24pin ATX melted with after installing SLI

[xivlia]

hi there, i have recently bought another GTX 480. and i added it to my rig yesterday, it was working perfectly fine untill today. my pc randomly shut off while playing Dragon Age: Origins. and it would not come on when i pressed the power button. so i left it there for about 15mins. and tried again and it powered on, but it wouldnt POST. all it would do is. power on for 3 seconds and power off for 3 second and then power on for 3 seconds and so on. and it would do this forever untill i turned off the PSU at the back. i have a be quiet dark power pro 1000W and thought the gtx 480 sli was darwing too much power. so i discconected everything and took the 20+4pin ATX cable out and on the 4pin part. one of the pins has litterally melted. so im guessing the motherboard is also screwed. however. when i take off one of the gtx 480 and connect the ATX cable it works even though one of the pins have melted. it still shutts off randomly from time to time though. and i have checked both of my 480's one by one and they are also working fine. but when it comes to SLI it just wont post... another thing is that everytime i take a component out and put back in, my BIOS settings get reset for some odd reason. so is my PSU not working or is my Motherboard not working?

oh and my CPU was NOT OC'ed at the time

here are some pictures. off the damage

[zalbard]

I would RMA both of them. You're lucky you haven't fried your rig.
I'm guessing your PSU is the current issue, though. But you can easily replace it and have the same issues with the new one on the same board...
That's what happens when a board doesn't have an auxiliary power connector (usually Molex) for PCI-E.

[zanzabar]

the 480 pull more than the 75W per slot limmit on the pci-e standard so u need an over engineered board or added power connectors (they pull all the slot power from the 24pin otherwise.) i cannot tell what board u have but the evga power boost connector should help http://www.evga.com/products/moreInf...0Hardware&sw=4

or your psu was not enough as zalbard said, u have a 6 rail, so u most likely broke the OCP since it looks like u have 20A on the 24pin so thats 240W and may have topped that.

[xivlia]

hmm okay my PSU has x6 12v rails at 20A each, my psu is one of the higher end PSU's so i dont think that is the problem. i have the ASUS P6T v1 and it does not have an auxiliary power connection. which is the problem i think

[TJ TRICHEESE]

I would RMA both of them. You're lucky you haven't fried your rig.
I'm guessing your PSU is the current issue, though. But you can easily replace it and have the same issues with the new one on the same board...
That's what happens when a board doesn't have an auxiliary power connector (usually Molex) for PCI-E.

This, I have 4 high end cards in my rig and i haven't burnt my 24 pin connector (original P6t7 without extra molex connector).

[little_scrapper]

Melting pins is caused by bad connections. Trust me, the small traces on the board are not anywhere near the current carrying capability that the wire has. But, when you have a CONNECTION, the high resistance caused by two dissimilar metal planes of a bad connection generates heat.

It is caused by current. Its just that the weakest point is where the highest resistance connection was and that was at the connector. I guess what I am saying is that we may be entering an age where we may have to, just like when you change your cars spark plugs, use a conductive gell to ensure a rock solid connection that cannot corode.

[sin0822]

i think you just have a crappy PSU, i think your board is fine, either way id buy a new PSU then test it in your board and if this crap happens again RMA the board.

[B-Shot]

Melting pins is caused by bad connections. Trust me, the small traces on the board are not anywhere near the current carrying capability that the wire has. But, when you have a CONNECTION, the high resistance caused by two dissimilar metal planes of a bad connection generates heat.

It is caused by current. Its just that the weakest point is where the highest resistance connection was and that was at the connector. I guess what I am saying is that we may be entering an age where we may have to, just like when you change your cars spark plugs, use a conductive gell to ensure a rock solid connection that cannot corode.

+1 had it happen 2 times in the past was just a poor connection psu is still in use today, just check rails with a multimeter.

[thebad]

This is actually a well known problem with SLI and early motherboards. The fact that it is actually a CPU feeding connector would rather indicate a connection problem (poor contact) but if you had the problem only after you implemented SLI, chances are, the motherboard was trying to pull more than the connector was rated for to feed the video card. This is the reason why many recent motherboards include a separate connection for the PCI-Express slots and also why eVGA released their Powerboost gadget. You should look into it.

[zsamz_]

i seen that on the p6t7 supercomputer and tri-sli

you gpus just suckin too much power from pci-e slot
and possibly your psu not givin enough juice from 6-8 pins

[xivlia]

i think you just have a crappy PSU, i think your board is fine, either way id buy a new PSU then test it in your board and if this crap happens again RMA the board.

i can assure you that my psu is NOT crappy its one of the msot high end PSU here in europe, you havent heard of this as its not sold in the US. i payed £200 for this psu. and its 92% efficient. so its not crappy

[TJ TRICHEESE]

i can assure you that my psu is NOT crappy its one of the msot high end PSU here in europe, you havent heard of this as its not sold in the US. i payed £200 for this psu. and its 92% efficient. so its not crappy

It could be broken though, which would cause it to become crappy.

[xivlia]

It could be broken though, which would cause it to become crappy.

Well that's true, fair enough. How can I tell that this is what's wrong?

Or could it just be a loose connection ? I've cleaned the burnt out connection on the motherboard and stripped the melted plug (so the metal connector is exposed). Now it fits better and the pc has been on overnight with no problems, yet.

Also, there is a hissing sounds. I think its coming from the psu but can't tell exaclty. Could this a capacitor leak ? Is this a major problem ?

[Olly_K]

what PSU is it ? if it broke OCP it should shutdown to prevent damage, no ?

I have run sli on 295's, 480's and 580's off my Fractal 1kW PSU and never used the extra molex connector on the motherboard, and I've been totally fine

[xivlia]

what PSU is it ? if it broke OCP it should shutdown to prevent damage, no ?

I have run sli on 295's, 480's and 580's off my Fractal 1kW PSU and never used the extra molex connector on the motherboard, and I've been totally fine

It's a Be Quiet Dark Power Pro 1000w. In the manual it says there is kinds of protection: over current, over voltage, etc.. I don't anything is damaged other than the socket on the mobo and possibly the psu itself.

[TJ TRICHEESE]

I think you may need to split the load around the rails more

[Olly_K]

interesting, I had that exact PSU before and had a few problems with it, it would shutdown the entire system when I loaded the first level of crysis. Changed to the fractal and had no such problems. I would try another PSU and take it from there

[xivlia]

I think you may need to split the load around the rails more

I'm not sure about that. My psu has 6 12v rails each capable of 20A. The gfx cards each had their own rail. So this kinda seems unlikely to be the problem.

[some_one]

If you have a meter you could try to measure the resistance between pins 10 & 11 (with power disconnected and 24pin connector disconnected) on both the PSU side and MB side to see if pin 10 is open circuit or high resistance. If so, this could lead to one pin taking all the load instead of it being shared between the two as designed. Just a thought.

I did try to make a drawing but gave up so hope this helps, MB side.

[markwrc]

tbh 20A per rail isnt much, only 240w of power, which is just enough to power the 480's PCI-E, my Antec is well over-engineered at 48A per rail, i think that the loading is your problem, 20A just isnt enough

[zoob]

Take some tweezers and try squeezing the motherboard connector on the PSU. It's a square receptacle and looks something like [].

On some of my older PSUs, it's gotten so loose that I could reboot the PC by tapping on the case. Tighten this up and it should help...

[xarot]

Huh, I ran 3xGTX580s on P6T6 WS Revo with 950 W Corsair PSU for some time...nothing like this happened! The P6T6 does not have an extra molex connector on board. Now running them in R3E in fear of this though... and AX1200.

[CedricFP]

You ran 3 580's fine on a 950W Corsair? What was your CPU clocked at?

[xarot]

I ran them to troubleshoot while waiting for my other parts to arrive...my CPU was clocked at 4.2 I think, I experienced shutdown in Furmark.. after that I didn't do more Furmark..

[CedricFP]

I see. The Furmark information is useful, though I'm surprised OCP didn't kick in? Did you unlock with Rivatuner, or did PC shut down without the unlock?

I may have bitten off more than I could chew in promising to buy a pair of 580s, but I've already got one so that puts me in a pickle as I've only got a Corsair HX1000W but wouldn't sell off one of the cards if I already had them...


...