• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Graphics card sudden death

Vince said:
Its arrived! : Very quick test, its got a short to ground on 12v. This fet is blown:



Of course I cant find any info on it... Edit: Worked it out they are an FDMS3604S Dual N-Chanel Fet and the short is on the memory VRM not Vcore.
Click to expand...
Impressive stuff, do you do that for a job?
 
Last edited:
Vince said:
Its arrived! : Very quick test, its got a short to ground on 12v. This fet is blown:

...

Of course I cant find any info on it... Edit: Worked it out they are an FDMS3604S Dual N-Chanel Fet and the short is on the memory VRM not Vcore. Down side is that I can only find them in china.
Click to expand...

Wow, quick work! :)

Does that give you any inkling as to what might have been likely to have caused it?

Vince said:
Ok no i cant. Ill need to order one of those fets in.
Click to expand...

Cool. Cheers. This is all indeed very impressive! :)
 
strumpusplunket said:
Wow, quick work! :)

Does that give you any inkling as to what might have been likely to have caused it?



Cool. Cheers. This is all indeed very impressive! :)
Click to expand...

Were you overclocking memory? These fets should be able to deliver around 224amps at peak assuming you could keep them under 125 degrees which realistically wouldn't happen as the chokes just wouldn't cope with that and that would require somewhere around 70/80w of cooling on the fets. What im saying is you shouldn't be able to push them hard enough to fail on a 1080ti. But that fet im pointing at has just failed it's got a short to ground at it's drain whereas the other one on that phase is showing 10k at the drain, as are the two on the other memory phase. I bet if I took it off the shorts on the card would all go away and it would boot.
 
Vince said:
Were you overclocking memory? These fets should be able to deliver around 224amps at peak assuming you could keep them under 125 degrees which realistically wouldn't happen as the chokes just wouldn't cope with that and that would require somewhere around 70/80w of cooling on the fets. What im saying is you shouldn't be able to push them hard enough to fail on a 1080ti. But that fet im pointing at has just failed it's got a short to ground at it's drain whereas the other one on that phase is showing 10k at the drain, as are the two on the other memory phase. I bet if I took it off the shorts on the card would all go away and it would boot.
Click to expand...

No, I never overclocked the card at all. Since cards started with all the boost clocks I gave up on overclocking GPUs.

Any cause for concern that the PSU might have deliverd a power spike that shorted out the fet?
 
strumpusplunket said:
No, I never overclocked the card at all. Since cards started with all the boost clocks I gave up on overclocking GPUs.

Any cause for concern that the PSU might have deliverd a power spike that shorted out the fet?
Click to expand...

I will whip that fet off in a bit and check that the shorts go away. Also because i cant find a datasheet on it's markings of GA7V1F (looks like an infineon part) Ill be able to check it's pinout and make sure it's pin compatible with the onsemi replacement. To clarify on fet power delivery above thats the data in terms of the whole vmem rail.

Honestly it's very unlikely to be psu related imo. Most of the time they go because of age, wear, heat... Looks like the cooler was a bit clogged with the old dust and that over a period of time can stop air hitting the fet. Prolonged that can cause failure.

Also this card is an absolute beast, what is it an auros extreme?
 
Last edited:
Vince said:
I will whip that fet off in a bit and check that the shorts go away. Also because i cant find a datasheet on it's markings of GA7V1F (looks like an infineon part) Ill be able to check it's pinout and make sure it's pin compatible with the onsemi replacement. To clarify on fet power delivery above thats the data in terms of the whole vmem rail.

Honestly it's very unlikely to be psu related imo. Most of the time they go because of age, wear, heat... Looks like the cooler was a bit clogged with the old dust and that over a period of time can stop air hitting the fet. Prolonged that can cause failure.

Also this card is an absolute beast, what is it an auros extreme?
Click to expand...

Just normal Gigabyte Aorus, I think. Though I threw away the packing ages ago so can't check to be sure.

Good to know it's not likely to be the PSU. I had been a bit paranoiod about my new card going the same way :)
 
strumpusplunket said:
Just normal Gigabyte Aorus, I think. Though I threw away the packing ages ago so can't check to be sure.

Good to know it's not likely to be the PSU. I had been a bit paranoiod about my new card going the same way :)
Click to expand...

Well I took the fet off, shorts are gone, I tested all of the vcore vrm, and the pcie power, everything there looks fine, core resistance looks OK. I think it should be all good.



Checked the pinout against the onsemi replacement and that will also work. Happy days.

And this is the reason the fets I have won't work even though the specs are basically comparable...



Different pinout... Below might go some way to explaining why these fets die.. no direct contact wit the cooler and directly under a part of the cooler that's bent over with no air flow through it... Nice...



 
Last edited:
@Vince nice work.

The flat part of the cooler looks like it was originally intended to contact all those fets, but that octagon thing between the two pairs stands taller, i wonder if that was added in post PCB design of a later revision of the PCB?
 
humbug said:
@Vince nice work.

The flat part of the cooler looks like it was originally intended to contact all those fets, but that octagon thing between the two pairs stands taller, i wonder if that was added in post PCB design of a later revision of the PCB?
Click to expand...

I think it's just an oversite. The fets are well over spec for application but I bet you they get toasty. couple of little heatsinksinks on them might help a lot. 3 week estimate on parts which i have already now ordered :( Card is back together in the box it came in and on my bench with about 50 other broken things that are also waiting for parts. :cry:

The board on this thing is seriously thick it was just sucking up heat from my rework station. Had to turn it up to 11 to remove that fet. Never good when you have alloy electros close by as them things like to explode under too much heat.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom