First post, by kool kitty89
Does anyone know what might cause the -12V line on an AT form factor motherboard to short to ground?
This is a 486 chipset board without onboard I/O or sound (it's a Symphony chipset board with 386 and 486 sockets) and I don't think any of the onboard logic makes use of -12v signaling or reference voltages.
I had it working with a few different 486SX and DX chips, 486DLC, AM386DXL-40 (which it came with), and an Intel 386DX-25. (the latter running at 33 MHz using a 66 MHz oscillator I swapped in for the 80 MHz one it came with)
I'd had it idling on 3Dbench for a while with the 386 and left the room to check on something else and found it turned off.
I thought I'd just forgotten I'd shut it off before I left and swapped in a i386DX-33 IV and it wouldn't power on.
I realized the PSU was staying off with either position of the switch, so I unplugged everything, then tried the power supply again and its fan spun up fine, tried a 90mm fan and it powered that on fine too. Tried plugging it into the board again and it stayed off. (the internal breaker popped, and reset after unplugging the PSU from the board and AC power for several seconds)
I tried the PSU in another board and that worked fine, then tried just the P9 half of the AT connector (5V, -5V, gnd only) and the PSU powered up fine, then P8 and it stayed off again.
So I probed the P8 side of the connector on the board using a multimeter in resistance mode and the +12, key (+5V), and power good lines all showed high resistance values as expected but the -12V line read only 0.9 ohms, so it seems like there's a short there.
I don't see any obvious damage on the board and I didn't notice any foreign objects in the ISA slots (or any flecks of scraped gold contact plating) that might short the -12v line there and no burnt or blown capacitors or diodes around the ISA slots or elsewhere.
I'm currently thinking it may be a capacitor that failed shorted and the PSU halted things before it could actually pop or burn. If it was one of the tantalum caps it obviously would've failed more dramatically if allowed to and had it been on the +5V line the PSU might have passed enough current to allow that (it's done that on an 8088 board before) but the -12V line has a low current threshold that may be preventing that.
All the tanalum caps are black, rectangular surface mounted things so also might not show partial burning (but there was no odor and no change to the white polarity markings on them) and I didn't see any discoloration to the ceramic caps around, even the tiny surface mounted ones. (though some of the tiny SMDs I was looking at were probably resistors)
To be clear: the board was unpopulated when I did the power on testing, so no cards in the ISA slots could be the source of the short. I don't think the video card makes use of it anyway, but the Vibra 16S I have does have the -12V pin connected and the multi-IO card I had hooked to a CF adapter does have it for the RS-232 ports (which were unused during these tests).