The beast is dying, heart or brain-which is it

[b700029]

Re: The beast is dying, heart or brain-which is it

I think the 12v connector developed high resistance due to dirt/corrosion and started heating up. The heat then worsened the connection leading to burnout.

What materials are the pins and sockets made of? Galvanic corrosion may be possible here.

[mariushm]

Re: The beast is dying, heart or brain-which is it

Nah, just overloaded connector...

With my previous system, I've bought a cheap cable from Dealextreme to keep my 4pin cpu cable from staying over the video card due to short psu cable.

Didn't realize the wires were too thin - found out about 2 weeks after the system started to occasionally smell like something's burning (I know, I was too lazy top open up the system)

Too thin cable, overheating, both ground and 12v wires' insulation burned.. electric arch.. system restart, everything working like nothing happened.

[Uranium-235]

Re: The beast is dying, heart or brain-which is it

I thought of getting one of those extension cables, and soldering a higher gauge wire across those 12v1's-but I said screw it and got a rma from corsair, I paid $150 for this power supply I expect better

then again, I do have these two heated pins on the motherboard. I'm wondering if the pins might be causing some heat

once I get a replacement i'll be checking to see how hot those get. I might try an eps connector replacement on the motherboard, i'm not sure if asus would warranty such a problem considering they would probably blame the power supply

maybe I should try and gold electroplate the replacement connector

[Uranium-235]

Re: The beast is dying, heart or brain-which is it

well I FINALLY got the replacement. I was almost hoping corsair would sent a 850a (modular) or even a 950w to be nice (I know WD has sent acquaintances of mine 500GB RE3's in replacment of a 320 blues)

but they sent me the same thing, new in box, which is fine with me, I'm going to check the heating of that 12v1 after I turn it off every night

I was a little bit perturbed to see a UCC KY inside it, but I know that series isn't as bad THE KGZ/J series

[c_hegge]

Re: The beast is dying, heart or brain-which is it

Chemi-Con KY isn't bad at all. They are just as good as Rubycon.

I wouldn't bother returning the Mobo TBH. After all, in the unlikely event that the offer a replacment at all, it won't be a Gigabyte or DFI.

[Uranium-235]

Re: The beast is dying, heart or brain-which is it

Chemi-Con KY isn't bad at all. They are just as good as Rubycon.

I wouldn't bother returning the Mobo TBH. After all, in the unlikely event that the offer a replacment at all, it won't be a Gigabyte or DFI.

haha you suck dude. I don't know why people badmouth asus-i've gotten dozens of asus boards and other then nvidia OEM HP's and compaqs, they have RARELY died on me

the 12v1 is getting warm but not hot like it did, I wonder if cleaning the MB pin will help keep the temperature down

[c_hegge]

Re: The beast is dying, heart or brain-which is it

For me, (and most badcaps members), they rarely last long, even with intel chipsets. In fact most of the ones we tried to use for new PCs had problems straight out of the box.

ASUS ranting aside, the ATX connector shouldn't even get warm, so I'd try cleaning the contacts.

[Uranium-235]

Re: The beast is dying, heart or brain-which is it

For me, (and most badcaps members), they rarely last long, even with intel chipsets. In fact most of the ones we tried to use for new PCs had problems straight out of the box.

ASUS ranting aside, the ATX connector shouldn't even get warm, so I'd try cleaning the contacts.

I don't know what to say, I've had maybe 90%+ success rate with asus boards for the past 3 years, I guess you've been unlucky

well, I have two factory OC'd GTX 285's drawing major amperage from the 12v1, I think it getting warm might be expected

[shovenose]

Re: The beast is dying, heart or brain-which is it

I'm not an Asus hater either

[c_hegge]

Re: The beast is dying, heart or brain-which is it

note how I said MOST (not ALL) badcaps members.

EDIT: Maybe this is a good topic for another thread

[Uranium-235]

Re: The beast is dying, heart or brain-which is it

*bump*

5 months, its starting to shut down again! I'm wondering if the overdraw from the 12v1 isn't weakening the overcurrent circuts making it shut down at lower amps

apparently my problem is not just my problem. Look here:

http://www.evga.com/forums/tm.aspx?h...749507&mpage=1

also have some fascinating fixes. including soldering a 12v molex cirectly to the underside of the atx connector, and evga has a pci-e boost card

[Keatah]

Re: The beast is dying, heart or brain-which is it

Once a connector has burned like this, it will never be able to carry the same amount of current again: no matter what you do with it with regards to cleaning and re-tensioning and crap like that.

