8800GTS shorted board? psu OCP

[Kram]

Hi
Ive been given an 8800GTS (g80) board that is unmodified, that has just stopped working.

It was working perfectly, then pc crashed/blackscreen, and then the pc would not boot at all. Swapped graphics card and pc works fine.


The card is tripping the psu's over current protection (only turns on for a milisecond, have to reset with mains switch). With 6pin cable removed, the psu doesnt trip, but screen stays black.
From the 6pin connector, 12v->ground is 1ohm.


Could this be the same issue people were having with nvidia gpu's in laptops (bga solder problems)?


I suspect it is more likely blown mosfets but not really sure how to test.

[largon]

First I'd desolder one leg of each vGPU inductor to determine if the short is in VRM or GPU.

[diredesire]

I had a similar issue, can you shoot a picture of the board? My card was an eVGA 9600gso. I replaced one of the power FETs, and it booted right up. I'd start by using your multimeter to probe the legs of each fet. Look up the data sheet, and then use the diode mode on your multimeter to probe from gate to drain, gate to source. Your DMM should beep or something, but only in one direction (gate out, or gate in), depending on type of fet.

You will probably need a hot air rework station to remove the fet from the board, since there has been solder paste applied to the tab. If you remove the fet, measure 12v/0v resistance, it should jump back up in to the K Ohm region. If that is the case, you can just replace the FET, and see what happens. I can't guarantee it will work, it really depends on WHY the FET failed.

This all of course assumes that the problem is what I described. I doubt it's the VPU, but if it is, check this thread out for a theoretical fix:
http://hardforum.com/showthread.php?t=1421792
(yeah yeah, i know..)

[diredesire]

I should mention that if you haven't used your multimeter before in diode mode, that you can just check resistance from gate to source/drain. If any fet is conducting (low R) in both directions, it's likely dead. This also assumes that the fet is OFF the board, but if you're seeing values in the 0-low hundreds range both directions, chances are you found your device.

When I "fixed" my board, I traced back to the fet because of one thing: I actually tried powering the card without it being plugged into the PCIe slot, and the FET smoked (literally). In retrospect, this is probably not a great idea, as I have no idea what the circuit topology of the card is, so I say don't try this unless all other troubleshooting is fruitless ;)

JD

[Kram]

Im having some problems with finding datasheets :/

I found a capacitor (smd cermaic) that looks like one leg is black (this may not mean anything, the solder joints look quite horrible in places). Its between a small 3x3mm BGA package 11BD FKFM (fet driver?) with a via going between two fets, but not to the outer layer. I will have to get a picture.


I assume the vGPU inductor would be the 1R0 labled component, bottom right of the front pcb picture below (between white connector and capacitor.

The fets appear to be labled K0302 6k4 (x8 on board) / K0305 6k3 (x4).
They have 4 pins + tab, but 3 legs are joined. On the K0305's the tab is +, legs goto the tab of K0302's, one leg goes to a via.
So Im not sure if the tab or the legs is vGPU from the K0302.

This is the same pcb layout,
http://www.ixbt.com/video2/images/ref/gf8800gts-scan-front.jpg
http://www.ixbt.com/video2/images/ref/gf8800gts-scan-back.jpg


Running the board without it in the slot wont help me - I think, the psu is seeing a direct short rather than the card deciding not to work :O it doest stay on long enough to burn anything :p
But Ive burnt things before, my r9700np died from a failed vmod, a fet was glowing red hot/exploded.. I am not happy when stuff just dies for no reson. I like to know why and preferably that I caused it :D
thanks
Mark

[diredesire]

Im having some problems with finding datasheets :/

I found a capacitor (smd cermaic) that looks like one leg is black (this may not mean anything, the solder joints look quite horrible in places). Its between a small 3x3mm BGA package 11BD FKFM (fet driver?) with a via going between two fets, but not to the outer layer. I will have to get a picture.


I assume the vGPU inductor would be the 1R0 labled component, bottom right of the front pcb picture below (between white connector and capacitor.

The fets appear to be labled K0302 6k4 (x8 on board) / K0305 6k3 (x4).
They have 4 pins + tab, but 3 legs are joined. On the K0305's the tab is +, legs goto the tab of K0302's, one leg goes to a via.
So Im not sure if the tab or the legs is vGPU from the K0302.

