Voltage tolerance

[usp8riot]

What's the acceptable amount of voltage drop allowed when benching? Hardware monitor shows less than .5v in any one category. I've had this erratic problem over the past few months that's seemed to get slightly worse with random freezing or reboots.

Things tried:

• Reinstalled the OS several times
• Switched between Windows 8 and 7
• Ran Memtest just fine
• Ran OCCT and Prime95 for a few hours each
• Ran GPU benchmark
• Different drivers for GPU
• Checked if cables were secure on all components
• Checked SMART status and ran chkdisk initially

I thought it was an HDD problem at first because one HDD failed to show up in Windows/CMOS. I then tightened the cable and it seems solved as far as that goes. My hardware is in my sig. I'm suspecting bad caps on the mobo or PSU though with this mobo, it supposedly uses the newer and supposedly less fallible caps so maybe the PSU? It's the oldest component in my system. What do you guys think? I have a 480watt PSU from my last build I could test it with but I'm lazy and that's a lot of work . I just want you guys' opinion first and even if the 480w Antec could handle my system with 4 7,2k HDD's. btw, system in sig isn't 24/7 oc'd like that. That was only for a month of bragging rights. I ran default speeds for much of this system's life.

Also to add, I leave the PC on almost 24/7 since it's an HTPC and it seems to never happen unless I'm using it. When it's in standby for the night or the 2 hours before and after, I can't recall it ever freezing or restarting.

 

[NiHaoMike]

Calibrate it against a known good multimeter. (You can't do that in BIOS, you have to edit the lm_sensors configuration files.)

 

[Bobnova]

HWMonitor is almost always wrong.

12v/5v/3.3v need to be within 5% of their rated voltage.

Your sig says you have a Corsair 750w PSU. Is it that, or is it a 480w antec?
If it's the Antec, how old is it? If it's more than ~3 years, odds are it has crapcaps and is dead.

 

[usp8riot]

HWMonitor is almost always wrong.

12v/5v/3.3v need to be within 5% of their rated voltage.

Your sig says you have a Corsair 750w PSU. Is it that, or is it a 480w antec?
If it's the Antec, how old is it? If it's more than ~3 years, odds are it has crapcaps and is dead.

I didn't notice any voltage differences out of the ordinary in bios monitor or with HWMon. Yes, using the Corsair 750w currently with three HDDs and sig specs though it's at default clocks as it has been usually. The past day I've hooked the front fans up on my case which obviously helps with temps and leads me to wonder if my PSU was getting too hot though I've never heard PSU's having that problem from heat.

 

[Bobnova]

BIOS monitor is generally out to lunch as well, the same sensors feed both it and HWMonitor/etc.
You need to use a digital multimeter if you want real voltages.

 

[usp8riot]

Just an update for anyone that has comparable issues: After testing all pins on my PSU, it seems my restarts were related to Firefox. I remembered I changed the gfx options in about:config and I guess my GPU didn't seem to take too well to it. Three days now and I'm running solid. I also had a problem with Firefox and my 560ti when I first got it. I'd get screen corruption when using FF but only, it seemed, when I changed my gfx options. I thought it was definitely a GPU issue. That was before I bios flashed it so I began to wonder if the guy sold me a dud.

So for anyone who doesn't know, leave Firefox gfx options alone unless you know what you're doing. My first thought as most of you if something like that happened is definitely not browser issues but thinking back, it only seemed to happen when browsing but that's mostly what I do with this PC so it was hard to pin down.

 

[larrymoencurly]

What's the acceptable amount of voltage drop allowed when benching? Hardware monitor shows less than .5v in any one category.

ATX specs say a maximum deviation of ±5% from the nominal values of 3.3V, 5.0V, and 12V is acceptable, but 0.5V is 15% of 3.3V and 10% of 5.0V.

Also look at how much the values change because if the +12V varies from 12.5V to 11.5V, it's still within the ±5% spec, but that's a 9% change and indicates something is wrong with the PSU or connections. If your monitoring software always reports reasonable values for the voltages, it should be accurate for reading _changes_ in voltages

Get a digital multimeter. Anything with at least 3 1/2 digits of resolution will be more than accurate enough, even the $3-10 ones.

 

[Bobnova]

The shift seen at the monitoring chip is also effected by the motherboard design, how much loud is on the same traces as the monitor, how big the traces are, etc.

Definitely not a good sign, but IMO there isn't much that the software is genuinely useful for.