CPU / RAM get *extremely* hot then pi shuts down

[nfman]

Hi forum dwellers,

My Son's Pi has started failing, and I thought I'd ask here before contacting Farnell for an RMA.

When power is applied, the CPU/RAM, regulator (RG3) and F3 get blisteringly hot within 30 seconds or so, and then the pi shuts down. Before it shuts down, the boot process appears to be normal.

My guess is that the cause of the shutdown is the polyfuse tripping, as it appears to be drawing current to the limit of the USB Hub. I measured ~680mA without an SD card installed (my Daughter's Pi draws ~12mA in the same state).

What's less clear is why it's drawing so much current, apparently via the CPU/RAM. Is there anything obvious we should check before I contact Farnell?

Tim

[johnbeetem]

Sounds like something is drawing way too much +3.3V current, which heats up RG2. RasPi has excellent heat conduction through the power and ground planes, so it's hard to tell which component is the main heat source.

Just for fun, it would be interesting to compare +3.3V (RG2 tab) between the working and failing RasPis. Does USB work before RasPi dies? If not, the LAN9512 USB hub/Ethernet could be the culprit, since it uses +3.3V.

In any case, sounds to me like an RMA.

[nfman]

Just for fun, it would be interesting to compare +3.3V (RG2 tab) between the working and failing RasPis.

I get 3.29V on the working pi (an RS model FWIW) which seems reasonable, and ~2.55v on the unhappy one.

Does USB work before RasPi dies? If not, the LAN9512 USB hub/Ethernet could be the culprit, since it uses +3.3V.

I'm not sure if USB was working. I'll be sending it back to e14, so will probably leave any further diagnosis to them. If I get a few spare minutes over the weekend, I might look at it again and let you know.

[johnbeetem]

Just for fun, it would be interesting to compare +3.3V (RG2 tab) between the working and failing RasPis.

I get 3.29V on the working pi (an RS model FWIW) which seems reasonable, and ~2.55v on the unhappy one.

3.29V is well within the 2% regulation one usually gets from a good regulator. Something hooked up to 3.3V on the unhappy RasPi is definitely broken. Sounds like RG2 is getting way hot trying to source 3.3V and F3 is doing its job preventing RasPi from catching fire.

Purely for amusement, you might try seeing if any of the capacitors connected to the Ethernet jack are shorted out by solder blobs, or if the any of the Ethernet jack pins (LAN9512 side) somehow got shorted to ground.

In any case, it's RMA time.