GTX 550TI Faulty GPU Chip keeps blowing MOSFETS?

[taxxin]

Dear Bad Caps,

Ive been trying to search extensively on how to diagnose a Bad GPU chip but cant find too much, I thought i might just tap into the wealth of experience here.

I have this GTX 550TI came with a short. I noticed that the Thermal paste was rock solid and took a lot of force to remove the heatsinks for both VRM and GPU.

Found that it was a mosfet blown in the main VRM to the gpu chip. Replaced that and found that a electrolytic cap was bad and was causing short. Replaced both parts and Tested again the short was gone so After Powering up and testing it turned on for a few seconds then tripped the psu short detection. the same mosfet was blown.

So ive removed all the high side and low side mosfets from that rail completely and tried to to test again, within seconds of turning it on the chip was blazing hot, burned my fingers twice. so i decided to put the heatsink back on and try again. It turned on again for a few seconds and it tripped again. Blew another short.

So my question is what could be causing this to keep happening? Is it a bad GPU chip?

Regards
Taxin

[taxxin]

Re: GTX 550TI Faulty GPU Chip keeps blowing MOSFETS?

So just a quick update,

Just tried to inject some more voltage into the Vcore circuit, it seems there is a short in the low side circuit, no short in the main 12v rail. there was a short in the pcie slot 12v but when i pressed on some of the nearby filter caps the short disappeared.

could be one of the vcore caps shorting to ground but no way to check, my power supply only goes gives 4A max before OC protection kicks in.

[sukizg]

Re: GTX 550TI Faulty GPU Chip keeps blowing MOSFETS?

Check your MOSFET drivers... what is your phase measure in ohms?

[dicky96]

Re: GTX 550TI Faulty GPU Chip keeps blowing MOSFETS?

That's not a very good pic but it looks like you removed the high side anfd low side mosfets on one phase of Vcore only.

Take the heatsink of the card and post some clear photos with the DC resistance to ground from ALL of the inductors you can find on the card please. Check on both sides of the card, though there are probably no inductors on the rear.

[taxxin]

Re: GTX 550TI Faulty GPU Chip keeps blowing MOSFETS?

That's not a very good pic but it looks like you removed the high side anfd low side mosfets on one phase of Vcore only.

Take the heatsink of the card and post some clear photos with the DC resistance to ground from ALL of the inductors you can find on the card please. Check on both sides of the card, though there are probably no inductors on the rear.

Hello Mitch, thank you for your reply my amigo, I have been watching your videos and they have been great. I Ive been loving the videos but they do go on for quite long.

My only suggestion is that maybe you could get someone of fiver to edit your videos for you for a couple squids. just to edit out parts where you need to grab a power cord and so on. but really i think that your videos deserve to get more views than it does. Very informative and entertaining, not like the american videos which just go on and on about really basic things

So just measured the resistance and it reads 0 ohms its definitely shorted and probably burnout another mosfet causing short to ground, I will remove it in a sec and try and see what the resistance is with the short removed.

What could cause something like this to happen? (Blowing of mosfets) another short to ground on the low side?

Ive removed one set of mosfets from the board to see if the card could run without them, but it just blew some from the other set.

[taxxin]

Re: GTX 550TI Faulty GPU Chip keeps blowing MOSFETS?

All coil inductors are measuring less than one ohm, i think it might be over for this card, 3 vcore phases and one memory phase.

[dicky96]

Re: GTX 550TI Faulty GPU Chip keeps blowing MOSFETS?

It is normal for the coil inductors on the GPU main supply (the one with multiple phases) to read less than 1 ohm

The memory phase is typically around 100 ohms but make sure you use ohms range on your meter - not continuity or diode test mode.

Check out this video from my YouTube channel - I am reading 0.8 ohms on the GPU Vcore supply and this turned out to be a good working card (you can watch also the later videos on this card)

[taxxin]

Re: GTX 550TI Faulty GPU Chip keeps blowing MOSFETS?

It is normal for the coil inductors on the GPU main supply (the one with multiple phases) to read less than 1 ohm

The memory phase is typically around 100 ohms but make sure you use ohms range on your meter - not continuity or diode test mode.

