Low voltages PSU, dangerous?

[FerK]

Hello, I do have the PSU listed in my specs at my profile. I remember when I bought it I had perfectly fine voltages and that's with me using a R9 290. However I realized a few days ago that my voltages were a little bit low, and considering I didn't change anything and I should be using less power now with a GTX 1060, it seems strange.

These are the voltages. As you can see they're all pretty low and the 3.3v is outside the ATX tolerance specs (I think the minimum is 3.13v or something). They aren't too high on idle neither. I have contacted XFX for a RMA (since it's still inside the 5 years warranty) but their answer was a little bit underwhelming. They said I had to pay for the shipment and I could be taxed from 50 to 100 euro for the return shipment from Hong Kong (which is 100% applied if they use DHL or something). And the whole process may take up to 8 weeks.

It doesn't seem worth it and I think I would be better off buying a new unit with a new warranty and from another brand that doesn't have a RMA system like this. However I wanted to ask, if I keep my current PSU, is there any danger for the rest of the components? Can it damage my whole PC?

Regards!

 

[NdMk2o1o]

Test with a multimeter, software readings don't mean squat

 

[EarthDog]

Yep... use a MM of some sort to verify. Software readings are notoriously off. Youll probably find it fine....

 

[FerK]

Thanks for your answers, I think I have one laying around here, I'll look for it this afternoon. In the meantime, if the readings are still wrong but not too wrong (similar to the picture above), is it a major problem to keep using the PSU?

 

[EarthDog]

If its in spec, keep it. If not, get rid of it (sooner rather than later).

 

[IceScreamer]

If you don't have a multimeter check in BIOS first. My readings with OpenHardwareMonitor and HWMonitor differ greatly from BIOS readings.

 

[dirtyferret]

Thanks for your answers, I think I have one laying around here, I'll look for it this afternoon. In the meantime, if the readings are still wrong but not too wrong (similar to the picture above), is it a major problem to keep using the PSU?

Is the PSU causing any issues when you are gaming?

 

[FerK]

If you don't have a multimeter check in BIOS first. My readings with OpenHardwareMonitor and HWMonitor differ greatly from BIOS readings.

In the bios they are pretty much the same.

Is the PSU causing any issues when you are gaming?

I was thinking about this right now. Sometimes my screen turns to black for a second, then comes back. I don't know if it's related to the PSU, probably not, it can be anything. Other than that, I didn't feel anything else.

 

[eidairaman1]

Hello, I do have the PSU listed in my specs at my profile. I remember when I bought it I had perfectly fine voltages and that's with me using a R9 290. However I realized a few days ago that my voltages were a little bit low, and considering I didn't change anything and I should be using less power now with a GTX 1060, it seems strange.

These are the voltages. As you can see they're all pretty low and the 3.3v is outside the ATX tolerance specs (I think the minimum is 3.13v or something). They aren't too high on idle neither. I have contacted XFX for a RMA (since it's still inside the 5 years warranty) but their answer was a little bit underwhelming. They said I had to pay for the shipment and I could be taxed from 50 to 100 euro for the return shipment from Hong Kong (which is 100% applied if they use DHL or something). And the whole process may take up to 8 weeks.

It doesn't seem worth it and I think I would be better off buying a new unit with a new warranty and from another brand that doesn't have a RMA system like this. However I wanted to ask, if I keep my current PSU, is there any danger for the rest of the components? Can it damage my whole PC?

Regards!

When you RMA you pay for shipping, that is how most if not all companies work. It has been that way since 2002 at least. Normal business practice. You also insure the shipment.

 

[FerK]

When you RMA you pay for shipping, that is how most if not all companies work. It has been that way since 2002 at least. Normal business practice. You also insure the shipment.

Yeah, but the return taxes though... They could as well have some warehouses in EU. I simulated this morning a shipping to a random postal code in Netherlands and it was about 25 euro, which is acceptable. For 25 + 50/100 euro, I buy a new one, for products with a price of about 100 euro, it almost invalidates the warranty because I don't think many people would choose this option.

______________________________________________

I'm doing some testing with the black screen for a second. I do think it has something to do with the cpu overclock, I'm trying to raise the voltage so to discard it's a PSU problem, because I remember that with Intel Burn Test, I suffer a black screen too, and I hear the CPU fan slow down during its duration, so maybe the voltage is too low.

However, no good news so far.

