0

I was using a Raspberry Pi Model 3B in my smart home system for about half a year now, which means it was powered around the clock - without any issues.

Yesterday, both of my living room lights (one of them controled using a relay) went crazy and randomly turned on and off. I immediately unplugged the Pi from the power grid and waited for a minute. Then I re-powered it and both lights went crazy just in the same moment when I plugged the power adapter to the grid, which is obviously caused by a faulty GPIO interface.

Then I realized a smell of hot plastic and saw smoke - the power cable of my Pi was melted (well, my power supply has a current limit of 4-5amps and I should have attached a fuse in series to it, I know...).

I almost burnt my fingers when I tried to hold the Pi in my hands and the processor and SD card slot were too hot to touch. The SD card cracked and when plugging it into another Pi, it also produces so much heat that you can smell it.

The curious thing about it is, that when using another SD card, the Pi still works (I get screen output and it even boots up), but the processor gets so hot just 2-3 seconds after powering it (you can not put your finger on it), even without an SD card, so it is not caused by the processor load.

I read much about overheated and cracked SD cards but now, but this time, it is probably not just an SD card, but also a faulty Pi.

Did anyone of you experience this problem? Does your Pi 3B get hot without having an SD card inserted?

And do you have any ideas how to prevent this fault in the future? This seems very dangerous to me as the cable could have cought fire and I even read about burning SD cards (with appropiate power supplies).

Ercksen
  • 147
  • 3
  • The cause of your problem is most likely the external circuitry you have connected to the Pi. As you have not detailed this no one can suggest a cause. Interfacing the Pi (or indeed any circuitry) to mains power requires special care. – Milliways Jun 27 '18 at 09:45
  • Wow... Fireworks! Glad to hear the damage wasn't any worse, and you didn't injure yourself. If you're looking for a post-mortem on this, it would be most helpful if you could: a) add a wiring diagram to your question, and b) add a photograph of the RPi. – Seamus Jun 27 '18 at 15:51
  • The relays I am using guarantee a complete galvanic isolation between the power grid and my Pi, and as they are still intact (and wired correctly), the reason must be somewhere else. My external curcuit is completely intact and no transistors/resistors/sensors/wires are damaged in any way and when attaching it to another Pi, it works as before. My Pi did not take any visible damage, it looks as before and the only fault you could see was the SD card. – Ercksen Jun 28 '18 at 20:59
  • I have a very new Pi (B+) and it has overheated and will not boot. The SD card is hot. I left it on over the weekend and could not login to it this morning. It was working fine on Friday. It had no external equipemnt plugged into it over the wk end. Not even a keyboard or a mouse. I was using it with "ssh -X" – Robin Oct 01 '18 at 07:50
  • What kind of PSU were you using? USB charger/generic 5V PSU/offical Raspberry PSU? Where you powering your Pi through Micro USB power input or backwards powered? Mine was backwards powered and I suspect a voltage spike or something like this to maybe have damaged it, as backwards powering omits this safety feature. – Ercksen Oct 02 '18 at 08:08

1 Answers1

2

The curious thing about it is, that when using another SD card, the Pi still works (I get screen output and it even boots up), but the processor gets so hot just 2-3 seconds after powering it (you can not put your finger on it), even without an SD card, so it is not caused by the processor load.

No, at this point it is likely caused by damaged circuitry in the power regulation on the board or in the SoC itself. I think this is not unusual with microcontrollers/microprocessors; they may still work, but too much current and/or voltage is passing through all or part of them. Or too little voltage, which can also lead to problems including excessive heat from some components.

See here for a discussion of some of the things that can happen after electronics have been subject to excessive voltage:

https://electronics.stackexchange.com/q/195573/52138

Destroying a transistor can lead to a short circuit (damage which may "not be visible outside of the IC"), resistors can break down in such a way that their resistance is altered, etc.

Of course, as long as you are watching it, it is probably safe to keep trying to plug it in, but I doubt the situation is going to improve.

do you have any ideas how to prevent this fault in the future?

As Milliways implies, if you are controlling a relay connected to mains power, this circuit is the likely culprit. And as you yourself observed, a fuse on the Pi's supply could well have prevented this.

You should also check your power supply with a multi-meter. I have at least one 9V supply which was connected to a short and now puts out 11-12V, without any other systems of a problem.

goldilocks
  • 58,859
  • 17
  • 112
  • 227
  • Thanks for your information. All components of my curcuit were tested and it works perfectly. My power supply does indeed have some problems - it is not regulated (in normal operation, it regulates 5 volts fine but when shorting it for example, voltage drops but current is flowing (6-7 amps), without a short curcuit protection at all), so I replaced it with a normal phone charger rated for 2 amps. I am convinced that the mains voltage did not cause the failure as the relay is completely intact and there are no connections between logical and high voltage side at all. – Ercksen Jun 28 '18 at 21:07
  • One thing I forgot about is the fact that my Pi's ground is connected to earth (which is not the case when using a normal phone charger), could this result in unexpected current flow or noise? – Ercksen Jun 28 '18 at 21:12
  • Well, whatever caused the problem, obviously it was not an intentional feature of anything (your circuit, your power supply, the Pi). I would presume the power supply and the Pi are also tested by someone at some point who said, "They work perfectly" ;) Possibly the relay was responsible -- although if it still tests okay, that's impossible to say. You could post a schematic of your circuit at our larger sibling site, Electrical Engineering and see if anyone sees a potential problem there. – goldilocks Jun 29 '18 at 12:02
  • WRT the Pi, I've been here almost every day for 5 1/2 years and read a high percentage of the questions, and certainly there have been reports of what probably amounts to "defective" but I don't recall a story like this one, where the Pi might actually have been a fire hazard. Of course I could easily have forgotten. Keep in mind more than 15 million of them have been sold. If your question is, "Why would this happen?" and you are sure nothing else could be responsible, then the answer is just component failure. It shouldn't happen, but it would be silly to say it is 100% impossible. – goldilocks Jun 29 '18 at 12:06
  • As it seems to be hard to investigate into this, I changed my power supply with one that offers an overcurrent protection and use a Model 3B+ instead. I added additional optocouplers to my curcuit to ensure a better galvanic isolation and now I'm gonna hope it makes things better :) Thanks for your tips – Ercksen Jun 30 '18 at 08:49