Asking for help diagnosing old GPU

Columbus43219

2545

Old Nvidia GPU that went dead a few years back. There are some heat sinks still attached to the VRAM(?) chips. The main chip is a G92-421-B1. I'd like to repair it to have as a backup in case my current one fails. Also for the practice.

From what I've seen on YouTube, I went through some basic diagnostics by measuring for shorts and checking resistance across some components.

Can someone who knows this stuff better tell me if one of the chips is reading as failed? There are two acting "different" than the rest I think. Upper MOSFET(?) on the left showing open, and the two on bottom right showing open.

That may be normal for those type of chips though, I'm not sure.

Left side "close up"

Right side "close up"

No shorts in the 6-pin power connector. Good continuity between the +12v pin in all the points marked in red.

Using my multimeter, set to 20k resistance, measuring both directions (switching red/black leads) the red chips read 2.5 each, the green chips read 10.0, and the yellows read open/infinite.

Chip types: 938 48 55NG(red)... 934 48 50NG(yellow)... D472A BA8Z15(green)... D452A BA8113(blue)

Back on the left, got good contact, 0 ohms, between 1st pin of PCI port and the capacitor (yellow). 0 ohms across resistor in red. 0 ohms on the Fuse (E).

When using the multimeter again on 20K resistance... the yellow chip read open/inifite. The red chip starts at close to 0, then ramps up rapidly to infinite in around 8 seconds.

Chips are D4120 BA9R11 and LD117AL ADA01WGR8

gpu

electronics

video_card

advice

SmokingGandalf

Was it the card that went dead or the slot in the MoBo? You try it on another board?

5 years ago |

I can't remember! I only had one pc at the time, and I traded in another used card (Radeon) and it worked OK.

tiderfish

I had an evga 970 gtx that had a known mofsfet issue, so I replaced it, and it worked until it let the smoke out. Stunk up the house

I was lucky it didn’t kill my motherboard.

You can't let the magic smoke out! How did you learn about the known issue? Might be something similar.

I just fixed it, at least temporarily. I had tested it powered up and all the voltages looked right. I tried flux and rework heat gun.

4 years ago |

innagaddavidababy

For gods sake, just throw it away! If your time is worth more than 10 cents an hour, it would be cheaper to buy a new one.

I agree that it's not a value/money decision. I am just learning something new and like to fix things if I can. I used to do more stuff

like this, but the SMD things intimidated me. Now I've been procticing those, and remembered this old card.

He is learning a new skill and keeping toxic materials out of the waste stream. Let bygones be bygones

Guess what! I tested voltages while it was on and everything looked right. I did a Hail Mary and "reballed" the GPU with flux and my rework

heat gun and it WORKS!!

stercusmoriturussum

So, nothing unusual from your measurements, Nextstep would be to dig up the datasheets for these parts (google the numbers) and check 1/?

2/ what it was that you measured. The one that's ramping up is a capacitor (enough of them in the area) charging up from your DMM.

That's cool, I didn't know it would do that

I got it working! I tested as much as I could and found nothing wrong. The voltages were there all the way from the power sources to the GPU

so I tried flux and a rework heat gun as a last resort. It started working. I used a low-viscosity flux designed for reballing.

Comments (20)