Tech help for potentially bad PSU

rvjones10

Distinguished
Jun 18, 2009
33
0
18,530
I recently had an issue with repeated reboots of my PC. Usually 1 or 2 right when I turn the PC on and then random restarts every hour or so. After running a number of tests and not getting errors (memtest and a few others) I reluctantly narrowed it to my PSU (CORSAIR HX Series HX850). I say reluctantly because its relatively new and a respectable brand. I have an ati HIS 5850 gpu which requires two PCI-E connectors, so I decided to unplug the non-modular(hard wired) PCI-E I was using and plug in two of the modular PCI-E connectors. Its been three days now and I've had no reboots.

Here's my question: Before I deal with customer support and shipping this thing back, I wanted to test the suspect PCI-E connectors. So I put them on my multimeter and all 6 (3 on each) tested at 12v. Now I'm confused. I thought I read somewhere that sometimes a bad connection might show bad unless its under load. Is this true? and How can I test that?

Thanks for the help and please note that before today I've only tested a wall socket for power.
 
Solution
Power (wattage) is a function of voltage times current. voltage in this case is constant and the current varies to the load. You would have to have a suitable volt/ohm meter that also measures at least 7 amps of current (75 watts= 12v x 6.25A), although the load is split among the 3 12v pin connectors. Problem would be getting the connecting probe attached to the 6 pin molex connector while it is plugged in and running. Then you could compare the 3 pins and see if they are all the same. Problem is I don't know if they are all equally shared or progressive in power delivery along the 3 pins. You would probably have to call Corsair to find out the best way to test the pins and how much current is supposed to be on each of the 3...

austing

Honorable
Jul 10, 2012
825
0
11,060


A way to test under load, is to obviously put the system under stress, run Prime95 and a GPU benchmark at the same time, and see what happens.

Also, have you tried measuring resistance in the wire?
 

avjguy2362

Honorable
Jun 21, 2012
732
0
11,360
Power (wattage) is a function of voltage times current. voltage in this case is constant and the current varies to the load. You would have to have a suitable volt/ohm meter that also measures at least 7 amps of current (75 watts= 12v x 6.25A), although the load is split among the 3 12v pin connectors. Problem would be getting the connecting probe attached to the 6 pin molex connector while it is plugged in and running. Then you could compare the 3 pins and see if they are all the same. Problem is I don't know if they are all equally shared or progressive in power delivery along the 3 pins. You would probably have to call Corsair to find out the best way to test the pins and how much current is supposed to be on each of the 3 successive pins. They may not want you to even try and getting the probe into the back of the connector may be difficult. It looks possible on the back of my molex connectors. I have an AX750. If your not extremely careful you could short the PS and end up worse than your are now!
 
Solution

rvjones10

Distinguished
Jun 18, 2009
33
0
18,530
Thanks guys for the help. In response to avjguy2362, thanks for the info, It seems it would be best just to isolate the fact that is has to be the PSU and let corsair deal with it. It would be best for me not to mess around with it. I just wanted to add that after 3 days of no reboot issue's, It started again. System was just sitting idle and I heard the beeps of the reboot from the next room. Its been happening very rarely, but still happening none the less. I've ordered a another PSU just to make sure its not the GPU. I will use one while the other is getting replaced and then keep it as a back up.