Really not trusting PSU - what to get?

Soldato
Joined
8 Nov 2006
Posts
9,237
Have had 2 GPU's die (7950/7970) this year already, neither pushed extremely hard, and now with the remaining 7950 in my system, I used to easily get 1125 with 1.174V, but now can not even achieve that with 1.2V and don't want to go higher.

I am also sure that when I first got my 2500K, it wasn't hard to get above 4.5GHz, but didn't try for all that long, as 4.5GHz was good enough, but now getting above 4.5GHz is a bit of a struggle.

Had sleep resume issues with X79 and 3930K when OCed as well.

It's not a cheap PSU and had good reviews, but I just don't trust it right now...

So?
Corsair AX 850W - £154
Silverstone Strider 850W gold - £143
OCZ ZX 850W - £124
Corsair HX 850W - £122
Antec High Current Pro 850W - £120

I know the general consensus is the more expensive the better, but is there really any point in going for, say, the AX over the HX, or the Silverstoen over the Antec?
 
The sleep resume issue you mention is most likely AMD driver related rather than your hardware (CPU/board). Been seeing quite a lot of people with 79xx cards reporting having the same problem.
 
The sleep resume issue you mention is most likely AMD driver related rather than your hardware (CPU/board). Been seeing quite a lot of people with 79xx cards reporting having the same problem.

I know of that particular problem, and it has been solved in driver update. These were completely different resume issues.
 
Thermaltake Toughpower 875W 80+ Bronze

That is a decent PSU. With 72A available on the +12v it should be more than up to the job. If it's faulty could you RMA it and get a replacement as tbh it's as good if not better than the ones listed... ?
 
Yeah, I know it is/should be decent, but all the issues have had/am having...

Just, how do I know for certain? Voltage readings always look good in monitoring software.

Maybe I just remember things incorrectly with regards to clocking the 2500K, but 2 dead GPU's (both which happened at power down/reset), and now a third starting to look dodgy...
 
You could connect a multimeter to the 12v line and monitor under load for any fluctuations. That psu has so many safety features I would be surprised if the psu was causing the issues, killing a GPU takes quite a bit of effort unless the psu is spiking badly at re-set or power down.
Saying that it could be faulty which could cause problems. I would get a meter on the 12v line and do a few shutdowns, re-starts and see if there are any obvious jumps in voltage.
 
Have sent an email to thermaltake.

In mean time, will find a cheapo multimeter, but will need help figuring out where to stick things - which is sad considering I did electronics in school!
 
To just give a bit more detail on when the cards died...

Both cards died on previous X79 platform.

Card 1: Direct CU II 7950 Crossfire primary card
Was during a restart. Suddenly system powered down and wouldn't power up again. Finally worked out which card it was, and it was only when power cable attached to the power connector furthest from the back of card.

Without a cable attached to that power connector, system would power up, but obviously would not boot up without 2 power cables attached. Tried different cables too though, so something specific to that power connector is what went.

Card 2: Reference 7970 reference primary card.
After shut down. Was in BIOS, and turned system off using power button.

System would power up with card connected, but wouldn't post or anything. Wasn't a power connector issue this time. Simply would boot.

Current card being used on 2500K system.

The current card which is no longer clocking as well is the remaining 7950 DCU II.
 
Have decided to simply purchase AX850 in mean time.

Am getting new GTX 670's and 3770K, and in case the Thermaltake is causing issues, I didn't want to chance using new stuff with it.

Will still test it, and will either send back if I find fault, or sell on if I don't.
 
Got the multimeter, but just not sure I got the will/desire/guts to start sticking probes in places while things are plugged in to test if this thing is spiking at shut down/start up/reset....
 
Is it ok to just test on a molex connector?

Have been doing that and voltages actually seem fine.

Tried restarting, shutting down, starting up, and about the only thing worth noting is that it takes a fraction of a second to get to 12V (I can see it go around 11.2/11.6) from cold start or resume.

I don't need to test against the motherboard power connector do I?
 
I have now also tested against the mobo 12V connectors, and also for the sake of being thorough, both pci-e power cables attached to main gpu (only one wire on each cable though).

Doesn't even seem to be any fluctuation whatsoever on the motherboard connectors between 12.0/12.1V. Its just a solid 12V

The only thing worth noting is what I pointed out earlier in that from power up or resume, it takes a fraction of a second to get to 12V. Just once I saw 10V for just a moment, but every other time it's been a fraction of a second anywhere between 11V an 12V beofre settling at 12V.

It really is jsut a fraction of a second though, and not sure if it should be instantly at 12V or not.
 
A meter wouldn't show you the ripple, which could be causing your gpus to die, could be caused by bad caps etc.. a dmm will show the voltage regulation to be just fine, you'd need a scope to check for ripple.
 
Back
Top Bottom