Jump to content
Corsair Community

HX1000W high defect rate?


permi

Recommended Posts

Hi,

 

I was wondering when Corsair was planning on doing something about their HX1000W model. I have found approx. 1/2 to 3/4 of all HX1000W power supplies that I test are defective. I am talking about personally testing and finding more than 30 defective HX1000's.

 

The problem lies in the PCI-E voltage. If you have a GPU connected using 1 6 or 8 pin pci-e power connection on one rail, and the CPU 4 or 8 pin power connected to the other rail, the PCI-E voltage output by the PSU will be too low. If the CPU power and GPU power are both connected to the same rail, either 12v1 or 12v2, then the PCI-E power is output at the correct voltage.

 

Not all the units are defective, some work fine, but from what I have personally experienced building systems over the last 1+ year with these PSUs, is that over half have this issue.

 

The computer may appear to function ok with one of these defective units, however you may get cold boot issues, where it takes several attempts to get the computer to post. Also may have issues with crashing when the GPU is under full load.

 

I have also had systems come in for repair that no longer post at all with these defective units, only to find that the incorrect PCI-E voltage has killed the video card, so both the PSU and video card had to be replaced.

 

I have contacted Corsair reps about this before, but have been told that "rma rates on this unit are normal".

 

If you have one of these units, please test it and post if yours has this issue, and maybe Corsair will finally fix the problem.

Link to comment
Share on other sites

  • Corsair Employees
Hi,

 

I was wondering when Corsair was planning on doing something about their HX1000W model. I have found approx. 1/2 to 3/4 of all HX1000W power supplies that I test are defective. I am talking about personally testing and finding more than 30 defective HX1000's.

 

The problem lies in the PCI-E voltage. If you have a GPU connected using 1 6 or 8 pin pci-e power connection on one rail, and the CPU 4 or 8 pin power connected to the other rail, the PCI-E voltage output by the PSU will be too low. If the CPU power and GPU power are both connected to the same rail, either 12v1 or 12v2, then the PCI-E power is output at the correct voltage.

 

Not all the units are defective, some work fine, but from what I have personally experienced building systems over the last 1+ year with these PSUs, is that over half have this issue.

 

The computer may appear to function ok with one of these defective units, however you may get cold boot issues, where it takes several attempts to get the computer to post. Also may have issues with crashing when the GPU is under full load.

 

I have also had systems come in for repair that no longer post at all with these defective units, only to find that the incorrect PCI-E voltage has killed the video card, so both the PSU and video card had to be replaced.

 

I have contacted Corsair reps about this before, but have been told that "rma rates on this unit are normal".

 

If you have one of these units, please test it and post if yours has this issue, and maybe Corsair will finally fix the problem.

 

Hi Permi,

 

Can you list your hardware configuration so we can attempt to duplicate this issue? On our HX1000 unit we've never seen the 12V rails drop below the ATX specification - and never has this happened in any of the reviews of the unit, either.

 

If you can list the hardware (CPU, Motherboard, Video Card, other hardware) and how you're testing the 12V line (Multi-meter? BIOS readings?) and what numbers you're seeing, we can attempt to duplicate it, and when we duplicate it we can do a root cause analysis and get it fixed.

Link to comment
Share on other sites

Sure. Unfortunately I just recently broke my multimeter, I will try and pick another one up, probably not until friday though.

 

The problems have been seen on so many different configurations that I have eliminated the other hardware as a problem.

 

You will only see a problem when you have single cpu systems with a discrete video card that requires an extra 6 or 8 pin power connector, and only if the cpu and pci-e power connectors are from different rails. This is the most common configuration, as the 24 and 8 pin connectors that are hardwired in are on 12v1, and the PCI-E cables that are hardwired are on the 12v2.

 

Although I don't currently have a functioning multimeter, I have two different brands of psu testers. One is the Antec model, the other is a generic. Both give me:

 

PG: 990ms (reports out of normal range)

+5v: 5.1 (normal)

+12v1: 12.0 (normal)

+3.3v: 3.3 (normal)

-12v: LL (error)

+12v2: 12.0 (normal)

5VSB: 5.1 (normal)

 

This is with only the non-modular hardwired cables plugged in, testing the 24 pin, 8 pin cpu, and one of the 6 pin PCI-E power cables.

Link to comment
Share on other sites

Of course the model of the psu tester would be filtered, I should have realized that. Anyways, that first psu tester is made by a competing company.

 

and again I should note that I have seen these same results on many many units. I have religiously tested every one of these that I have sold for over a year now because of this. Not all are bad, but way more than there should be.

Link to comment
Share on other sites

  • Corsair Employees
Ok, on this particular defective unit, the two yellow +12v connectors read 12.00 and 12.01, and the blue -12v fluctuates between 11.77 and 11.78.

 

Should I test any others?

 

Yellow are the +12V power wires you should be concerned with for video cards - anything else is largely unrelated.

 

+12V has a +/- 5% ATX spefification, meaning that anywhere between +11.4V and +12.6V it is within acceptable ranges.

 

-12V has a +/- 10% ATX specification, meaning that anywhere between -10.8V and -13.2V it is within acceptable ranges.

 

Your multimeter shows that the PSU is performing up to ATX specifications.

Link to comment
Share on other sites

Ok, giant brain cramp. I had my red/blue multimeter cables plugged in backwards.

 

-12v (blue wire): 11.74 - 11.75

+12v (yellow wires): 12.02

 

yellow 12v on 8pin (hardwired, so 12v1 rail) cpu connector: 12.02

 

yellow wires on 6+2pin pci-e connectors (12v2 rail): 12.12 - 12.13

 

yellow wires on 6+2pin pci-e connectors (12v1 rail) - 12.01

 

So technically within spec ... but I have seen power supplies with fluctuations of + or - .1 v cause stability problems before, and with this brand new high end corsair psu, I would expect a better voltage output on all 12v rails than 12.12?

 

edit: I noticed Wired asked above what model of multimeter, it is a Mastercraft 52-0060-2 Digital Multimeter

Link to comment
Share on other sites

On any system that is otherwise functioning normally, there is no way a fluctuation of .1v, or in this case from 12.0 to 12.12v going to cause instability. If a fluctuation that small results in instability, most likely the Vreg on the MOBO is hosed. And, since your meter was broken, I can only assume you saw these fluctuations using software voltage readings which are notoriously unreliable and cannot be relied upon.
Link to comment
Share on other sites

Yeah, I suppose a difference of 1% should never cause any real problems ...

Actually my multimeter itself wasn't broken, one of my cables was, so I had to replace that cable. These results were taken using a multimeter.

 

Do you have any idea why these units are giving me an error on every psu tester that I have used? It's not like it's every single HX1000w gives me an error, and why I do seem to see boot up issues and video card issues on the units that give me those psu tester errors?

Link to comment
Share on other sites

Basically what I'm asking, is why when I use a PSU tester on this, it tests fine with the 24pin and 8pin plugged in, but when I then also plug in the 6pin pci-e from the other rail into the tester, it suddenly gives me a -12v and pg error?
Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...