I am sure your 24 pin melt down has nothing to do with your 4+ GHz overclock and inferior power supply selection at the time. Now I suppose this goes without saying, but having the cards at stock and the processor as well you would have never experienced this mishap.

thanks for such great info guys..Exact same thing happened to me and i have a couple of questions.
 first of all, these are my pc specs
i7 950@4ghz ( now running stock everything)
6gb ddr3 corsair xms3 1600 ram
windows7 hp 64bit
gtx 460 1gb SLI
a lot of peripherals like joysticks, sound system, cameras( almost use up all of my USB ports on the back of the PC)
and a 4 month old OCZ zx1250w 80plus GOLD PSU
 
Everything was fine for these first 4 months until i had troubles booting up my pc once in a while..1 out of 10 times the screen would just stay black and windows would not boot..then today, once i tried to turn on my pc, no power was getting to ANY device that was hooked up to the motherboard...i noticed this because all of the fans, including the cpu, case and video card fans, were not spinning, while the fans which were connected to the molex connectors via the PSU were.
 
I unplugged the ATX connector and sure enough the 10th and 11th pins were burnt, with melted plastic that dripped onto the motherboard pins.So after a while of sand papering and scraping off the plastic from the pins, i plugged everything back in and my pc started up(but now im running only one video card). After reading this post i kinda understand why this may occur, but im not sure what to do from now on.
 
Should i get a new psu or motherboard to prevent this from happening again? Or am i stuck with using just one GPU so that not so much power is going through those wires?
 
Roman
 
PS: i had an antec 850bronze psu before, and while running the same set-up i did not encounter this problem. unfortunately with my high overclock and many usb devices connected at once, that psu was not enough to run that set-up, so i sold the PSU. 
 

yes i've looked into the power boost, and would certainly buy it, but i have an asus rampage 3 gene motherboard, with x2 PCIE slots and one PCI4 slot in the middle of them(the tiny black slot)..but since its a micro ATX motherboard, this small PCIE slot is completely covered up by the two video cards..

thanks ill give it a try...but it looks like a VERY tight fit..i mean it seems like i have 3-4mm to work with. Can you confirm that this "power boost" works, KMoore?

thanks a lot KMoore, i guess ill be ordering this power boost soon...but i got another question: the ATX cable i have plugged into the motherboard is the burnt one..should i replace this cable? and if i have to, do i need to buy a whole new PSU or can i just buy any 24 pin atx cable separately? also, how bad is it to operate the pc with this burnt cable?

yup but in my case, the motherboard pins are clean ( after sanding them) What im worried about is the the atx cable.  the plastic melted off of it and there is exposed metal on 10th and 11th 12v wires, kinda of like the metal housing in which the motherboard pins are inserted...
 
could this whole problem  been the reason why my pc was sometimes not booting?

hey hey, finally found something on this problem, and have some news about it. First here is my box
 
I7 920 (right now at 2.8)
Asus P6T V2 Deluxe
2xXFX factory OC 285 SLI
Corsair 850 tx
 
pics..

 

 
(I know its not as bad as some of yours)
 
My first 850tx I had for..like...2 years. I didn't even notice this problem. Suddenly, after two years, my power supply started shutting down during gaming. I used online PSU calculators and it says my system is drawing about ~770 watts
 
I found the connector like that. One thing I noticed is the heat somehow caused some kind of corrosion inside each of the female connectors on those two pins. I would get my plyers, close the connector tight, take a paperclip, clean out inside the pins, and connect it and re-connect it over and over, scraping some of that corrosion off. Playing after that the wire temp would drop for a day or two. But the power supply kept shutting down more and more. I would guess that this corrossion is essentially decreasing the contact surface between the female atx plug and the make atx connector on the motherboard. essentially taking the overall guage of the connector down. With the added power draw, this is whats causing it to get so hot, at least one of the reasons.
 
I RMA'd it to corsair and got a new one about 5 months ago, shutting down problem was solved...till now. It started two days ago. Sounds like over current protection is kicking in, even though I dont' think i'm drawing more then 850. I'm only guessing somehow the 12v1 is drawing so much from that rail that its weakening some of the overcurrent protection components.
 
tell me what you think. never tried this soldering  a 12v molex. I'm not only a technician but I have this hobby of replacing caps (notice those pics are hosted from badcaps . net). So soldering skills are very much no problem with me ;)

Uranium-235

damnit, pics aren't showing up, i'm guessing since i'm a new member and have the same restrictions as url's

I bet if I post the url it will go bye bye too

once you get your 50 posts you will be able to post Pics.

