.12V on ground?

[jalek]

I have an Antec SU 430 that's seemed ok. I started having regular crashes and problems with 3d rendering on a pretty new 8800GTS.

I looked at speedfan and it showed the 12V supply at 11.3. BIOS showed it at 11.97. I tried checking voltage with a multimeter, but if I just attach the ground to either black wire on a molex connector and either the 12 or 5 volt connector, I get a spark. Instant overload on the 5V.

I checked ground to case, and there's +.12VDC. Am I missing a ground to case or something? That just doesn't seem right, but making a direct ground I highly suspect would just be making a direct short and take out whatever's leaking voltage and it seems to be working ok for now.

Is a voltage on the ground wires normal?

 

[lucuis]

Ground wires shouldn't have current unless something is wrong. Ground is essentially a safe path for current to flow if there a short, that way the user or other components don't get fried.

 

[RetiredChief]

A quick ck to see if we are on the same page. You connect the Black lead from the DVM to the black wire on the molex connector, and the red lead from the DVM to the +12 or +5 molex pin and get a arc and cause the +5 PSU rail to go into an overload condition.

This is more an indication of defective DVM (or as I have done, tried to measure voltage with the meter set to measure current. In voltage mode there is normally 10 megohms resistance between the leads (max about 12 microamps current.

Normally I attach the black DVM lead to the Case (alligater clip) and then make my measurements to the various voltage points including to the return lines..

Reference to +0.12V on black molex to case - Power off computer and unplug PSU power cord. Ohms check Case to (1) Black lead on molex connector, and then (2) case to Ground on PSU AC Power receptacle Both should be 0, or very near zero ohms. Also measure the case to AC neutral at PSU power receptical

Added
Lucuis: The black wires on a molex (as all black wire) are really return lines that are connected to Ground and as such the have the same current flowing thru them as the +, or neg voltage lead. I think you really meant ground wires shouldn't have a voltage on them.

 

[jalek]

That's correct. With the cord pulled, there was nearly no resistance between the common ground on the molex connector and the case or the case and the ground on the PSU receptacle.

Also, after restarting, the odd voltage is gone and I got 12.04 and 5.10 from the molex connector without any issues. Speedfan voltage readings didn't change, they're as reliable as always.

The ground differential didn't return with a 3d app hung, but 12V dropped to 11.68. I think I know the culprit.

Thanks. I was thinking about upgrading anyway, now it looks necessary.

 

[lucuis]

It's usually normal for voltage to drop some under heavy load.

The quality of the PSU will determine how much.