Weird PSU Voltages...is this OK?

[Oversized Rooster]

I have noticed lately that my PSU is putting out weird voltage at -5V, +5V and +12V.

It's a PC Power & Cooling Silencer 750W PSU. I bought it new about a year ago and it's been running fine.

Here is what the PSU is running:

Intel Core 2 Quad Q9450 @ 3.8GHz 1.33V
Asus P5Q-E Mobo @ 1900FSB
2 x 2GB OCZ Reaper HPC DDR2-1066 @ 1144 2.18V
1 x WD 300GB VelociRaptor HD
3 x WD 500GB Caviar SE16 HD
2 x 512MB Sapphire Radeon HD 4850 Cards @ 700/2100
Auzentech Prelude X-Fi PCI Sound Card
20X LG DVD-RW
Floppy Drive
5 x 120mm case fans
1 x 200m case fan
4 x USB power-drawing devices

What worries me most is how the -5V is at -0.17V. And also by how much the +5V and +12V are off. The +5V is at 5.56V and the +12V at 12.46V.

Also, is the -5V supposed to be -0.17V? I think there is something horribly wrong here. Isn't the -5V supposed to be -5V?! I mean that's what it's called, ain't it? :lol:

So given my 11.2% +5V overvolt, 3.8% +12V overvolt and 96.6% -5V undervolt, should I RMA? Should I get a new PSU?

The picture below is when the PC is idling. When it's at load, the CPU Core voltage goes to 1.33V, but the weird -5V, +5V and +12V are still the same.

http://gergin.net/posts/PSU%20Idle.JPG

[johnsmit90210]

Go get Everest Ultimate 4.60 and try monitoring again. If that doesn't do it, try a different monitoring application such as NextSensor to compare the results. It supports Winbond I/O LPCs. The other thing that I highly recommend is buying a good power supply tester; Something that will test +3.3V, -12V, PG, +5VSB, +12V, -5V, +5V outputs. This is kind of hypocritical of me because I don't even have one (not yet anyway-I've been meaning to buy one for years). It could be that the PSU has developed a bad 'rail'. Or who knows, it could be that the sensor is defective or doing bad reporting--in which case the physical tester would provide some clarity.

Also, (for some beneficial insight) traditionally voltage flux on a PSU is/was completely normal. That being said I don't know how normal a negative voltage is (sounds like a software blip to me). Quality of a PSU used to be in part measured by the units ability to provide steady 'reliable' power. Voltage is supposed to be thought of in terms of 'pressure' on a wire. That pressure or voltage wavers regularly anywhere between 100V and 130V on a 120V circuit. Simply put, power is what they describe as 'dirty'. But that dirt is part of what's considered normal. This is where a power or line conditioner can come into play to provide voltage regulation among other benefits. Brands such as APC or a Monster power bar/UPS (what have you) can condition the power by providing better power quality. Read this: http://en.wikipedia.org/wiki/Power_conditioner

Check your existing surge bar or UPS specifications to see if you've already got some of these features. Nowadays they are becoming more common. APC is #1 in my books for being power fanatics. They implemented noise supression/voltage regulation in a variety of their products long before Monster came along and incorporated 'stage X clean power', so I would stay with them if you're thinking about picking up something. All that being said it's quite possible that a high end PSU already has incorporated these advanced features in attempt to create a good product. There are so many damn possibilities in this world. Just remember it's all in the specifications.. =)

Let us know how it goes and what you find out!

[KorruptioN]

Good post^

All I wanted to say is that software voltage readouts are not to be trusted.

[movieman]

Is the -5V line still used for anything in a modern PC? I'm pretty sure that my Antec PSU has no -5V supply; I know the voltage monitoring program that came with the motherboard complains about a negative voltage line being out of spec with a measurement around -0.1V.

Either that or maybe I forgot to plug it in :).

[matkun]

Yep, just because it's giving a bad reading does not actually mean there is a problem. Those are more or less normal +- 10% fluctuations, which most components are designed to expect.

Has your computer been flaky? If not, I wouldn't worry about it.

Also, don't worry about the -5V rail. It's not used for anything nowadays, so ATX12V 1.2 spec and higher dropped it. I.E. your PSU probably doesn't even have a -5V rail.

See here:
http://www.dansdata.com/io053.htm

[toalan]

I would not worry about it, the program is probably wrong. Even if it is right as long as your comp works without problems then you are golden. To be sure you should use a good quality multimeter and see what the voltages really are.

[i6s1]

Twice in my life I've noticed voltage readings around 10% after the computer was acting flakey. Both times I replaced the power supply, and both times it fixed the problem. My voltages were also bouncing around, they weren't stable. After I replaced the second one (a 5.5 year old Antec 350) I noticed that one fan blade was broken, and one of the MOBO connector pins was black, and the plastic around it was melted.

[KorruptioN]

Is the -5V line still used for anything in a modern PC?

No - it was removed in the ATX12V v2.0 spec.

[Oversized Rooster]

Guys thank you very much for the comments and suggestions. I've looked into the topic a little further.

I also noticed that the -5V has been scrapped by the more recent ATX specs and nothing "uses" this.

I've also installed Sandra which shows a completely different story in the voltages. Sandra in fact shows that my +5V is around 4.8V and the +12V is at 11.9V.

I also went into the BIOS and there things looked to be within 5% of the correct values. Therefore, as you said - software readings are not to be trusted.

I bought this PCP&C PSU exactly in the hopes that I'm getting a quality product. I guess everything is fine. My computer has NEVER given me any symptoms of power problems (I've had to deal with this a few years back with a bad Antec PSU). So all is well. :cheesygri

[balou911]

buy/borrow a volt meter, and open up your case, and take some measurements. thats is the only way to get accurate readings.

[johnsmit90210]

Everest seems to have trouble properly reading mulitple versions of the Winbond W83627* sensors. I don't know what versions of Everest etc. or other details, but if you google "Everest Winbond W83627" or variations of, you'll get postings about incorrect readings.

I don't know what it is that actually reports active PSU voltage to the system; If someone knows please post. There are thermo-couplers, diodes, thermoresistors, different embedded sensors types etc. To completely eliminate the whole chain of dependancy and get a true and accurate reading for any power supply, I think the no fail method is to do as the above poster states. Get a good multimeter/multitester and see what's coming out first hand. I gotta get one of those too. haha!

[Oversized Rooster]

Everest seems to have trouble properly reading mulitple versions of the Winbond W83627* sensors. I don't know what versions of Everest etc. or other details, but if you google "Everest Winbond W83627" or variations of, you'll get postings about incorrect readings.

I don't know what it is that actually reports active PSU voltage to the system; If someone knows please post. There are thermo-couplers, diodes, thermoresistors, different embedded sensors types etc. To completely eliminate the whole chain of dependancy and get a true and accurate reading for any power supply, I think the no fail method is to do as the above poster states. Get a good multimeter/multitester and see what's coming out first hand. I gotta get one of those too. haha!

It seems that in version 4.6 they fixed the Winbond issues. Here is what version 4.6 reports:

http://gergin.net/posts/PSU%20Load%20New.JPG

[infamouskid]

multimeter the damn thing. if you really want a definite answer.
but personally i had issues like this before. alli did was purchase a brand new mushkin psu and voila all problems gone. go reliable brand psu. dont buy the cheap stuff. SAY NO TO POWER UP PRODUCTS!! or any other chinese brand.