Suggestions on hot D17?


6 posts
by kohsuke » Sat Jul 27, 2013 8:02 pm
Hi,

I think I've damaged my Raspberry Pi, and I'd love to get people's opinions about the diagnosis.

I believe what happened was that I have accidentally connected 15V DC power to one of the GPIO pins. Since then, Raspberry Pi stopped working. When I disonnect everything and just try to power it on from USB power supply, the red power LED doesn't come on, and D17 gets very hot in a few seconds to the point that I can't touch it. Disconnected from any power source, multi-meter reading shows D17 has infinite ohm. I assume it means this isn't shorted. F3 read about 1.6ohm. resistance reading between TP1 and TP2 shows infinite as well, so I think this means there's no obvious short in the circuit.

I'm puzzled to understand what's going on. If D17 is eating that much current to make it untouchably hot, shouldn't that trip F3? Is it worth de-soldering D17 and see if that brings back Pi? I understand that there is a pretty good chance that other more critical circuits have been destroyed as well, but if it's broken any way, it wouldn't hurt to try a few things, no?

Any thoughts/feedbacks would be greatly appreciated.
Posts: 4
Joined: Sat Jul 27, 2013 7:42 pm
by drgeoff » Sat Jul 27, 2013 8:33 pm
D17 is a protection device and an undamaged RPi will work fine without it.

Your F3 resistance is higher than desirable. It may reduce in resistance if left in peace to recover.

F3 is also a protective device and an undamaged RPi will work fine if it is bypassed. (Same effect as powering via the GPIO header or backpowering from a naughty hub into the USB sockets.)

Are you certain that D17 is generating the heat and not RG2 on the other side of the board?

Be prepared to write this RPi off and consider it a bonus if found to be operational/repairable.
Posts: 2723
Joined: Wed Jan 25, 2012 6:39 pm
by kohsuke » Sat Jul 27, 2013 9:24 pm
Thank you for your suggestions.

I think you are right, it is RG2. That'd make more sense. Nonetheless I went ahead and desoldered D17, and things didn't improve. Looks like I'm going to have to buy a new one.

I was using Pi to drive an LED array that needed 15V. I just inserted the connector in a reverse direction, which caused 15V power line to connect directly to a GPIO pin. That was one expensive lesson!
Posts: 4
Joined: Sat Jul 27, 2013 7:42 pm
by kohsuke » Sat Jul 27, 2013 9:33 pm
According to the schematics at http://www.raspberrypi.org/wp-content/u ... .2_027.pdf, RG2 is NCP1117, and according to its datasheet at http://unihedron.com/projects/sqm-le/PD ... 1117-D.PDF it should tolerate 20V on the input line, so my understanding is that it should tolerate 15V coming in, and so I'd think the board should be able to light up the power LED.

... perhaps unless there's a short between 3.3V line and GND, I suppose.
Posts: 4
Joined: Sat Jul 27, 2013 7:42 pm
by drgeoff » Sat Jul 27, 2013 9:36 pm
kohsuke wrote:According to the schematics at http://www.raspberrypi.org/wp-content/u ... .2_027.pdf, RG2 is NCP1117, and according to its datasheet at http://unihedron.com/projects/sqm-le/PD ... 1117-D.PDF it should tolerate 20V on the input line, so my understanding is that it should tolerate 15V coming in, and so I'd think the board should be able to light up the power LED.

... perhaps unless there's a short between 3.3V line and GND, I suppose.

Or one of the further voltage regulators feeding into a short, eg a 'fused' SoC.

It wasn't 15v getting to RG2 that was the major problem. 15v will likely have destroyed the SoC. RG2 is probably still functional if asked to do its normal job.
Posts: 2723
Joined: Wed Jan 25, 2012 6:39 pm
by kohsuke » Sat Jul 27, 2013 9:42 pm
Sorry for a repeate self-reply.

Indeed, one pin of RG2 reads 0ohm to TP2, another reads 5ohm to TP2, and the last one is infinite, so I think that explains the behaviour. (The pin number denotation in the schematics and the datasheet doesn't match up, but in a functioning Pi I believe only one pin should be connected to GND/TP2.)
Posts: 4
Joined: Sat Jul 27, 2013 7:42 pm