The connector needs to be upgraded or jumpered with wires that are soldered. Just don't run it out of spec and you won't have these issues. Have fun!

[Uranium-235]

Re: The beast is dying, heart or brain-which is it

then why was it having overcurrent shutdown issues? if it carried less, I would figure it would not shut off due to too much current

well I did clean it and clamped it down tight and has yet to heat up like that again or shut off. but we'll see how long this lasts

[ben7]

Re: The beast is dying, heart or brain-which is it

Maybe a ceramic surface mount capacitor popped, and it might be partially short circuited, drawing lots of current.

The reason why your 3.3v and 5v went high is cause those voltages are coming from the same transformer, and to keep the 12v from going under 12v, PSUs can output excess voltage on the other less loaded rails.

-Ben

[Keatah]

Re: The beast is dying, heart or brain-which is it

There's a "big science" behind the physics of connectors. It becomes important when they're run right to the limit mostly.

And simply put, here, the surface area of the contact points is/was not enough for the load.

Perhaps the connector was faulty in the first place. OR you blew it up with too much overclocking.

I don't know the answer, I only considered the connector aspect of the problem.
Power-on & Power-off surges (normal day to day use) may have overcome the connector resistance.
There might have been improper tension.
Could have been a literally microscopic dust spec acting as oxidation/insulation.
Too much or too little dielectric compound.

I'd try monitoring the connector temperature with a thermocouple probe, or an infra-red imager (expensive) and see if there are hot-spots.

Minuscule hotspots can grow as the spot oxidizes and resistance increases, soon it spreads out. And at some time it runs away and goes poof and the magic smoke escapes.

Thing about retensioning and repositioning is the surface geometry is sometimes different than what the mfg wanted. And that makes hotspots.

And finally, once a connector gets like this, mostly, your motherboard is dealing with AC power. Not AC mains of course. But the onboard regulators and filter networks are fighting against a fair sized DC pulse operating at the frequency of the vibration of the arcing connector.

Without going into more detail, all this stuff snowballs and shortens the life of other parts. Typically it affects the complex IC's like cpu and gpu. And the simpler chips are less affected. Harmonics from the struggling regulators get into them and they become unstable. Resonances wave across the chip and other strange things occur. It's like putting bits of sand in a gearbox. And rocking it back and forth the clear the mechanism. Temporary lockups occur, you don't even see these, because, the internal ECC (cpu,gpu,memory bus) is working. Working overtime when it should be having a party!

When overclocking a system, try to keep as much as within mfg spec as possible. The exception being the components you're trying to force beyond their specification.

I've never "overclocked" a connector. But, using the one on the motherboard, plus another one soldered to the bottom of the motherboard and used in parallel would double the contact area which is the root problem.

[c_hegge]

Re: The beast is dying, heart or brain-which is it

then why was it having overcurrent shutdown issues? if it carried less, I would figure it would not shut off due to too much current

Two possibilities:

1 - worsening connections. If the connection gets worse, the resistance will increase, which will lead to the devices pulling more current from the PSU to overcome the increased resistance.

2 - The motherboard is shutting off, not the PSU. I know some newer boards from ASUS and now Gigabyte will shut off if they are unhappy with one of the voltages.

Now, this is why I don't like single rail units. The connectors melt BEFORE they shut down, and then this kind of thing happens.

[Uranium-235]

Re: The beast is dying, heart or brain-which is it

Could have been a literally microscopic dust spec acting as oxidation/insulation.

I know it has oxidation or some other corrosion on it. Scraping it off the best I can has pretty much stabilized it

I'd try monitoring the connector temperature with a thermocouple probe, or an infra-red imager (expensive) and see if there are hot-spots.

I know there are hot spots, but only on the two 12v1 rails. Every few hours I feel down on those wires during play. after I 'clean' it, they are barely warm, but after about 3-4 days they become just as super hot as before

When overclocking a system, try to keep as much as within mfg spec as possible. The exception being the components you're trying to force beyond their specification.

this problem is actually known (at least I found out a few days ago its known).

