Topic: Burned out PSU - need advices - (Read 2247 times)

hi

that's what i've decided: i bought a 1000 W PSU
and since there's no problem ...

thanks for the confirmation

Here's your problem, sir. You're running your PSU at 85.4% total load, no wonder it burnt out! PSUs are fine with this for a few hours at a time if you're gaming or doing something intensive, but not for mining for any good stretch of time. I'd recommend at least a 1000W PSU if you're going to be mining with three cards.

I've burned a few PSU's mining, but hey. I broke even.

im struggling to find the reliable info on the pin layout.... so far it could be the PSU on. 5V or 3v.... the only way to get any proper answer would be ask corsair or follow the with to the other side and see where it leads...

my conclusion from this investigations is its BROKE... 

If that really was the PS_ON wire then yes. However, the pinout of the PSU connector is proprietary so it may very well be the two 12V lines...

so its the 4th and 5th pin on the top row? which if im correct is the green wire for PSU on and a black ground.. which if im also correct dosnt have anything what to do other than making a circuit from the PSU to the motherboard which also means that its unlikely to be a motherboard or powerdraw issue its more than likely cause by either
A. a major PSU fault during power up
B. a surge during power up

In which case I would just remove the PSU side connectors and just solder the wires directly.

Hi guys

Thanks for the answers, to be exact the mother board connector hasn't burned
this is only on the PSU side

The motherboard connector was also melted. This is most likely due to too much power than because the PSU is modular, though I share your dislike for modular PSUs even though I use a few.

Bad connection lead to high resistance, which leads to heating, which leads to meltdown.

This is why I avoid modular PSUs like the plague.  It's another set of connections, which leads to the possibility of dodgy connections.

PSUs cause so much trouble for everyone and their families.

I could be that the three video cards need too much power from the motherboard. The PCIe spec supports up to 75W per slow, so 3 slots = 225W or ~20A at 12V. However, the motherboard connector has only two 12V pins (the other connector may be used just for CPU power). So 10A per pin = melted pins (and in your picture just the 12V pins melted).

I would just replace the connector on the motherboard and PSU. Then I would probably buy the PCIe extenders with additional power connectors for one or two cards. Or at least solder some 12V wires directly to the motherboard (underneath the connector).

hello


Something very strange happened to my 6 month old Corsair AX 860 PSU
I'm mining steady since months with no problem @ 1830kH/735W av

I've got 3 undervolted + a little overclocked sapphire 7950
with one mobo Asrock970 extreme4 + normal PPCIe risers

But one morning i woke up and the rig was unpowered !

I tried to restart several times but the miner doesn't want to
Sometimes it switch on, worked maybe 1 sec and then off ...
But all the rest seems working

When i disconnected the power cable from the ATX 24 PIN cable from the PSU  
it was like "glued" ... i finally managed to unplug it





I don't understand what happened ...

After some investigations, i've noticed that this kind of problem may come when
too much power is taken from the mother board PCIe ports, but normally mine
can handle 3 x 7950 with no problem ...

example of a typical burned mobo




I noticed everytimes this happens, this is on the mobo side ...
Not on the PSU side connectors like me, does it means something ?


3 days before i had 1s instant down/restart of the whole house mains because of the bad weather
fortunately nothing in my equipments was damaged and the miner restarted without any problem
I think house mains power problems would damage differently the equipment, and i have a APC power surge protector on the miner...

Maybe, the only mistake i made is to forget a common ground for GPUs, PSU an mobo as i got a plastic/wood open case...
 
Maybe some "conductive" dust like metal (f*ck knows why it's here) created some dangerous connections between electronic parts and have made something ... but i haven't noticed anything like this
and but i don't really trust it ...

Or maybe it is a issue due to software, because it can ask the graphic cards to work more than expected
(cgminer, trixx, drivers ?)

One day before, i've just maid a fresh reinstall of SKD and catalyst cause i lost a litttle from my "normal" perfomances ( it worked - normal performances were back ) after a break of one month mining.
And also i tested some new versions of cg miner but i decided to not use them because of no change of perfs or not working at all ... (last version for ex asked me "waiting for a usb hotpug" ??)
 
So i re ordered a PSU + some powered risers, and do a common grounding, but the mystery is still here ... do you have any ideas of what happens ?

thanks