just tell them that their solder has become nippled.

  2. My original xbox had this problem, well within it’s life cycle. When it refused to turn on one day, I found wiggling the plug made some oddly satisfying crackling noises and on taking it apart I found that all the solder had mysteriously vanished from where the power supply cord connector was attached to the power supply board (through hole) – I resoldered it with generous amounts of leaded solder and about a year later the problem recurred. MS’s “fix” apparently was to send out power cords with a breaker built in to prevent fires. I’ve just called it “solder migration” since, though I have no clue if I’ll find my missing solder in Florida or not.

  3. I see this stuff often in solder thru hole and even big hot surface mount resistors. It’s weak then things start moving and then there is a ring of failure starting to form even before it goes. Most of my repair work I fix and finish with a gun or iron. Big things like TO220 transistors and connector pins don’t solder fast on that quick wave of molten solder.

  4. My original xbox had this problem, well within it’s life cycle. When it wouldn’t turn on one day, wiggling the power cord produced a crackle. On disassembly I found the solder had all vanished from where the plug for the power cord was soldered onto the psu board, re-soldered it and it was fine for a year or so before the problem recurred. I guess it was an issue with certain models, as MS issued cords with built in breakers as part of a recall over some fires. I’ve just referred to it as “solder migration” though to this day I have no clue if I’d find that missing solder relaxing in Florida.

    1. Wouldn’t the alternating current prevent migration since you’re just moving things back and forth instead of in one continuous direction? Same reason you can’t electroplate with AC.

      1. As much as I’d love to understand the why of it thus also where the solder actually went and more importantly whether it increases fire risk, all I can say for sure is that it vanished from where it was supposed to be even after cleaning and re-soldering with leaded. I’ve also heard of this at least once on the A/C plug of a much more recently made large cell battery charger.

  5. Dunno if it’s electromigration, or just some leadless shit, but I had to redo solders on my Samsung washing machine as it seems to fail right after two year due to sone soldering failing on the power traces… Some case hacking and a few solder saves several hundred bucks on those…

  6. Electrolysis of impurities in conducting material could separate conductors when energized by flow of electrons, Curie temperature, and boiling coused by pressure. Yet there is a component of 8.77e-9 grays (Gy) background ionizing radion every hour in the form of Radiolysis .

  7. To me the brass bus bars are an ugly fix for a problem that took over 40 years to manifest itself. I think I would have heavily resoldered the joints with some high quality 60/40 (seems to last better than 63/37 in my experience) and called it good for the next 40 years.
    But that’s just my opinion, of course…

    1. In another forty years, will anyone around be able or available to diagnose the problem? There have been substantial changes (not for the better) in solder, who knows what might change in the future? We like to see it still functioning, but there might not be the money or the will to diagnose it in the future. Part of the magic of museums is they inspire one to think long-term, and an ugly fix that’s obvious might be the best way to let future caretakers aware of the problem in the future (ie if documentation gets lost). Somewhat ironic, but it’s a consideration.

  8. Electromigration is the cause of both the Intel P67 chipset’s SATA III degradation and the funny thing known as “Sudden Northwood Death Syndrome” where the earliest stepping Northwood Pentium 4s with a 400MHz FSB would just cease to work, especially in an enthusiast overclocking environment. Any overclocking is bound to introduce electromigration if temperatures are not kept in check, and it can also happen to mobile CPUs a lot quicker if they aren’t kept within a decent temperature range. the further you are from the thermal junction max, the better.

    1. The visit of Steve Jobs, which led to the development of Apple into one of the richest corporations in the world, and the neglect by Xerox, which was already a giant corporation will always be a business, or even life lesson, for techies who want to change the world. That story may be as important as the actual tech advances that the Xerox Parc team made.

  9. Decades ago, before printed circuit boards, there were tag strips and the prevailing teaching was that a wire had to be wrapped around a tag (and if possible threaded through a hole in the tag) so that a good, long term, electrical contact was made by each wire on the tag and only then was the joint soldered (to keep the rain out). I have often wondered why that rule was taught (replacing a component was a PITA) but perhaps it was early experience of electromigration (or whatever it is).

  10. Solder has a much, much higher electrical resistance than copper wire. The contacts, if running high current, should be mechanically joined, such as the screw terminals found in every single electrical socket in your house, or a mechanical crimp style connection, as found in nearly every joint in your car.

    Running a dozen amps through a solder joint without the copper touching other copper will boil off the lead solder.

    1. Although, take into consideration that Alto probably wasn’t made to last 45 years, and didn’t have safety issues of cars. Probably it wasn’t planned obsolescence, but… I’m not sure if you intended to post a kind of educational or public service announcement…, but in some cases manufacturers don’t care and will cut corners, and that’s a lesson worth remembering, too.

      1. Take a look at the NASA wire loom guidelines. All connections should be crimped, not soldered. Nothing to do with electromigration, all has to do with vibration. Solder wicks up the wire and creates an unsupported point where the wire starts to flex, eventually work hardens, and fails. That’s why the battery cables (and all the wires in your car) are crimped. Just lucky it is both cheaper and more reliable. (The NASA guidelines also call for wire tie downs within a certain distance of the connection.

        If you are passing enough current to heat the solder up enough to melt, never mind _boil_, you have far larger problems. Your wire insulation is probably starting to get crispy, soft, and runny. If you happen to have chosen teflon or other high temp wire covering, the wire will pull out of the connector. Another reason to crimp.

        Crimping is the stranded wire equivalent of wire bonding used in chips. Done right, it forms an airtight metal-metal connection that is second only to welding similar materials.

    2. Nonsense, that would imply you could never run a current over “a dozen amps” through a PCB with soldered components, which is exactly what happens inside many industrial frequency drives, servo drives, large switchmode power supplies, and even on your computer mainboard, which has a converter that usually supplies somewhere in the range of 60-80A to your CPU.

      What matters is the size of the connection (cross section and length), and the heat dissipation; if properly designed, a large current should not be a problem.

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.