This is the same pcb layout,
http://www.ixbt.com/video2/images/ref/gf8800gts-scan-front.jpg
http://www.ixbt.com/video2/images/ref/gf8800gts-scan-back.jpg


Running the board without it in the slot wont help me - I think, the psu is seeing a direct short rather than the card deciding not to work :O it doest stay on long enough to burn anything :p
But Ive burnt things before, my r9700np died from a failed vmod, a fet was glowing red hot/exploded.. I am not happy when stuff just dies for no reson. I like to know why and preferably that I caused it :D
thanks
Mark

If you can get some very close up pictures, that'd be helpful. I think the pictures you linked will be fine to refer to for full-card shots. Since the PCB is black, it's difficult to tell where the traces/planes lie, if there's any way when you take closeups, try shooting at an angle or something so we can see the division of traces... As far as datasheets, yeah, it's going to be tough, I had a different package for my FET.
I doubt the SMD cap is the culprit, I think it's just a coincidence.

Are there any other markings on the fet? I think this might be what you are looking for:
http://www.alldatasheet.com/datasheet-pdf/pdf/247451/RENESAS/RJK0305DPB.html

It's a very interesting package, I don't think I've seen one like it before...
Here's the K0302: http://www.alldatasheet.com/datasheet-pdf/pdf/247448/RENESAS/RJK0302DPB.html

I actually didn't think running my card out of the slot would make any difference either, because I was getting a short with the PCIe connector in, but none without. I wanted to see what happened, just for :banana::banana::banana::banana:s and giggles, and it broke the fet. Looking back, I have no idea why it would have. (it makes no sense, right? :ROTF:)

I know what you mean by not being happy when stuff dies. I dont' think i've ever killed a part via modding, etc, probably expedited the death of, but I've brought several things back from the dead :)

I'm starting to think though, that the time investment to do this sort of thing is getting less and less worth it.

[largon]

vGPU inductors are those marked as:

BL
121

Lift the other leg up of each of them and measure Ω between GND and both inductor pad.

[Kram]

Thank you both!
Found a few problems:
Q15 K0305 6k3, source/drain and gate all shorted.
Q5 and Q6 are k0302 6k4. Source/drain shorted.


Q5 + Q6 are memory (the gpu is the BL121 inductor phases?, the phase at top without is memory?)


All others the ohm rises like a capacitor with only a very short bleep.

So I will need both type of fet :O

I lifted the inductors (- lead was easyer, I hate Jlead smd's) I got 502-506ohm on the good phases and 480 on the dead phase.


I would really like a suggestion of a replacement, since I really have no idea with fets.

I was going to ask if its ok to test with the gpu phase fet removed, but since the memory phase is bad too theres little point.

****Im getting pictures now*** :)

[Kram]

0-0051.jpg (http://jellard.pwp.blueyonder.co.uk/8800gtsgfx/0-0051.jpg)
0-0052.jpg (http://jellard.pwp.blueyonder.co.uk/8800gtsgfx/0-0052.jpg)
0-0053.jpg (http://jellard.pwp.blueyonder.co.uk/8800gtsgfx/0-0053.jpg)
0-0054.jpg (http://jellard.pwp.blueyonder.co.uk/8800gtsgfx/0-0054.jpg)
0-0057.jpg (http://jellard.pwp.blueyonder.co.uk/8800gtsgfx/0-0057.jpg)
0-0059.jpg (http://jellard.pwp.blueyonder.co.uk/8800gtsgfx/0-0059.jpg)

[largon]

Those fets will be sort of... an experience... to solder.
¦)

[Kram]

http://www.vishay.com/docs/73011/73011.pdf

Not the same package but got 6 of them ready to use, specs look good enough to me *not that I know what Im looking for*... Would this work?

In this pcb layout would they both be N channel type? it annoys me that 0305 is probably the manufacturing week :o


also if the fets are shorted, does that mean the gpu has seen +12v and dead too?

[largon]

Could be.

[diredesire]

yeah, it could be, depends on what happened. It's tricky to troubleshoot problems like these, because there has generally been something that caused such a "catastrophic" failure.

As far as replacing the FETs, they're generally n channel. As Largon has said, it's going to suck if you're trying to remove the FETs with just a soldering iron, the tabs that are attached to the back of the FET are going to be soldered down to the board, unless you have a hot air rework station, you're likely going to do more harm than good...

[diredesire]

yeah, it could be, depends on what happened. It's tricky to troubleshoot problems like these, because there has generally been something that caused such a "catastrophic" failure.

