VOGONS

Common searches


First post, by pentiumspeed

User metadata
Rank l33t
Rank
l33t

Other day, repaired the blown components (literally blew off and melted inductor SMDs, on this one used 5 small inductors in parallel at power input), and replaced the shorted power MOSFET, did check stuff and is not abnormal anymore, even the gate pin is good means gate driver IC is fine on a EVGA GTX 960 mini. GPU and memory shows low ohms as normal due to their silicon design. Today, took some time to test the repaired GPU and still not detected during POST and nothing blew up. Previously last year, I attempted repair by replacing mosfet and gate driver on a co-worker's GTX 970 and it worked but degraded condition that it wouldn't get to full potential.

Personally, I feel that I should give up on buying cheap modern GPU that needs repair and save up for one that is reliable. IE: GTX 980 and up.
Another reason, sellers does not show us the GPU card with heatsink removed to judge the condition of the PCB board to see is there cratered burn holes in the PCB etc.

How common is EVGA and other brands GPU prone to blowing up? Just blew. Not from liquid cooling spill or lightning strike.

Honestly, I don't think price is reasonable at any rate these days even post-etherum merge price drop. My "wall" is around 300 CDN. If pushed, I could go for 500 but that it. But to get 3080 at 600-700. Ahhh no. That exceeds total cost of a new PC parts. To double the cost of PC just because of a cost of a GPU. That made me very reluctant. Consoles like Series X or PS5 is cheaper to buy but PC is high quality resolution is main draw.

Sigh. I have long experience on repairing stuff for a living (work), and even I had to repaired several consoles that had shorted mosfets and they simply worked fine and no other damage made, thanks to high quality power supply tripping on shorts instantly. But not in this PC, power supplies have reputation to have different quality of protective circuit not tripping properly.

Cheers,

Great Northern aka Canada.

Reply 1 of 11, by Shponglefan

User metadata
Rank Oldbie
Rank
Oldbie

Personally, if I am buying hardware I'm intending to use over time, I just buy something that I know is working.

Otherwise, it's a big risk in buying broken hardware that may either be unrepairable or even with repair not particularly reliable.

The only time I would buy hardware I know is broken is either for the challenge / fun of trying to repair something, or if I am cannibalizing it for parts.

Tandy 1000 TL build
286 Epson build
Ultimate Windows XP build

Reply 3 of 11, by pentiumspeed

User metadata
Rank l33t
Rank
l33t

What can I do to lessen the chances of these so called "exploding"? Anything else I can do about this? Buying new one at full price is not feasible.

What about AMD GPUs?

Cheers,

Great Northern aka Canada.

Reply 4 of 11, by BEEN_Nath_58

User metadata
Rank Oldbie
Rank
Oldbie
The Serpent Rider wrote on 2022-11-21, 00:36:

All modern Nvidia cards are prone to exploding.

Thankfully I got a 10 series card 4 years ago and never needed a upgrade

previously known as Discrete_BOB_058

Reply 5 of 11, by The Serpent Rider

User metadata
Rank l33t
Rank
l33t
pentiumspeed wrote:

What can I do to lessen the chances of these so called "exploding"?

Undervolting and curve editing via GPU tweakers.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 6 of 11, by TrashPanda

User metadata
Rank l33t
Rank
l33t
The Serpent Rider wrote on 2022-11-21, 00:36:

All modern Nvidia cards are prone to exploding.

Citation needed.

Oh noes, the cap let the shmooo out 😁

Reply 7 of 11, by The Serpent Rider

User metadata
Rank l33t
Rank
l33t
pentiumspeed wrote:

What about AMD GPUs?

AMD have tendency to overbuilt their stuff. But some of their cards are very prone to die quietly, without any success to repair, like everything with HBM2 memory.

TrashPanda wrote:

Citation needed.

Lack of OCP in any meaningful form should be enough. Nvidia cards now are repeating GTX 590 fate basically. If something goes wrong on software end (driver) - hardware can't protect from VRM failure. That's why "killer games" like New World are a thing.

You can add safety net with curve editing though, when GPU absolutely won't go beyond set clock speed and voltage, instead of vague average power limit with crazy overshoots over target.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 8 of 11, by BitWrangler

User metadata
Rank l33t
Rank
l33t

On the other hand, if you can figure out one with a common problem that's basically a "fuse" then fixing one and knowing how, will mean you can keep it in service many years.

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 9 of 11, by The Serpent Rider

User metadata
Rank l33t
Rank
l33t
BitWrangler wrote:

On the other hand, if you can figure out one with a common problem that's basically a "fuse" then fixing one and knowing how, will mean you can keep it in service many years.

Fixing a fuse without touching anything else in most cases will lead to more damage, because something else on a board tripped it. If a card even have one. Fuses only protect from catastrophic failure, like total PCB meltdown scenario, not from hardware damage itself.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 10 of 11, by ODwilly

User metadata
Rank l33t
Rank
l33t

It seems like every single Nvidia 700 series card Iv had from EVGA has failed from blown power circuitry or VRMs.

Main pc: Asus ROG 17. R9 5900HX, RTX 3070m, 16gb ddr4 3200, 1tb NVME.
Retro PC: Soyo P4S Dragon, 3gb ddr 266, 120gb Maxtor, Geforce Fx 5950 Ultra, SB Live! 5.1

Reply 11 of 11, by The Serpent Rider

User metadata
Rank l33t
Rank
l33t

GeForce 780/780Ti/Titan/Titan Black/Titan X/980Ti had weak inductors on GDDR5/GDDR5X rail. They tend to melt when overstressed.

I must be some kind of standard: the anonymous gangbanger of the 21st century.