(Topic ID: 293596)

How / What / Why does a CPU die?

By goingincirclez

2 years ago



Topic Stats

  • 9 posts
  • 8 Pinsiders participating
  • Latest reply 2 years ago by BigAl56
  • No one calls this topic a favorite

You

Linked Games

No games have been linked to this topic.

    #1 2 years ago

    My Time Fantasy (WMS System 7) has been trouble-free for over 2 years now; original MPU with zero battery damage and no Scanbe sockets. I rebuilt the 40-pin interconnect and power supply in 2019. The game ROM is custom to support an added feature using a relay to turn on a UV light to react with a custom playfield, but even that was proven stable for an entire show weekend in March 2020. No issues then; game has been played a few hours/mo at home between Mar2020 and this past weekend.

    So this weekend I turned it on and played some games, then left it on... after a while I noticed the displays were out. Power cycled and they came back but the game had lost its mind: scrambled info on displays, no attract mode, no audit mode, etc. Further investigation revealed the MPU diagnostic display digit saying either "0" or "gibberish" or nothing at all every time I turned on the game. Displays would rarely if ever even flicker, random sounds would play or lock, etc. Just a whole range of random haunted weirdness.

    To cut my rambling down: I wondered if my custom ROM had flaked out so I put the original one back in, but the symptoms were little changed. If I removed the driver board, the MPU would appear to stably boot to the "Zero Digit blink" (which traditionally points fingers at the driver board)... but not *consistently*. Still, I swapped driver boards with another game which confirmed the driver board was NOT the issue.

    Finally I realized the MPU had a socketed CPU chip and I just happened to have a spare 6808... swapped the spare chip in and everything went back to normal. Put the original one back in and the game freaks out.

    So, Yay... BUT! Thus far in my pinball tech "career" of 7 years, every chip failure I've experienced has generally traced to a direct cause-effect relationship. I.e., something (mechanical or already-stressed electric part failure; boneheaded troubleshooting fallout, etc) happened that caused a cascading failure at the chip. I never had a chip - much less a 40-year proven CPU! - just "die while sitting".

    Now I realize that nothing lasts forever, and maybe that 40-year-old CPU chip finally just somehow "wore out" (It was an AMI brand which I understand might have been on borrowed time anyway)... but part of me isn't willing to think this is actually fixed.

    How often does old, proven, otherwise reliable silicon "just die"? FWIW I've inspected the boards and connectors and there truly isn't anything obvious lurking. But I'd like to be at ease!

    #2 2 years ago

    It didn't die just sitting, it died when you turned it on. The cpu has been used over many years, possibly the legs are oxidized and sitting for a year they got more so and when you turned it on, high resistance on that leg, heat is generated and chip fails. If this were the case it would have happened even if you played it every day.

    Just a theory but I think the first point is the accurate one.

    Although the symptoms you described sounded more like corrupted ram.

    #3 2 years ago

    Is your failed CPU just by chance an AMI brand 6802/6808?

    #4 2 years ago

    CPU chips are generally the least likely to fail, but on rare occasions they do.

    Usually because of material failure (caused by age, heat, or both), or significant overvoltage.

    #5 2 years ago

    Could be many causes. Power supply spike or moisture getting to the chip
    inside the package are the most likely. People seem to be under the impression
    solid state devices should last forever. Not so.

    #6 2 years ago
    Quoted from Tuukka:

    Is your failed CPU just by chance an AMI brand 6802/6808?

    Quoted from goingincirclez:

    Now I realize that nothing lasts forever, and maybe that 40-year-old CPU chip finally just somehow "wore out" (It was an AMI brand which I understand might have been on borrowed time anyway)... but part of me isn't willing to think this is actually fixed.

    #7 2 years ago

    I appreciate the chatter. At very least I can claim due diligence toward making sure my repairs are sound

    I was indeed surprised at the amount of "oxide dust film" on the sockets when I removed the chips for inspection / troubleshooting - it's amazing to think that stuff can even get in such tiny crevices. (I'd seen that on other boards of course, but this particular one was clean enough otherwise with nice socket exteriors and unoxidized chip legs). So a failure induced by gradual buildup of funk and heat and such, does make theoretical sense. I've just (fortunately?) not experienced that myself just yet - it's always been "X happened so the chip went poof".

    Funny enough I've resurrected dead / project boards and always assumed "X then poof" originally caused those inherited failures, but maybe more of those really were due to simple incidental age/usage than I assumed.

    #8 2 years ago
    Quoted from ForceFlow:

    Usually because of material failure

    This was always my thought. Material purity and doping within the silicon chips and transistors can effect longevity. I imagine things weren’t quite as good back in the late 70s and 80s, so overtime with heat and vibration, things just breakdown I assume.

    #9 2 years ago

    They are made out of sand. It's a miracle they ever worked.

    Reply

    Wanna join the discussion? Please sign in to reply to this topic.

    Hey there! Welcome to Pinside!

    Donate to Pinside

    Great to see you're enjoying Pinside! Did you know Pinside is able to run without any 3rd-party banners or ads, thanks to the support from our visitors? Please consider a donation to Pinside and get anext to your username to show for it! Or better yet, subscribe to Pinside+!


    This page was printed from https://pinside.com/pinball/forum/topic/how-what-why-does-a-cpu-die and we tried optimising it for printing. Some page elements may have been deliberately hidden.

    Scan the QR code on the left to jump to the URL this document was printed from.