What component is this?

[irq]

Hey everyone! I'm a bit new to electronics and have been really enjoying Dave's blogs and especially the people and information on this forum. I was hoping someone here might be able to help me identify a component.

Long story short, My Mac Pro just died, and some magic smoke came out. After tearing everything down, I found the problem... a component labeled L12 on the board for my video card, which is an Nvidia Geforce GTX285.

The component obviously got a bit hot, destroying its label and melting some of the solder holding it to the board, as well as warping some plastic on the heatsink that attaches to this board. The component itself says R10 on it but being labeled L12 on the board, I really doubt it's a resistor. Is it an inductor? I'm still definitely an electronics newbie...

Anyway, here's a photo:


You can see the full photo here: http://dan.nyip.net/videocard/full.jpg.

[Alex]

It's an inductor. I doubt this is the only component damaged.

Can you take a picture of the underside of the board at that location? Look for some black components marked D-something.

[irq]

You may be right, but as far as visually-obvious damage goes, it's the only one. It's cheap enough to try replacing it and seeing if the card functions again. It's unfortunately out of warranty, so I don't really have much to loose. I actually have decent skills with SMT soldering, so I'm confident I can physically replace the component.

What does R10 mean? 100nH?

[mikeselectricstuff]

It's an inductor. I doubt this is the only component damaged.

Yeah - if this smoked it will have been a symptom of excessive current elsewhere. If you're lucky it might just be one or more of the MOSFETs to the right, but it could have toasted the GPU it's powering.

[irq]

Is there any good way to test the MOSFETs?

[Alex]

This looks like a 6 phase step down converter. The inductors are 0.1 micro Henry. Unless the diodes are on the other side (need a pic), they ahve used the MOSFETs as diodes from the same controller on the right.

The damage might be caused by the MOSFETs connecting the inductor across the power supply for too long. If that was the case and the inductor was blown, the GPU/memory would have been fed with the full 12V from the PSU. My bet is that this is a dead board beyond repair.

[irq]

The interesting thing though is that the board does still work, but only for about 5 minutes, then the computer freezes. Before it freezes, everything works fine, including my 2560x1600 display. That makes me think the GPU & RAM might not be damaged? Do you think it's worth trying to replace the inductor?

[Alex]

Interesting. Does the GPU get hot?

[irq]

It does get hot, but I can't really tell if it's hotter than normal or not. This card is actually a 2-slot-wide PCIe card - there's a humongous full-board-length heatsink-with-fan assembly that attaches to the card with about 20 screws. It covers the entire board, with separate little heatsink widgets for the main GPU, another nvidia chip, and each memory chip.

Under normal operation the card gets hot enough to heat up the room the machine is in. It's an extremely high-end graphics card. So yes, it does get hot, but abnormally hot? Hard to say.

[Zero999]

100nH seems a bit small for an SMPS. It's either a filter or it could be 100µH which makes more sense for an SMPS.

[Alex]

Can you touch the heatsink that the GPU is attached on during operation? You should be able to touch it comfortably. I wonder if it goes into thermal shutdown. Make sure you have a small amount of fresh thermal paste on place.

Yeah, I know, 100nH is small. But this is the base unit for this notation, have a look here for example:
http://www.tdk.co.jp/tefe02/e531_spm5030.pdf
http://www.bourns.com/data/global/pdfs/SRP4020.pdf

Maybe its an LC filter at the output of each phase? Can you take a picture of the other side of the board irq.

[irq]

Believe it or not, you can't actually touch the heatsink once it's installed. It's completely covered on the top by plastic, its only interface to the outside world is through a bunch of metal foil vents on the side of the card.

Below is a link to a photo of the back of the board. The dead inductor is almost directly underneath the "do not throw out" garbage bin/recycle bin icon. Thank you so much for looking at this guys!

http://dan.nyip.net/videocard/back.jpg (beware, 14MB JPG)

[Zero999]

Yeah, I know, 100nH is small. But this is the base unit for this notation, have a look here for example:
http://www.tdk.co.jp/tefe02/e531_spm5030.pdf
http://www.bourns.com/data/global/pdfs/SRP4020.pdf

Yes I know, but I've seen inductors marked in mH before, although µH seems to be the norm.

[Alex]

Thanks. What made you think the graphics card causes the PC to hang? Have you tried another card or the on board graphics? If it's the graphics card, try running it without the plastic cover and with one or two more fans pushing air through the fins. Touch the heatsink without the cover, how hot is it? Some cards have on board temperature measurement that you can use.

As far as the PSU goes, it seems these are the only inductors around. Maybe another phase has failed and the controller is working this one hard. Who knows. It is possible that there is nothing wrong with the PSU of the graphics card.

Do you have an oscilloscope, multimeter?

[irq]

I'm positive it's the graphics card because the machine freezes within minutes every time I power it on with this card. If I put in a different graphics card, the problem never happens. Also, I think the melted plastic on the fan casing and obviously overheated inductor are good indicators.

I do have an oscilloscope and a multimeter. However, I just discovered that this card still has 3 months of warranty left on it. I thought it was just 1 year, but it's actually 2, so I'm going to see if EVGA will replace it for me. From what I can tell, they should.

Thanks again, Alex, and everyone else, for all the help. This is one of the best online communities I've seen and it's guys like you that make it so good.

[Zero999]

The bad news is you've removed some components so the warranty will no longer be valid. The good news is, if you solder them back on they'll probably not notice and will have no problem returning it.

[irq]

Um, what components are you referring to? All I did was take off the heatsink to inspect the board, and put it back on again... and there was no warranty seal stopping me from doing that.

[Zero999]

I noticed there were some empty spaces for C68, C69, C75 and C76. I assumed you'd removed but they're obviously just there in case they were needed which is common practise in PCB design.