CPU OC at 4.4ghz, 1.32v-1.34v --> 3.3v at 3.08v-3.10v and black screen
CPU stock --> 3.3v at 3.13v-3.15v, no black screen

I'm going to increase the vcore a little bit more, my 2500K doesn't like to be overclocked but I think, given the voltage provided, should be stable enough to not to cause a black screen. If I increase the vcore and I still suffer for black screens, can it be possible that the PSU is at fault?

 

[EarthDog]

Your CPU runs off the 12V rail (though all are important) which, according to software (grain of salt) is OK. Its possible the PSU is bad, but I don't think this manner of testing will show it.

Seems to be the overclock was causing some instability if it is not doing it at stock.

Again, you are not reading the voltage with a MM so it is likely not accurate the voltage you are reading.

 

[FerK]

Your CPU runs off the 12V rail (though all are important) which, according to software (grain of salt) is OK. Its possible the PSU is bad, but I don't think this manner of testing will show it.

Seems to be the overclock was causing some instability if it is not doing it at stock.

Again, you are not reading the voltage with a MM so it is likely not accurate the voltage you are reading.

I guess we learn something new every day. I looked for a multimeter but unfortunately I ended up not having one.

 

[EarthDog]

Try loading the CPU only at stock and see if it happens.. if not, add a GPU test in... loop Unigine Heaven while you run CPU stress tests. If it happens then, Id look more into drivers first than the PSU.

Since its difficult for users to PROPERLY test a PSU, see if you can find a known good PSU and put it in there.

 

[peche]

OC video card, little Oc on proccesor, game a bit and list weird experiences, ask a friend for his PSU, do the same and compare results, thats the easiest way to test a psu without tools or knowledge,

Regards,

 

[FerK]

Try loading the CPU only at stock and see if it happens.. if not, add a GPU test in... loop Unigine Heaven while you run CPU stress tests. If it happens then, Id look more into drivers first than the PSU.

Since its difficult for users to PROPERLY test a PSU, see if you can find a known good PSU and put it in there.

OC video card, little Oc on proccesor, game a bit and list weird experiences, ask a friend for his PSU, do the same and compare results, thats the easiest way to test a psu without tools or knowledge,

Regards,

I just tested and it happened at stock too (auto voltage at 1.2v, from 34x to 37x multiplier), with no GPU load. The flame in intel burn test freezes, followed by a black screen, then a fucked up screen (the content but distorted) and then goes back to normal. With the distorted screen I can see Intel Burn Test in [Idle] in the top window and not burning, that's why the fan slows down, because it has stopped. It's strange.

I do have my old PSU around here, I changed it because the voltage was low'ish and I wanted to buy the R9 290 and it wasn't enough (it's 650w but poor brand). I can go and grab my friends PSU too, who just happens to have changed his own, but that'll take more time.

So can it be the PSU? The CPU? Because when this happens, and when I finish Intel Burn Test, it doesn't say is not stable, it says it's all good.

 

[EarthDog]

Run the GPU only and see if it happens...

You need to think a bit and figure out how to binary test and isolate components to find the issue.

It really sounds like a GPU issues be it the GPU itself or a driver if you are seeing borked screens.... that has nothing to do with th CPU.

 

[FerK]

I tried to reproduce the same issue five times (with a reboot between each test) but it worked fine each time. The only change I did was to close MSI Afterburner from the tray. As you can see, I'm taking shots in the dark, but I would say this issue is resolved (hopefully I don't have to correct myself later).

Back to the PSU, as you say, without proper tools...

 

[eidairaman1]

Yeah, but the return taxes though... They could as well have some warehouses in EU. I simulated this morning a shipping to a random postal code in Netherlands and it was about 25 euro, which is acceptable. For 25 + 50/100 euro, I buy a new one, for products with a price of about 100 euro, it almost invalidates the warranty because I don't think many people would choose this option.

______________________________________________

I'm doing some testing with the black screen for a second. I do think it has something to do with the cpu overclock, I'm trying to raise the voltage so to discard it's a PSU problem, because I remember that with Intel Burn Test, I suffer a black screen too, and I hear the CPU fan slow down during its duration, so maybe the voltage is too low.

However, no good news so far.

CPU OC at 4.4ghz, 1.32v-1.34v --> 3.3v at 3.08v-3.10v and black screen
CPU stock --> 3.3v at 3.13v-3.15v, no black screen

I'm going to increase the vcore a little bit more, my 2500K doesn't like to be overclocked but I think, given the voltage provided, should be stable enough to not to cause a black screen. If I increase the vcore and I still suffer for black screens, can it be possible that the PSU is at fault?

Black screens typically are gpu related.