Check out this video from my YouTube channel - I am reading 0.8 ohms on the GPU Vcore supply and this turned out to be a good working card (you can watch also the later videos on this card)

The memory was good before, not sure why the resistance is so low now?

[dicky96]

Re: GTX 550TI Faulty GPU Chip keeps blowing MOSFETS?

I had one card (8800GT) that was reading about 130 ohms on the VRAM supply, that suddenly went to about 2 ohms, at first intermittent then permanently. This is one the series of videos I made on 8800GT (I have two cards the same to compare). You can see in hte earlier video the resistance was high, then it changed. In the end after a lot of tracing it turned out the near short was in the GPU (I proved this by removing it). This is on the video I made called 'short circuit tracing'

I'm glad you enjoy my videos, yeah they can get too long but I'm trying to do 'reality' here LOL. I will try to edit some future repair videos to get them shorter while still concentrating on the important stuff - and then check the metrics to find out which style is getting the most views and play-hours. Cheers for the suggestion

[taxxin]

Re: GTX 550TI Faulty GPU Chip keeps blowing MOSFETS?

I had one card (8800GT) that was reading about 130 ohms on the VRAM supply, that suddenly went to about 2 ohms, at first intermittent then permanently. This is one the series of videos I made on 8800GT (I have two cards the same to compare). You can see in hte earlier video the resistance was high, then it changed. In the end after a lot of tracing it turned out the near short was in the GPU (I proved this by removing it). This is on the video I made called 'short circuit tracing'

I'm glad you enjoy my videos, yeah they can get too long but I'm trying to do 'reality' here LOL. I will try to edit some future repair videos to get them shorter while still concentrating on the important stuff - and then check the metrics to find out which style is getting the most views and play-hours. Cheers for the suggestion

Yes that was a great video, I love how you used all the different kinds of meters and also explained why each tool has its place.

I want to buy a FLIR thermal imaging camera, but its a bit too expensive to buy for my hobby. the other thing thats cool is the SHORTY device to find short circuits by being a highly sensitive ohmeter reading up to like 6 decimal places.

But its thanks to your video i know what i need these things.

[momaka]

Re: GTX 550TI Faulty GPU Chip keeps blowing MOSFETS?

Hi, taxxin,
Not sure if you are still troubleshooting this video card or not. However, for a future suggestion, DO NOT test video cards or motherboards with MOSFETs removed, particularly if the MOSFET(s) are part of a single voltage rail and you didn't remove all of them for that rail.

The reason for this is that inductors resist changes in current. So if you removed a "lower" MOSFET and the "upper" MOSFET pulsed, there will be "current in motion" going through the inductor with no low-impedance current path to continue pulling current from. This will result in either damaged MOSFET driver or damaged upper MOSFET. The same applies if you remove an upper MOSFET but don't remove the lower one - then the lower MOSFET will pull current from the voltage rail to ground and then when it switches off, the inductor will try to push current through that MOSFET, despite being off. This again, will end up destroying either another MOSFET or the MOSFET driver.

Also, if you remove MOSFETs from a voltage rail with multiple phases (more than one set of MOSFETs and inductor), the other phases will end up trying to work harder to keep powering up whatever needs powered. So again, you will likely end up with friend parts.

Therefore, my suggestion is to NOT test hardware with MOSFETs removed.

The only exception to that is if you removed all of the MOSFETs for whatever voltage rail they pertain to. Thus, in a multi-phase design like the GPU VRM, you could test the card with power ONLY if ALL of the MOSFETs in the GPU VRM for ALL of the phases are removed.

Otherwise, the best approach is to replace the shorted MOSFET and also the MOSFET driver too, if you can source it somehow. Blown MOSFETs don't always take out their drivers... but when they do, the card/motherboard will just keep "eating" good MOSFETs and puffing out magic smoke. I know because I learned this the hard way too.

[taxxin]

Re: GTX 550TI Faulty GPU Chip keeps blowing MOSFETS?

Very useful information there!!!

yes I didnt have a good feeling about only removing only one of the mosfet pair, but thanks to you my theory is confirmed.

I will try to see if i have a spare driver lying around, im doubtful, but i do have some compatible mosfets on another lenovo board that i have, perhaps that driver might do the trick? from what i can see there is total 3 phases for the Vcore.