3.3v rail not at spec - could this cause boot/GPU probs?

[MichaelD]

I have a Corsair HX1000 PS. All the rails are spec except the 3.3v rail. It idles at 3.28 and under load goes down to 3.26v. All other rails stay at spec under load. I have no system instability once it's up an running. Here's the problem:

I have a 4870X2 GPU; a power-sucking beast of a card. For the past couple of months, the MB will not boot most of the time with the 4870 in there. It will boot with a 8800GT or other lower-powered card. I have tried three different motherboards (getting tired of swapping them out!) and the problem is present with all three MBs.

This means either my GPU is bad or the low 3.3v rail on my PS is causing the problem. I'd like your opinions before I either order an expensive new GPU or an expensive PS. Thanks.

 

[blastingcap]

3.26V is well within 5% of 3.3V (it's about a 1.2% deviation) and thus it is unlikely to be the source of your problem. If it's outside of 5% then it's out of spec.

I have a Corsair HX1000 PS. All the rails are spec except the 3.3v rail. It idles at 3.28 and under load goes down to 3.26v. All other rails stay at spec under load. I have no system instability once it's up an running. Here's the problem:

I have a 4870X2 GPU; a power-sucking beast of a card. For the past couple of months, the MB will not boot most of the time with the 4870 in there. It will boot with a 8800GT or other lower-powered card. I have tried three different motherboards (getting tired of swapping them out!) and the problem is present with all three MBs.

This means either my GPU is bad or the low 3.3v rail on my PS is causing the problem. I'd like your opinions before I either order an expensive new GPU or an expensive PS. Thanks.

 

[MichaelD]

Thanks, blastingcap. So, I guess it's the GPU then? Damn. It's always the expensive part. LOL! This was a replacement card for the orginal 4870X2 that died on me. "Factory recertified" products are crap 90% of the time, IME.

 

[Modelworks]

The video card powers off the +12V . If the board will not boot sometimes with the card installed I would still look at the PSU. The reason is that when a pc starts it demands a large amount of current, more than any other time when it is running except maybe with everything maxed at 100%. HDD pull max amps at power on putting even more strain on the PSU.

Unfortunately measuring power quality at power on is not an easy thing to do without some fairly expensive equipment like an oscilloscope.

 

[MichaelD]

I definitely have been considering that exact scenario, Modelworks (nice avatar, BTW!). The Corsair HX1000 is a pretty beefy PS. I had it running a dual-quadcore-Xeon workstation with 10 hard drives and it didn't flinch at all.

When I first started troubleshooting this problem I actually did the standard "MB won't boot, so I take out everything except the videocard, CPU and 1 stick of RAM"thing. So, I had no HDs (instead of 3), 1 stick of RAM (instead of 4) and no PCI cards. It still wouldn't boot. So I sort of tried minimizing the bootup electrical load already.

I've tested the PSU at idle, out of the case with a DMM and the measurements line up with those of Windows-based utilities like CPUID's "Hardware Monitor." It kills me to think this expensive PS isn't cutting it. Maybe I'll post up on Corsair's forums and see what they say.

 

[doubleOseven]

Michael, where are you getting the voltage readings from? If you're not already, get the readings directly from a digital multi-meter, as BIOS readings are known to be inaccurate.

 

[Computers_Plus]

The variance on the 3 volt rail is only 2 tenths of a volt not enough to cause an issue.

Do you have the computer plugged into a UPS or power strip or directly into the wall?

 

[xd_1771]

The HX1000 could power 2 4870x2's with your CPU quite easily; I don't see how there's a problem.

 

[Computers_Plus]

Even with the best power supply available if you feed it dirty AC you can have trouble I see it all the time with old houses with bad wiring and the cheap power strips and UPS systems they can introduce a lot of line noise.

The computer power supply is very sensitive I would suggest using an AVR (auto voltage regulator) type power outlet to keep AC voltages regulated so the power supply is fed a steady amount of power.

 

[MichaelD]

At Corsair's suggestion I used a DMM right at the 24-pin connector while the system was running. All rails are above-spec. I was pretty shocked! I thought there was no way that the BIOS could read wrong, but it does. All the rails are above-spec and stable, according to the DMM. Now, what I don't have the answer to is the question: Is the PS healthy enough that upon startup it can maintain a healthy 12v supply to the GPU and the MB? (as Modelworks suggested, above).

Right now there's a 7900GTO in my gaming box; not exactly a big power-sucking card. There are no issues at all with it. I'll be buying another GPU in a few weeks. Hopefully, my PS is OK. No way to tell until then (unless you guys have suggestions?)

BTW: The PC is plugged into an AVR power supply. Wall power sucks and I don't trust it at all.

 

[fffblackmage]

How far above spec are the voltages? The voltages are almost never going to be exactly 3.3V, 5.0V, or 12.0V.

 

[Modelworks]

The way motherboards measure voltage has been poor for years. Intel is one of the few who get it right most of the time. The rest just put a voltage monitor on the board and it may be all the way across the board from where the power enters the board. By the time it gets to the other side it isn't close to what it was at the wire. They also tend to use low accuracy chips which are best for determining if there is 5v present if you don't care if it is 5.5 or 5.1



Pull the card and under very bright light look for cracks in solder , poor connections, etc. Look it over carefully around the power connectors , the slot tab, both back and front.

Solder cracks are well known to cause hard to find behavior. They make or break depending on temperature, pressure on the board, angle , or just day of the week