Back your OC to stock though and just run the igp, if no crashes, increase the cpu oc back, if it crashes then its the cpu/board, if not then its the gpu. Check other voltages, tolerance is -5/+5% on psu rails.

 

[FerK]

Black screens typically are gpu related.

Back your OC to stock though and just run the igp, if no crashes, increase the cpu oc back, if it crashes then its the cpu/board, if not then its the gpu. Check other voltages, tolerance is -5/+5% on psu rails.

Hi, I think I can confirm the issue was caused by MSI Afterburner, I tried two more times now, one with MSI Afterburner on, and one with it turned off, and with it working I had the black screen issue, without it everything works as it should. I made this tests with the CPU at 4.4ghz and 1.32v (it's a lazy 2500K, can't go to 4.5ghz without a big jump in voltage).

The voltage issue may be left unresolved I think. I don't have a way to measure the voltage accurately, in fact OCCT reads the 12v rail at 8v or so, so I can't trust software readings. I don't have any apparent issue apart from the black screen talked above, and I'm not willing to spend 75+ euro and be left without a PSU for 8 weeks without knowing if it's at fault or not. It would be hilarious (not for me though) to sent it to XFX and get returned the same PSU because they find no faults. I looked for multimeters and they seem to be a little bit pricey, there are some that are like 'portable' but I don't think those would work for me, because I may be wrong but as far as I know, those can't put a decent load into the PSU, right? Or would they?

 

[Ferrum Master]

Just get ANY MM... Even an ebay 2$ will do.

 

[FerK]

Just get ANY MM... Even an ebay 2$ will do.

I got one that isn't digital, it works like with 'watch handles', move a switch to X volts and plug a red and a black 'something'... Would that work? If not then I'll buy one off ebay.

 

[NdMk2o1o]

I really wouldn't worry if you have found the culprit of the black screen issue (maybe premature to think that's the case but give it some more time to be sure anyway and you'll know one way or another) and no other issues persist, forget what readings software is giving you for the voltages unless you have reason to believe your PSU is an issue.

Just out of curiosity, why did they say you would have to pay taxes on a replacement when it's not like you have bought an item and imported it for it to incur taxes, it's a a faulty item (may or may not be the case, but regardless) that was sent back and returned or a replacement supplied to you, you've already paid the applicable taxes when you bought the original unit, should really be shipping cost which shouldn't run you more than 10-20 euros tops, though the 8 week thing is a deal breaker, so yea, I'd probably do the same and buy another unit from a different mfr if I was in the same position, shitty turn around time and would stop me buying their products again if that was the case.

if it's an analogue multimeter it should still work as intended presuming you know how to use it

 

[Ferrum Master]

I got one that isn't digital, it works like with 'watch handles', move a switch to X volts and plug a red and a black 'something'... Would that work? If not then I'll buy one off ebay.

Yes. They work fine. You do not need 0.001V precission. For this task it is fine.

 

[FerK]

I really wouldn't worry if you have found the culprit of the black screen issue (maybe premature to think that's the case but give it some more time to be sure anyway and you'll know one way or another) and no other issues persist, forget what readings software is giving you for the voltages unless you have reason to believe your PSU is an issue.

Just out of curiosity, why did they say you would have to pay taxes on a replacement when it's not like you have bought an item and imported it for it to incur taxes, it's a a faulty item (may or may not be the case, but regardless) that was sent back and returned or a replacement supplied to you, you've already paid the applicable taxes when you bought the original unit, should really be shipping cost which shouldn't run you more than 10-20 euros tops, though the 8 week thing is a deal breaker, so yea, I'd probably do the same and buy another unit from a different mfr if I was in the same position, shitty turn around time and would stop me buying their products again if that was the case.

if it's an analogue multimeter it should still work as intended presuming you know how to use it

I don't know. The 8 weeks thing is the worst case scenario according to them. I must say I didn't buy this psu from a store, it was used but not registered in their website. I registered it with the purchase date from the person who sold it to me, and, as such, I told the same thing to XFX, that I had no 'proof' of buying for those reasons. Maybe that's the reason for this kind of 'RMA', since they can't verify the date is correct. I don't know, maybe, after all, they were doing a good action, without proof they could have said no, or charge me for a repair/something, but the taxes though...

About the multimeter, if there's something that has been evident in this thread, is that I know literally nothing about psu's. I obviously don't know how to operate it (I've seen a few guides in google and it doesn't seem extremely hard). It's not an issue because if I can't I have someone who knows about this.

 

[TheoneandonlyMrK]

I would be fine with it, thats the minimums the average is fine and the minimum wont hurt anything if enough current is there ,it would crash if it's having issues.