Need urgent help! Motherboard smoking!!

[fractal9]

Hello to all, my first post since I can't remember my old handle.

I was having power supply issues where the psu would make power fluctuating noises (hard to explain) and the amber power lights on my nForce4 would flicker on and off very quickly. After turning on the psu switch I would be able to turn on the computer for about 3 seconds (after which it would boot fine), but if I waited too long the power would not be there for it to start. So I decided to yank a few things, which didn't work. After reinstalling my SCSI controller, however, upon system start I saw and smelled smoke coming from the mainboard. So I decided it was time to finally get an RMA on my PSU.

I recieved my new unit today. After cleaning and plugging into the motherboard, the amber lights were solid, so far so good. So I went to start up the system, and once again I started seeing smoke billowing out of near where the lower amber LED power indicator is (I think it was coming from a capacitor). So I removed the motherboard from the tray and saw that there was a burn mark on the tray and on the back of the motherboard. I have pictures if that is necessary.

What the hell is going on? I am afraid to turn on the system again, as I believe it already destroyed my SCSI controller. Here is my setup:

Antec P160
Mushkin 800W PSU
DFI nForce4 Ultra-D
Opteron 148

That is all that was installed this time and the motherboard started smoking.

Here is some of my other gear:

2 x 1GB OCZ PC3200 (was not installed)
2 x 7800GT (were not installed)
LSI U160 (not installed)
Seagate 15k.3 36GB (not installed)
Maxtor Atlas 15k II 147GB (not installed)
Hitachi Deskstar 250GB
WD Caviar 750GB
NEC DVD/RW

Thanks in advance, I've never had an issue this serious before...

 

[bruceb]

It sounds like something is causing a short either on the mobo or between the mobo and the case.
Did you recently install the mobo ? ? Check that all the metal standoffs are in the correct spots and
not touching any circuit foils.

 

[fractal9]

The motherboard is installed correctly. After removing it and plugging it in without turning it on, I notice the area right under the lower amber LED gets hot even when the system is not turned on (but power is flowing through it - PSU switch is on, amber lights are on). The other amber light (near DIMMs) is not warm nor heating up. ???

 

[fractal9]

I found what is getting hot (even with the system off!). It is a small black chip at the bottom of the motherboard, just below the lower PCI slot.

Picturehttp://marc.cusumano.googlepages.com/LPnF4SLI-DR.jpg

 

[bruceb]

Can't tell exactly from that photo, but if it is a chip mounted to the mobo then you need a new mobo.
They are not replaceable by the average computer tech, even if you could find the proper replacement.
Based on what you said, apparently when the PSU went out, it took that part & maybe others, with it.

 

[fractal9]

OK, thanks for your help. Guess it will be a while before I get my system back up :brokenheart:

 

[fractal9]

Update - I went for it one more time. The damn thing actually boots!!! I got into windows and everything. The smoking has stopped, but the area which was fried (literally - you can see it on the back of the motherboard) still gets pretty damn hot (can put finger on it for no more than 5 seconds). The system is stable so far, but I am afraid of an electrical fire at this point. Also, my SCSI card was killed by whatever the hell happened.

Do you think it's safe to run this system like this? I can't afford to buy a new motherboard, especially one that is nearly four years old.

PS Can anybody recommend a PCI-E SCSI Controller that is a reasonable price?

 

[mindless1]

Originally posted by: fractal9
Update - I went for it one more time. The damn thing actually boots!!! I got into windows and everything. The smoking has stopped, but the area which was fried (literally - you can see it on the back of the motherboard) still gets pretty damn hot (can put finger on it for no more than 5 seconds). The system is stable so far, but I am afraid of an electrical fire at this point. Also, my SCSI card was killed by whatever the hell happened.

Do you think it's safe to run this system like this? I can't afford to buy a new motherboard, especially one that is nearly four years old.

PS Can anybody recommend a PCI-E SCSI Controller that is a reasonable price?

Is the motherboard still under warranty? If so, do nothing to it, just seek an RMA replacement.

If not under warranty still, on that (transistor?) that is getting hot, desolder and pry up both legs on it, or if you have no way to desolder, find a way to snip those legs breaking the circuit.

Now I'm going to speculate a bit, some guesses. Original PSU probably didn't damage it, either a card that was plugged in was failing which shorted it pulling too much current, the transistor itself was defective and shorting itself, the capacitor next to it on it's output is failing and has very high losses, or the board layers had a defect such that a power trace and ground plane short started occurring after it ran warm for awhile.

From one of the above, it was probably drawing excessive 5VSB current and that's where the startup problems and flickering board LED issue came from, that it was too great a load on the PSU. That might've killed the original PSU, or maybe the original PSU still works ok or somewhat ok but can't provide much 5VSB anymore. You might compare it's 5VSB rating with that of the new PSU for comparison purposes.

I'm suggesting that transistor is part of a switching circuit to provide 3.3V to the PCI slots, that if you don't need PCI power-on events from these cards you may not need that transistor at all. At any rate if you have no warranty it's worth pulling up the leads on it and seeing what functionality the board still has then if it seems ok from that point onward, plug in some low value PCI cards like a NIC or whatever and retest, power off and add another, retest, etc.

Here's a better pic of that area, but unfortunately it's at enough of an angle we can't see the markings on that and what's probably a regulator chip just to the right of it, and thanks to the dark colored PCB it's harder to trace it. I kinda liked the old days when more boards were like Asus color schemes, golden colored so traces were easy to see.

If you happen to have a multimeter you might also probe the two legs and tab while power is off to see if they're common to the 5VSB lead from the PSU (with system AC power disconnected).

