Hi,

Working on a dead 980TI, came with blown fuses for 6+8 PIN connectors and shorted few FDMF6823A. While "A" usually means 3.3V PWM logic 6823A is a 5V one according to datasheet since I am aware there were 3.3 & 5V PWM versions.

Long story short replaced all 8x 6823A with brand new SIC789 from digi-key (did not have the other part) - 60A, 5V PWM pin compatible what could go wrong. The card has started, displayed image, booted to windows, installed drivers without issues, no error 43. All was good. GPUZ was showing clock adjustments as normal, temp 50-60C. Started FurMark and before even picture came on the card "puffed" and did not smell right. Mosfet - 3rd from the top exploded and took half of pad with it, looks like it could still be repaired with copper plane but number of tries becomes limited...

Did I do wrong with SIC789 (not A) replacement for 6823A or there's something else at play?

After boot to windows I checked all coils with oscilloscope and phases were active with pulses. Resistance of core 2,4Ohm, Mem 69Ohm, all 3 fuses survived - 1. (12v from pcie 4k and growing) 2. 40k, 3. 2.3M

I did the test with cooler on.

One thing I did not check was BIOS, maybe someone loaded customized freq/voltages but would that cause immediate explosion? Dont think so.

it is unclear in what place they changed it . at least a boardview was posted or a photo of the board indicating the place of replacement

This one comes from Palit, 980TI jetstream. Essentially I replaced all integrated buck converters/mosfets (drmos) for Vcore (8 of them) from 6823A to SIC789. They are fed by 3 phase NCP81162 doublers and single 81174 controller like on a reference board.



Exploded 3rd from top which I assume is 3rd phase (powered from external power connector).

I have two other dead 980ti from evga with holes in PCB. One is fitted with 3V PWM and 6820A mosfets (drmos) and the other has 5V PWM and was fitted with SIC789.

Don't think change to SIC789 alone was the issue as why would it explode only when starting furmark? It is also 60A capable, and used by EVGA in their power stage with same doublers and controller. Thinking about changing PWM doublers but here same question, if they would be bad it should explode immediately after replacement on power on. I also checked PWM pulses on coils for all phases hence did not think about changing PWM doublers.

Maybe something else important, it did not burn its way to 12V power line like they typically do. It burnt its path to the coil and Vcore but there is no short, measuring 2.4Ohms and before starting furmark card was operating normally. Now with mosfet removed there's no short.