Uranium-235

seriously? you delete my post? I can't post pics, have a valid problem, and need pics to show it. You KNOW i'm not a spammer but am treated like one. Are you going to ban me next?

evga just went -1 in my eyes

Thats just the way the system is set up, 50 posts for pic's, or to enter the seller section , Who are you acusing of deleating, I saw no mod post in the thread,

I blew up my 24 pin using only two GTX 580s...STOCK too
 
From now on, I am using the EVGA power boost or motherboard external power plug.  I don't like risking it

I had a similar problem:
 
Configuration:
 
Asrock G31M-S motherboard with E5200 and 2GB DDR2 ram
Asus ENGTS250 with 512MB
Lousy Q-tec 500W PSU (+12V@18Amp single rail)
 
The PSU connects to the motherboard through an ATX 20-pin connector in a 24-pin slot. This is acceptable according to the Asrock site. The graphics card connects to the PSU via a standard Molex to 6-pin converter cable. While running, the PSU is measured with a DVM and +12V fluctuates from 11.98 to 12.12 (probably worse due to the DVM delay, and I have no scope available) under stress (FurMark). The motherboard sensors are then reading as low as 10.1V while the +12V lead on the ATX connector is browned because of the heat.
 
DVM reading on the graphics card +12V leads is the same as directly on the PSU. Surprisingly the 4-pin auxiliary power connector for the CPU also gave the same +12V reading as both the PSU directly, and the connectors on the graphics card. There seems to be a design flaw in both the motherboard and the graphics card. The Winbond chip is measuring the voltage correctly at its leads (much lower than delivered by the PSU), so there seem to be several flavors of +12V available on the motherboard.
 
The graphics card is pulling way too much current from the motherboard PCI-E connector (max. 75W) while the PCB strands are unable to supply that current.
 
The Asus (Nvidia) GTS-250 card is build to comply with the ATX 12V 2.2 specification. On a multiple (+12V) rail power supply that means that power comes from the PCI-E slot (+12V1DC through the ATX 24-Pin motherboard connector, max. 75W) and the 6-Pin additional connector supplies a second +12V circuit (+12V2DC) that supplies the remaining power (max. 150W). For single rail power supplies, the card should come with a jumper that connects both +12V circuits to the 6-Pin connector. With a single rail supply there are two paths to the same supply; this results in most current being drawn through the path of the least resistance.
 
This also results in a higher resistance from the PCB traces and the voltage drops near the PCI slots and the temperature across those traces (and the ATX 20-pin connector) rises; hence, the lower sensor reading on the Winbond chip, which is located near the PCI slots.
 
I remedied this situation by adding additional +12V wiring on the soldering side of the motherboard PCB (24-Pin ATX), same way as the author of the thread.
 
Having tested the system with this modification, voltages near the Winbond chip (and the PCI slots) still dropped too much, though less than before, so I decided to connect an additional +12V wire to the graphics card PCI-E +12V supply circuit.
 
Finally, I have a stable +12V on all required locations and the Winbond sensor reports an acceptable reading (slightly lower than the DVM reading; the difference being equal under stress and idle conditions).
 
Test: FurMark burn-in test at native resolution (1680x1050)
 
Result: negative fluctuation: 1.44% - positive fluctuation: 0.3%  (HWInfo log).
 
Maximum allowable fluctuation +/- 5% (from +11.400 to +12.600)
 
Before the modifications, the minimum (sensor) voltage was 10.1V (15.8%)
 
Cannot post pictures (yet).

Yes, that is what I did; I tried to find some additional +12V soldering points near the PCI-E slots and the Winbond chip, but those contacts are too small to solder a wire without ruining the motherboard. I did solder an additional +12V to the PCI-E circuitry of the graphics card and that will probably function the same way as the EVGA power booster. It is not that I'm unwilling to spend $10, but I wanted to analyze the problem and confirm that the problem originated in the double +12V path to the graphic card. Measuring the +12V leads on the PCI-E connector and the 6-Pin additional +12V connector confirmed that those circuits were separate.
 
Users of the VGA power boost should be aware that the Molex connector connects to an appropriate rail, as the motherboard gets its power from +12V1 and the 4-Pin CPU connector from +12V2; connecting the EVGA power boost to any other rail than +12V1 will combine that rail with +12V1. It should connect to the +12V1 rail.

And this is exactly why I have a good psu and surge protector.

Well, I can't post a link due to the restrictions, but here's a device that will not take up a PCI slot and will not require soldering. I got it at the xoxide website.

sorry dude that will not work, the issue is not that there is not a good power supply , but more of the mother board standard connectors are under rated for the current being drawn throw them.