The problem has to do with the 12v1 rails (one on the 20 and one on the 4 right next to each other). The 12v1 are rails for pretty much the main pci/pci-e bus. The problem comes in where most video cards draw most of the power from the seperate pci-e 8 and/or 6 power connectors, some video cards still draw a sizeable amount from the PCI-E bus, which puts more stress on those 12v1 rails then it was ment to handle. This is considering the fact I have two gtx 285 factory OC'd cards. Pulling several more watts from the 12v1 rails. This problem has showed up in more cases. Take a look at the link above to evgas site, the'll have some picture of people that have burnt thier rails out even more than mine

its not CPU overclocking, its a simple video card issue. But not much I can do about it other than take one of my cards out of SLI. Which I shall not do. Other solutions include getting a evga power boost, which takes one of the 5v rails and hooks into a pci-e slot (any slot) which will allow the system to take some stress off the 12v1's. Other than that another solution is soldering a molex 12v line to the 12v1's under the atx connector.

[Keatah]

Re: The beast is dying, heart or brain-which is it

Well once it's stabilized, KEEP it stabilized with dielectric grease.

Otherwise. Then. I guess one fix is to increase the contact and surface area. Multiple contact points in parallel would work. Increase the surface area till the connection(s) no longer get warm. If you need to hack it then hack it!

Or, If after a few days, like you say, they are warming up, dielectric grease may help. This forms a seal against the normal oxygen in the air from reacting with the metals. It helps curb the electrolysis/corrosion process. If your contacts are good and tight this might be enough to curb the increasing snowballing resistance. But it works best as long-term protection on properly spec'd connections. You just pack the grease in there, wipe off the excess. Dielectric grease is totally non-conductive and quite the good sealant and insulator.

INSULATOR?!!?!?!?! WTF?? Yeh, the contacts cut through it, it's like vaseline and jelly. And the excess forms an airtight barrier. Be sure it is not silicone grease, that becomes conductive. Make sure it's dielectric grease.

http://www.permatex.com/products/Aut...ric_Grease.htm

..is one choice of many! Loctite makes the shit too.

If oxidation is the problem, and there's sufficient contact area, dielectric grease will definitely fix the problem. In this case, the connector blocks may still get a bit warm, but it will take more current to push them into runaway than without the grease.

Bending and sanding and burnishing and cleaning. Pfffaagghh. It's a short term fix. You'll keep doing it till you have no contact left. I know, I've done it before.

You may also try gold or platinum contacts. Or screw terminal contacts.
And be sure the contact metal is the same. Copper to copper, or tin to tin. Whatever.

And regarding overclocking, seems EVGA pushed the stuff beyond the limit a bit. Is it Nvidia approved - what EVGA is doing?

While this won't help you directly, I always prefer to build on the conservative side with last year's tech. It's been debugged. What's on the market has maintained its presence because it works. It's cheaper. And all the early adopters have documented the problems for me. Thanks guys!

[Joun]

Re: The beast is dying, heart or brain-which is it

Well I have to add my experience as an electrical engineer( to anybody wants to hear it)
Once a contact goes bad there is no reviving it no matter what you do.. It may work for some time but it will break again sooner or later..
The only way to fix it permanently is to solder the cables directly on MB and even connect the two 12V pins..I'm suspecting there was nothing wrong with MB connector or the supply connector, you just draw too much current from them..As there are many ground pins on the connector to share the current the weakest link on current's path was the 12V pins so they broke..

[Keatah]

Re: The beast is dying, heart or brain-which is it

Yes that is generally true for contacts run right up against max spec.
And in the cost-cutting world of PC component mfg there's little wiggle room.

When I've "repaired" failed contacts I've either had to re-plate them, or run them at, say perhaps, 90% max rating. In many cases the dielectric compound worked wonders provided the contact didn't fail from overstress.

It's interesting to note that over-bending and over re-tensioning the contact creates little ripples or waves in the surface - thus eliminating the smooth touch points. And therefore reduces the surface area AND IT LOWERS the max current capability.

This is something few if any PC builders know about.