12

I've had a Fluke 8050A bench multimeter sitting on a shelf in my cold, damp garage for about 30 years. It was new in 1981, last powered up in the 1980s, and the display was dead. I came across these fine instructions for display repair and thought I'd try. I was encouraged to find that the original Fluke manual included the schematic, the layout and the calibration process.

I replaced the 40-year-old NiCads and wired up one digit of the display. By this time, the DMM had had about two minutes powered on in the last 30 years. My one digit seemed sensible so I wired up the rest and turned it on. It worked! All the buttons worked and the display looked credible in all modes, so I gave it some DC input and compared it with a 6-month-old DVM.

old and new readout

Great! Not bad drift for 40 years since last calibration!

Unfortunately, everything went downhill from there. The Fluke DMM has a Mostek MK3870/20 single chip processor which drives the display. That has 2kBytes of mask programmed ROM and 64 bytes of RAM. It's in a 40 pin plastic DIP package.

After a total of maybe 5 minutes of power-on time in 30 years, the processor started to misbehave. First it displayed voltage correctly, but confused the decimal point. Then it would display the correct voltage for a fraction of a second before then displaying all zeroes. Then it displayed random output as the Fluke buttons were pressed. Then it failed to write the display at all. And finally, its 4Mhz built-in oscillator stopped oscillating. It was dead. All this happened over perhaps another 10 minutes. Measuring everything, I'm pretty confident that it's the processor chip itself that failed.

A tragic tale, but at least the old DMM got to measure one last voltage before it died.

Is this sort of failure in a 40-year-old plastic chip to be expected? What might the actual failure mode be? Water ingress? Is there any way I might have avoided it?

Those who wish to close this question as opinion-based, please note that I am not asking why my DMM is broken. I am asking what failure modes exist for a 40-year old plastic chip, long unused, and what mitigation might help.

Thanks!

emrys57
  • 1,042
  • 7
  • 12
  • 13
    "A tragic tale, but at least the old DMM got to measure one last voltage before it died." +1 – Arsenal Dec 21 '21 at 09:18
  • 4
    What makes you think the CPU failed? What if the problem is somewhere else, like failed power supply? Or dead capacitor somewhere? – Justme Dec 21 '21 at 09:21
  • There is just one power supply to the single chip micro and it is working correctly. It's difficult to imagine how the chip can get power, have the oscillator not run, and not have failed. – emrys57 Dec 21 '21 at 09:22
  • How do you know power supply is working correctly? Have you verified it with oscilloscope that it has no excess ripple? The CPU depends also on other external components. What if the capacitor on reset pin failed? It is not very difficult to imagine that a single point of circuitry being out of spec causes processor to not run. – Justme Dec 21 '21 at 09:25
  • Yes, I checked with scope. Yes, I looked at the reset pin. – emrys57 Dec 21 '21 at 09:26
  • 4
    How do you know the crystal hasn't failed or got excess leakage on the PCB? Or that your scope probe is not loading the oscillator? – Spehro Pefhany Dec 21 '21 at 09:28
  • 2
    We all know that **cold and damp** isn't good for electronics. I would just leave the PCB in a warm and dry place for a week and then try again. We can only **guess** about any failure modes as you haven't proven anything yet. If you would replace the MK3870 and that would fix the issue, only then would you have proven that it was the MK3870. – Bimpelrekkie Dec 21 '21 at 09:31
  • When the processor first misbehaved, I checked the oscillator with the scope: a nice 4 MHz signal, no problem. So the crystal was not failing at that time, yet the processor was producing the wrong output. I conclude that the fault does not lie with the crystal. I'm not asking what's wrong with my DMM. I'm asking, what might cause an old plastic chip to fail, and could I have avoided it? – emrys57 Dec 21 '21 at 09:38
  • most chips have plastic housings, they are good in collection moisture, but might not transport it back to the outside fast enough (this is why good designed atomic bombs have e.g. ceramic chip housings...). Also ROMs have a data retention time... 40 years is certainly a possible time-span for ROM failures. In the Chip there might also be an aging of the points where the die is bonded... there are all sorts of possible failures. – schnedan Dec 21 '21 at 09:45
  • 10
    `Is there any way I might have avoided it?` - yes, you could have looked after it better and cared for it more and taken it on a walk once a day to read a volt or two. – Andy aka Dec 21 '21 at 09:46
  • 3
    Somewhat relevant (from retrocomputing): https://retrocomputing.stackexchange.com/questions/5479/is-it-safe-to-turn-on-a-40-year-old-trs-80 – MarkU Dec 21 '21 at 09:56
  • Yes, nice, thanks MarkU. I did look at the electrolytics before I started, and couldn't see any physical change. And the power is still OK. I have an even older computer out in the stables somewhere... I'm not tempted to turn it on! – emrys57 Dec 21 '21 at 10:00
  • 1
    Electromigration, ionic contamination, bond wire failure…. This list goes on. Without examining the chip it is anyone’s guess. Could you have done anything about it? Probably not. Besides, even with supposed abuse it lasted 40 years. – Kartman Dec 21 '21 at 10:54
  • 2
    Since it ran worse the longer it was on, think thermal. Maybe pc board expansion stressing old solder joints. Also, "exercise" the switch(es) vigorously with or without tuner cleaner spray. – AnalogKid Dec 21 '21 at 12:52
  • 5
    AND - I don't think his focus on the uC makes this "opinion-based", just inexperienced. DUH - that is what this forum is for. (PS - *all* questions are opinion-based. So are all moderator actions.) Please re-open for a full discussion and answers. I grew up in the era of $20 "MIL-grade" hermetically-sealed IC's, and would love to hear about the failure modes that drove and justified their existence. – AnalogKid Dec 21 '21 at 12:54
  • 3
    Seems like attention span is too short - the question is clear "What are the failure modes of an old IC and what are possible actions to reduce the risk?" even if the premise might be wrong (e.g. in this case it's some other fault). This question could be stripped down to only the last sentence - maybe putting it on top would reduce the risk of people mistaking this for a repair question or whatever. – Arsenal Dec 21 '21 at 13:23
  • 3
    @AnalogKid - Hi, Re: "*all questions are opinion-based. So are all moderator actions.*" True - but you seem to be implying that there has been a moderator action on this question. There hasn't yet! The question was closed by *community voting*, which is how the vast majority of closed questions get closed. Please see [here](/help/privileges/close-questions). Also you said: "*Please re-open*" but you did not cast a reopen vote yet. Please see [here](/help/reopen-questions) and cast a vote, if you want it reopened. This is community voting in action - people can vote to reopen, as well as close. – SamGibson Dec 21 '21 at 13:33
  • Sam - oops. I did not know any of that. This forum's operations are significantly different from the others I'm on. – AnalogKid Dec 21 '21 at 16:01
  • Are you sure that is a Fluke 8050a? I had one and it has an LCD display without backlight. Yours appears to have LED 7-segment digits. – GT Electronics Dec 23 '21 at 18:18
  • Failure of the LCD is the normal way a Fluke 8050A dies. Changing the display to LEDs is the fix, documented in the link in th equestion. – emrys57 Dec 24 '21 at 08:38