As far as replacing the FETs, they're generally n channel. As Largon has said, it's going to suck if you're trying to remove the FETs with just a soldering iron, the tabs that are attached to the back of the FET are going to be soldered down to the board, unless you have a hot air rework station, you're likely going to do more harm than good...

[railmeat]

interesting thread.tho i must admit i shot apple juice out of my nose when and i quote from op "a fet was glowing red hot/exploded" lol....i can relate as i had MANY caps blow in neon light kits.lol for no reason at all hooked up when i get them brand new i hook them up to a good external p/s as a tester and seems like 2 out of 10 sets blow up for no reason :o scares the crap outta ya tho cause u dont expect a m-30 to go off inches from you,lol.

im following this thread and see what happens.u guys know your stuff and i wanna slowly learn it as well. l:o

[diredesire]

Were you able to repair your card?

I ended up removing my FET fix, it looks like it can't handle the current, I'm looking for another suitable replacement. Anyone have a spare APM2509N? The replacement I found was similar in specs, but obviously isn't up to the task. I can boot up with the card, but after about 10 minutes in windows my screen just blanks, and I can feel the VREGs/MOS section of my vcard are pretty hot to the touch.

le sigh...

[crashZ]

I just bought a new DELL S2409W monitor and was using it with my old Geforce 8800 for about 2 hours on my pc when I put on Call of Duty and suddenly when the game resolution changed the screen just remained black, I restarted the machine and the screen didn't come on.

Then suddenly smoke started coming out from the PC, I put everythign off then observed that the APM 2509N burnt. Is there an alternative I can use for this mosfet? Also I dont know why this burnt? Could it be the new monitor?

How do I remove the mosfet? If I some how get the mosftet replaced could some other component have burned as well?

Any help would be greatly appreciated.

[diredesire]

I just bought a new DELL S2409W monitor and was using it with my old Geforce 8800 for about 2 hours on my pc when I put on Call of Duty and suddenly when the game resolution changed the screen just remained black, I restarted the machine and the screen didn't come on.

Then suddenly smoke started coming out from the PC, I put everythign off then observed that the APM 2509N burnt. Is there an alternative I can use for this mosfet? Also I dont know why this burnt? Could it be the new monitor?

How do I remove the mosfet? If I some how get the mosftet replaced could some other component have burned as well?

Any help would be greatly appreciated.

I am sure there is an alternative for the mosfet, but when I replaced mine, my card would overheat and eventually die. I am going to venture to say that it is unlikely to be stable again after replacement, unless you find another APM2509N, and if you do, le tme know where you found it, i could use one :)

Can you snap a picture of the card? What brand is it? I'm curious to see if our fried APM2509Ns burned in the same place.

I'm not sure the cause of the death :(

To remove it you will need a hot air gun or a hot air rework station, the metal tab on the back of the FET is soldered down to the graphics card.

[crashZ]

If only i knew which card or devices have apm2509N on them I would start searching. I have got some old AGP cards including a Geforce 4200 Ti and a Radeon 9000 they have similar looking mosfets on but I don't know if they'd work or be compatible and since i dont have a heat gun I cannot even take them off. Anyways here is the image you requested.:confused:

http://www.bis.edu.pk/apm2509n.jpg

http://www.bis.edu.pk/apm2509n2.jpg

[diredesire]

My card is an EVGA 9600 GSO

Look up the datasheets of those other MOSes, and see if the specs are similar. It might just post if you replace it. Mine would post, but would crap out (and die) after a few minutes, with severe heat issues.

[undrtkr]

i have a 8800gtx with the same mosfets shorted

making a short search, i found the datasheet for the mosfets, i think it will help
the full name of the mosfets are:

RJK0305DPB and RJK0302DPB

the 0305 are rated for 30A and 120A peak
the 0302 are rated for 50A and 200A peak

now i have to found similar components and try it

good luck

[Kram]

its still dead,

I had a quick look, would these work?

VDS Polarity RDS(ON) ID cont. IDM PD VGS(th) Manufacturer Device
(V) (Ω) @25°C (A) pulsed (A) @25°C (W) max. (V)
60 N 0.09 30 120 70 4 ST STP36NF06
60 N 0.018 50 200 110 4 ST STP55NF06

http://www.rapidonline.com/Electronic-Components/Discrete-Semiconductors/MOSFETs/TO-220-Power-MOSFETs/66264