Is This Normal For a Power Supply

[bayflats]

Hi all. I have a Corsair HX850 Professional PSU. Its a lot more than I really need, but I decided to pay up as I know a dodgy power supply can really screw with stability.

Recently my PC has been randomly restarting without BSOD. If I try to restart using the 'Reset' button it normally gives me the beeps telling me my ram is screwed, but I have run MemTest on it several times and the ram seems fine. The only way to restart the machine is to unplug and wait for 20secs, then restart the machine. Looking online, the issues I have sound similar to those with PSU issues, so I looked at my bios to check the voltages. I saw the following:

CPU Voltage - 1.288v
3.3v - 2.896v
5v - 6.033v
12v - 14.104v

These dont look normal - I'd expect the numbers to match the voltage ratings of the rail. Is my PSU all messed in which case I'd be really quite disappointed.

This might be related, but I am not sure. My GPU is also acting up, with fans constantly at 75% - 90% and idle temps at 85c going to 100c under load. This also kicked up in the last few months. Its a standard HIS 6970 which should not (and did not) struggle to run my games. At the moment I get anywhere from 6 - 30 FPS on games like Far Cry 3 on low settings 1920x1280 which is way below what it should be doing (with fans at 100% and 100c).

Thanks in advance.

 

[BobbyBubblehead]

normal tolerance is 5%

Supply [V] Tolerance Range (min. to max.) Ripple (p. to p. max.)
+5 VDC ±5% (±0.25 V) +4.75 V to +5.25 V 50 mV
−5 VDC ±10% (±0.50 V) –4.50 V to –5.50 V 50 mV
+12 VDC ±5% (±0.60 V) +11.40 V to +12.60 V 120 mV
−12 VDC ±10% (±1.20 V) –10.80 V to –13.20 V 120 mV
+3.3 VDC ±5% (±0.165 V) +3.135 V to +3.465 V 50 mV
+5 VSB ±5% (±0.25 V) +4.75 V to +5.25 V 50 mV

even for dubious bios readings that's "goodnight vienna" high voltage.

you may have a grounding problem in the case or the power supply is delivering way too much.

i`d break the whole thing down to try and identify any ground problem.
failing that send the power back.

the motherboard cant rectify that much so it may be doing damage

 

[bayflats]

BobbyBubblehead - many thanks for the quick reply.

How would you suggest I begin looking for a grounding problem? Could it be that I simply overloaded something or got the wiring wrong? Sorry, dont know much about this.

Do you think its also the cause for my GPU acting up all of a sudden?

Thanks again.

 

[BobbyBubblehead]

in effect you get a short on the casing. could be a motherboard screw in the wrong place or a molex cable un-plugged with a pin touching the case somewhere. Or a trapped wire thats sleeve has exposed the wire etc.

check round all the screws and make sure no unused cable pins are touching the case or electronics.

corsair power supplies are generally very reliable (only make I entertain buying really)

note. your in a bit of a dangerous situation if it is live for yourself and the electronics.

I had a bad ground once and touched the radiator with my other hand standing up... I got x amount of volts across my arms. not planning to do it again. lol

edit: dont use the computer until we have figured it out

 

[EarthDog]

While your bios is likely more accurate than software, you really need to take a multimeter to the board to confirm those voltages. I wouldn't worry too much about those voltages until you confirm them with a MM.

 

[BobbyBubblehead]

While your bios is likely more accurate than software, you really need to take a multimeter to the board to confirm those voltages. I wouldn't worry too much about those voltages until you confirm them with a MM.

I second that. maybe a good time to pick up a multimeter. always a handy bottom line test.

And it will come in handy time and again once you have one

 

[EarthDog]

It should be the first test, honestly, seeing as how software is usually quite bad at reading it.

To make any decisions without testing it properly first is premature at best.

 

[BobbyBubblehead]

i've never seen software that far off before. but I see your point if the psu is a little high and the software exacerbates the situation. (I give me a brownie point for spelling exacerbates right... did nothing for my dyslexiar erm dislexia erm dyslexia )

doddyduddleheab - peace out

 

[Silver_Pharaoh]

i've never seen software that far off before. but I see your point if the psu is a little high and the software exacerbates the situation. (I give me a brownie point for spelling exacerbates right... did nothing for my dyslexiar erm dislexia erm dyslexia )

doddyduddleheab - peace out

HWmonitor says my 12 volt rail is 4 volts, and the -12 volt rail is -10 volts
So software can be very innacurate


A MM is the best.