1 Answers1

2

This is a very interesting topic and actively researched e.g. DOI:10.1088/0268-1242/11/2/002 ( https://www.researchgate.net/publication/230944335_Corrosion-induced_degradation_of_microelectronic_devices ). It plays an important role for next-generation implants as well DOI:10.1088/1741-2560/10/3/031002 ( https://discovery.ucl.ac.uk/id/eprint/1395063/ ) and it is well known that powered vs unpowered (i.e. with or without bias) is very relevant DOI:10.3390/electronics9111884 ( https://www.mdpi.com/2079-9292/9/11/1884/htm ). This can be observed with e.g. watered electronics like whoops-it-fell-into-the-toilet-phones. If one manages to get them unpowered quickly i.e. remove the batteries, they can often be carefully dried and fully recovered. However, this is not limited to electronics exposed to humidity. Corrosive media (e.g., gasses), temperature, radiation, etc. can act the same way.

Long story short: Almost all processes have an equilibrium and/or require certain activation energy. If the parameters are suddenly changed (e.g. present electric field) those processes can be activated and tremendously accelerated. In your case assuming it was moisture or corrosive media, careful rinsing (optional), drying using getters and maybe sightly increased temperature (not too much - because temperature can lead to degradation as well) might have helped.

Christian B.
  • 1,851
  • 1
  • 5
  • 15
  • 1
    This leads me to think: when a device is powered, the internal heat helps humidity to go away. May be 1 hour a day help in preserving electronic devices from humidity? – linuxfan says Reinstate Monica Dec 21 '21 at 15:36
  • But is there really much of a moisture path in a DIP packaged chip? The resin thoroughly coats the wire frame, and the only path towards the silicon would be the tiny tunnels of the bond wires right? Sure with some differential thermal expansion in time there might be a slight path, but I'd think one could consider the silicon pretty much hermetically sealed? Or is the plastic itself porous/permeable? – parkside Dec 22 '21 at 11:58
  • 2
    parkside, water can get into plastic packages. eg https://www.semanticscholar.org/paper/A-rapid-technique-for-assessing-the-moisture-of-Chwastek-Shaw/5f52ad6955ca150e52aa16d53c153d2e3db75f75 – emrys57 Dec 22 '21 at 16:09
  • Christian B, great answer, thanks! Those papers are very relevant. And yes, I probably should have tried to dry the DMM out, perhaps for months, before powering it up: more fool me. Thanks for your help. – emrys57 Dec 22 '21 at 16:11