PC shuts down like power cut

[Trigeminal]

OK guys, just completed new build about a month and a half ago and need some advice. My new build is as follows:

Silverstone FT02-B
GIGABYTE GA-X58A-UD5
Core i7 930
Thermalright Ultra-120 eXtreme-1366 RT Rev. C CPU Heatsink
OCZ 6GB (3 x 2GB) DDR3 1600 RAM
SAPPHIRE Radeon HD 5850
SeaSonic X650 Gold 650W Power Supply
Western Digital Black 750GB HDD
Western Digital Green 2TB HDD
Win7 x64

For the first 3 weeks I had only 1 crash. Then after using it for 3 weeks(about a month ago now), I had 1 shut down on Friday, then 2 on Saturday, and then 5 or 6 on Sunday. I say "shut down" because it was just like what would happen if you pulled the power cord, and it required me to press the power on the front to start it. The crashes seemed to be more frequent to the point I was lucky to boot into windows, sometimes it would crash on the bios load, and several times in memtest (though no ram errors were ever present). I thought it was the power supply, but I wasn't positive, so I changed the setting in the bios to turn on with power loss, but the computer still required me to push the power button on the case to restart. So I assumed it was the mobo, until I chatted with a friend who stated that power supplies and mobo have fault protectors that require manual restart to prevent damage. Windows was set to run error log, but this only occured once despite the many shutdowns, it just "shut down".

I had no overclock on and then went to standard motherboard bios settings with lower voltage for ram and QPI. The crashes were less frequent, but they still occured quite often. I was quite busy with school so I pulled out my old rig and set this aside until my big class was done.

When my class finished last week I tested it and it still had the same "shut down" problem. So I ordered a new power supply, it came on Monday. I was installing it and noticed two of the Peripheral/IDE/SATA modular power cables were not latched in and easily fell out. I wondered if this was the problem, but figured I'd try the new power supply first.
That power supply resulted in 72 hours straight with no crash. Then yesterday evening I tried the old power supply. I plugged everything in and double checked it, guess what, no crash for 24 straight hours and counting. So...

Can a loose Peripheral/IDE/SATA modular power cable ause this???

Does this sound like the Power Suppply (in this case my shit install skills) are at fault for the crashes?

The old power supply has a 12V line that read 12.58V in the bios, the new one read 11.9V Should I return the old one for this variance?

Any other thoughts???

 

[Puffnstuff]

Yes a loose cable can cause problems like that. If you have access to a digital multimeter measure the rail output while the system is on and while under load. A ps will reduce output as heat builds and a failing unit will loose output that much sooner. Testing it will help you determine if the unit is bad or if a loose cable was to blame for your problem.

 

[Trigeminal]

Yes a loose cable can cause problems like that. If you have access to a digital multimeter measure the rail output while the system is on and while under load. A ps will reduce output as heat builds and a failing unit will loose output that much sooner. Testing it will help you determine if the unit is bad or if a loose cable was to blame for your problem.

Unfortunately I don't have access to a digital multimeter. Will the bios readings do any good?

 

[TemjinGold]

The fact that your old PSU registered 12.58 means it's on its way out if not already there. I would definitely return it.

 

[Trigeminal]

The fact that your old PSU registered 12.58 means it's on its way out if not already there. I would definitely return it.

It's brand new. So you mean it is broken or is headed that way???

12.58V is that bad???

 

[Trigeminal]

The stickied thread states that "For 12V, 11.4 to 12.6". Thoughts???

 

[0roo0roo]

bios reading of voltage is not something to really care about from what i've read.
you are probably avoiding the hard work..we all do that.
like it or not the real way to test this is to strip the system down to minimum, stress test with software tools like memtest95 and cpu burn for hours at the least, and slowly add back components back one by one repeating said tests until you find the breaking point or faulty component.

 

[Trigeminal]

bios reading of voltage is not something to really care about from what i've read.
you are probably avoiding the hard work..we all do that.
like it or not the real way to test this is to strip the system down to minimum, stress test with software tools like memtest95 and cpu burn for hours at the least, and slowly add back components back one by one repeating said tests until you find the breaking point or faulty component.

Thanks for the advice but I tested for 24 hours in memtest before install of windows and 72 hours in prime95 once windows was installed. I am busy, but very rarely take short cuts, that's why the "looseness" of these cords really struck me.

 

[bryanl]

The fact that your old PSU registered 12.58 means it's on its way out if not already there. I would definitely return it.

The fact the old PSU registered 12.58V means nothing by itself, especially when the reading is from the motherboard hardware, which is not known for being highly accurate, and not from a digital multimeter.

I would first raise any temperature alarm/shutdown limits in the BIOS or hardware monitoring program. If the computer still shuts down unexpectedly, operate it with the side panel removed and a large fan blowing into it. Power supply capacity drops as temperature rises, and its possible your system is consuming too much for the 650W Seasonic because the Radeon 5850 draws about 150-300W, the Intel i7 930 about 120-220W, and everything else may add 50-100W.

A loose power connector can cause undervoltage, and it may make it hot.

 

[Trigeminal]

The fact the old PSU registered 12.58V means nothing by itself, especially when the reading is from the motherboard hardware, which is not known for being highly accurate, and not from a digital multimeter.

I would first raise any temperature alarm/shutdown limits in the BIOS or hardware monitoring program. If the computer still shuts down unexpectedly, operate it with the side panel removed and a large fan blowing into it. Power supply capacity drops as temperature rises, and its possible your system is consuming too much for the 650W Seasonic because the Radeon 5850 draws about 150-300W, the Intel i7 930 about 120-220W, and everything else may add 50-100W.

A loose power connector can cause undervoltage, and it may make it hot.

Temps it in the mid 30s for both GPU and CPU during idle. Not sure what load was hitting as crashes rarely occured under any load at all. As I stated mostly during the boot process or while surfing. Tried complete reinstall of windows with no luck. HD tests were normal unless restart occured during shutdown or defrag corrupting data.

Definately not overload as the crashes tend to occur in low load states with zero overclock.

36 hours with no crash.