 

[Bobnova]

Get a multimeter, there's a guide in my sig.
If your PSU really is putting out those values, it is very dead indeed and needs to be replaced NOW.

It can be a cheap pile of Harbor Freight multimeter, they're more than accurate enough and cost a whopping $7 when not on sale, and $2 on sale.

Software is off more often than it's correct, and way off quite often as well.
The BIOS is usually closer than the numbers in the OP, so that is concerning. I've seen it pretty far out to lunch too though.

 

[bayflats]

Hi All, many thanks for your guidance.

I will head to the shop and pick a MM up and report the numbers I get. Will also in the meantime check the inside of the case and screws for any of the instances as mentioned by BobbyBubblehead.

Tks once again.

 

[Enablingwolf]

It should be the first test, honestly, seeing as how software is usually quite bad at reading it.

To make any decisions without testing it properly first is premature at best.

It's not that any software is terrible or bad.

The resistance via power supply path to the sensor. Can vastly skew a soft reading. So as you've already wisely suggested. Take a direct hard reading.

Any mass produced item, can and does have imperfections. How much will it be magnified...

 

[Bobnova]

Software has to know what the sensor is as well. The sensor reports a value 0-255 or 0-1023 or something like that, the software has to know how many volts reads as zero and how many volts per step the sensor outputs. The software also has to know the SMBus address of the sensor, if it reads 0x55 and the sensor is actually 0x83, you're going to get some interesting readings. Applying the "proper" voltage formula to 0x55 when it's the current status of fan header #4 isn't going to give you the 12v rail voltage.
That's why HWMonitor and Speedfan and such come up with such outlandish voltages sometimes.

The BIOS is out to lunch because the sensors are extremely cheap and inaccurate.

 

[bayflats]

Hi All,

Got a multimeter and here are the readings:
+12v = 12v
-12v = 12v
+5v = 5.3v
+3.3v = 3.6v

All seem OK and within tolerances. So that's a relief!

Also re-checked screws and re-oriented all the wiring inside. Lets see how we go.

Thanks for the help. Anything I can do to stress the system? Prime95?

 

[Silver_Pharaoh]

Hi All,

Got a multimeter and here are the readings:
+12v = 12v
-12v = 12v
+5v = 5.3v
+3.3v = 3.6v

All seem OK and within tolerances. So that's a relief!

Also re-checked screws and re-oriented all the wiring inside. Lets see how we go.

Thanks for the help. Anything I can do to stress the system? Prime95?

PSU is NOT within spec.

5 volt rail MAX is 5.25 volt.
3.3 rail is 3.465 volts MAX.

And I really hope you meant to type -12 Volts for the -12 volt rail!

Sooooooo



EDIT: Did you measure during full system load (GPU + CPU)?

 

[Ben333]

5.3 is a bit high, but I've seen worse. Your MB has many on board regulators anyway...

And -12 was used for RS232... wayyy back. Now I think any standard motherboard can make -12V on it's own, without a lead from the PSU. So I wouldn't worry about that one.

But as everybody above said, ALWAYS confirm with an actual DMM, cause I've often seen 12V rails read out anywhere between 8 and 14 volts in bios / software. Right now, with my good one MIA I have an $8 special from the local dollar store. Works fine, accurate enough for most of the stuff I do. Came with batteries! lol

 

[Mr.Scott]

PSU is NOT within spec.

5 volt rail MAX is 5.25 volt.
3.3 rail is 3.465 volts MAX.

And I really hope you meant to type -12 Volts for the -12 volt rail!

Sooooooo



EDIT: Did you measure during full system load (GPU + CPU)?

PSU is fine.

 

[Silver_Pharaoh]

PSU is fine.

I stand corrected.

 

[Bobnova]

Yeah that's all OK. 5V is a bit out of spec high, but not to the point where it's likely to cause issues.
3.3v is right at the high spec, but again that should be OK.

 

[bayflats]

Thanks for the help.

One thing I should mention was this was all measured at full idle (I disconnected the PSU from the PC and just powered on with the MM jacks)- I am reading the MM guide contained in Bobnova's signature which will instruct me on how to use the MM with the system under full load with P95.

Last bizarre thing is that once I reconnected the hard drives and rebooted the machine, everything went fine except that a lot of my game icons on my desktop have disappeared and any games contained within the Windows Games location do not load at all. Also when I tried to replay any of my Steam games (or any others), it acted as if it was the first time I was playing them and started re-initiating DirectX setups etc, though my savegame data is all fine.

Will report the MM findings with machine on full load.

Thanks again.