 

[fractal9]

Mindless - thanks for your thorough reply!! The motherboard is indeed not under warranty so I will attempt to tinker with it when I get home. I can't believe I am going to get out of this with a working motherboard!!

 

[mindless1]

Originally posted by: fractal9
Mindless - thanks for your thorough reply!! The motherboard is indeed not under warranty so I will attempt to tinker with it when I get home. I can't believe I am going to get out of this with a working motherboard!!

Well maybe you won't... I could be wrong or only partially correct and snipping those transistor leads could kill the board instantly by removing a crucial power supply to something. However, I gave the suggestion within the context that if that doesn't work and you have nothing else plugged in that's causing a short, it seems the only alternative left short of replacing the whole transistor and probing the rest of the board for faults - and if you're replacing the whole transistor anyway it wouldn't hurt to snip the leads beforehand, you still have the option either way to desolder it and actually it's typical to snip leads when desoldering something like this unless using a hot air rework station.

There's also the possiblity that the damage to the board was extensive enough that later one of the layers will short with another, and that could kill the board or disable some feature if there are data traces running over that area. It's worth a try though, since the transistor is obviously overheating the area you just snip it out of the circuit then see what was effected, meaning you may lose some board feature(s) whether it be 3.3VSB for PCI card power-on events, or something else.

If you have a multimeter handy and can measure voltages while the PSU is very briefly connected to the board and supplying 5VSB power, you might measure the voltage on the capacitors to the immediate left and right of that chip. Having the output voltage would help narrow down what it might be powering.

Note also that it might be possible the input is 5VSB, but the output powers the board's power-on logic. In other words, the board might never turn on again without that supply but if that is the situation you have a more serious short that would kill the board eventually since the transistor can't be left to bake a hole in the board, left overheating it would fail (if it hasn't already, it may already be dead by a partial short so you may only be disconnecting a part acting like a small heater) or damage the copper traces around it without any other resolution.

So basically it's all or nothing when you snip the leads, there's a good chance (based on what info we have) that will help, but there's still a chance the board is beyond salvage.

 

[fractal9]

I still haven't had a chance to do anything. I have no tools (multimeter, soldering iron, etc.), and I don't really want to go tinkering around with the board since it works...

If I use the board without doing anything else to it, do I risk damaging other components? I had the system running OCCT for 2 minutes with no problems, but overnight is a different story...

 

[mindless1]

Originally posted by: fractal9
I still haven't had a chance to do anything. I have no tools (multimeter, soldering iron, etc.), and I don't really want to go tinkering around with the board since it works...

If I use the board without doing anything else to it, do I risk damaging other components? I had the system running OCCT for 2 minutes with no problems, but overnight is a different story...

It's likely the board already killed the first PSU, and could be a similar strain on the new one - though maybe the new PSU is more robust and the strain won't matter.

If the transistor fails shorted, it will cause nearly the full input voltage to be applied to whatever it's powering which could damage whatever that is.

If the area keeps overheating it may cause further damage. If the part itself keeps running too hot it may fail itself.

What I would do is find a friend with a soldering iron and desolder and gently pry up both legs on that transistor then see if the board still works. If it doesn't, solder them back down onto the pads and hope for the best. On the other hand if you gently enough snipped the leads and left some lead on the transistor then someone could come along later and solder the remainder of the lead to the pad with a small piece of wire (a slightly more difficult soldering job since soldering so close to the transistor body will heat it up more, but not terribly difficult to someone with reasonable soldering experience).

Then there's the 3rd option, write down the transistor part #, see if you can find an exact (or near equivalent replacement) replacement for that transistor then even if the old one can't have it's legs soldered back on, the new one can be soldered in to replace the old one. It's really a question of what is going on which we don't know, if the transistor is leaking like crazy and already some downstream part is consuming too much current which may cause a secondary part failure, or if a downstream part is consuming too much current from it's own fault and the transistor overheating is merely caused by that.

Keep in mind I'm still only assuming it's a typical transistor rather than some more integrated IC, if you post the part # off of it (if you can still read that, the more it overheats the harder it may be to read off the #) then someone may come closer or be able to identify it.

You haven't yet shown us detailed pictures of the front and back of the board. Maybe they won't do any good but maybe someone will see what else is around that area on the front or back and that would allow more predictions about what will happen if the area keeps overheating. For example some boards route data traces along the edges and if those delaminated from the layer they're over it could eventually cause them to break or short to ground.

With the area getting hot enough that it's smoking and causing burn marks, I would not try to run the system like that more than seconds at a time to get a voltage reading unless you place your finger on the part before turning the system on and leave it on the part when system starts running to confirm it is not getting excessively hot anymore.

 

[Dillybob]

Dude. That's an old motherboard...still usable, but old. What socket is it? What CPU?

Yeah, I know I'm not really helping anything. But by the sound of things, you need a new motherboard. If it's not under warrenty, get a new motherboard. Is the NForce4-D intel or AMD?
AMD boards are REALLY cheap for some pretty decent hardware.....unless it's 939.

That's all I really have to say.....I'm not an electrician or engineer, so if you want to fix it yourself, listen to mindless (no insult intended).

If you want help finding a new board, let me know though!

 

[fractal9]

Thanks again mindless, you've been more than helpful. And yes, the board is a socket 939, and I haven't been able to find a different board at a reasonable price. So I am kind of stuck. Based on what mindless said, even if I did the repairs / tinkering suggested, I would still not be comfortable running the system for risk of damaging the other components, which are still worth a considerable amount of money. I guess I will be stuck using my parents old system (emachines) for a while :disgust: