I accidentally put 8.4V through my Pi

[alexeames]

This morning I was testing some new batteries and accidentally ended up connecting 8.4V to the 5V pin of my brand new Rev 2 Pi

I wasn't properly awake and it was a dumb and retarded thing to do.

But guess what? It still works.

I'd connected the +8.4V to the 5V pin and -ve to the GND pin. The Pi started up, but within about 20 seconds I realised I'd forgotten to use the 5V regulator, so pulled the plug.

There was no magic white smoke, which all electronic components need to keep inside in order to function.

I realise it's quite possible/probable that I've shortened the life of the Pi. Anybody care to speculate on what might be the effect of doing something so dumb?

[redhawk]

The SoC is regulated down to 3.3v therefore even if you're slightly over 5v it doesn't matter, I believe the maximum rating for RG2 is something like 9v (but don't quote me on this).
The 5v rail is used to power your USB devices so if anything they are more than likely to suffer the ill effects of your high voltage supply.

Does your USB ports have polyfuse protection??

Richard S.

[mahjongg]

The ground and 5V pins of what? The mini-B USB connector (fused) or the GPIO pin header (unfused)?
The PI has an overvoltage protection device that will short voltages above 5,6V to GND, but if you do not have some current limiting in between (such as the polyfuse) then you just run whatever the PSU can supply through it, and it will burn out if your supply can deliver too much current for it to cope with.
The PI does use 5V itself, as the SoC uses it to make its internal programmable low voltage core supply, yes it has a programmable switching power supply built right into the SoC!

[dknute]

Did you connect the power through GPIO or micro USB? My guess would be GPIO as I'd expect the polyfuse to fully trip when the protection diode starts drawing too much current.

With some luck your setup had enough resistance on the power supply path to form, along with the protection diode, a crude voltage regulator. But the diode will not survive this for long, it'll quickly go too hot and destroy itself. That and/or the PCB copper tracks will burn. If the exposure was brief enough there will be no further problems once things cool down again.

[alexeames]

The SoC is regulated down to 3.3v therefore even if you're slightly over 5v it doesn't matter, I believe the maximum rating for RG2 is something like 9v (but don't quote me on this).
The 5v rail is used to power your USB devices so if anything they are more than likely to suffer the ill effects of your high voltage supply.

Does your USB ports have polyfuse protection??

Richard S.

There were no USB devices installed. No polyfuses as it's a Rev 2.

The ground and 5V pins of what? The mini-B USB connector (fused) or the GPIO pin header (unfused)?

GPIO pin header.

The PI has an overvoltage protection device that will short voltages above 5,6V to GND, but if you do not have some current limiting in between (such as the polyfuse) then you just run whatever the PSU can supply through it, and it will burn out if your supply can deliver too much current for it to cope with.
The PI does use 5V itself, as the SoC uses it to make its internal programmable low voltage core supply, yes it has a programmable switching power supply built right into the SoC!

They were lipo batteries, but what might have saved me was that they have overcurrent protection circuits. I do know what I'm doing with lipos - just had an early morning retarded moment.

Did you connect the power through GPIO or micro USB? My guess would be GPIO as I'd expect the polyfuse to fully trip when the protection diode starts drawing too much current.

Correct

With some luck your setup had enough resistance on the power supply path to form, along with the protection diode, a crude voltage regulator. But the diode will not survive this for long, it'll quickly go too hot and destroy itself. That and/or the PCB copper tracks will burn. If the exposure was brief enough there will be no further problems once things cool down again.

I think the overcurrent protection circuit on the lipos saved the day

[carlvk]

I did that to my original board and now I can use it as a pretty red light.
My luck is not nearly as good as yours it seems

[alexeames]

I did that to my original board and now I can use it as a pretty red light.
My luck is not nearly as good as yours it seems

You should send it to the "will it blend?" guy.

[pygmy_giant]

I would incase it in clear resin to make a pretty paper-weight.

[Jim Manley]

You're probably well-acquainted with the statistical distributions associated with electronic components' individual abilities to withstand concentrated application of our stupidity. It appears the components on your particular board were all on the higher side of the peak in the distribution curve (not really a "bell" curve for the uninitiated, but a large, lump-shaped curve, of sorts, with unevenly-sized upper and lower areas around the mean). Without incrementally jacking up the juice until something actually fails, there's really no way to know how many more lives your poor Pi may actually have. Even for nominally-average semiconductor components, they usually last either seconds/minutes or tens of thousands of hours.

I have lots of fun taking engineering students around on tours of semiconductor fabs and other electronic device manufacturing facilities and someone inevitably asks, "So, which line are the highest-quality components made on, vs. the cheapie ones?" The answer is, of course, that they're all made right next to each other, right down to the individual wafers. Even after all this time, we still can't predict which specific components will be capable of handling the highest clock speeds or otherwise maxing out in performance, and which will be DOA, and they are often fabricated side-by-side.

It's true of light bulbs, microprocessors, or microwave resonant cavities (you have to make on the order of 10,000 cavities in order to get one that meets the specs for things like medical imaging machines - the rest are just melted down and you make another 10,000 to get another good one - that's why the good ones cost hundreds of thousands to millions of dollars ... each). This also applies in varying degrees to anything manufactured in quantity, whether it's the electronics, automotive, aerospace, or other industrial sector.

[alexeames]

Thanks Jim. I think the current limiting circuitry on the cells saved the Pi. It certainly would have helped limit the current that was shorted to ground (Still would have been up to 3 amps in bursts though). That and a robust set of components as you suggest.

I've put a proper connector on now, so it will only fit directly to the reg and not to the Pi.