VOGONS


Fixing geforce cards

Topic actions

First post, by Warlord

User metadata
Rank l33t
Rank
l33t

So I have a few geforce 2,3,256 etc cards that display have vertical lines, or stripes. It varies in severity per card. I read this is usually because the vram has gone bad. But I do not know enough about fixing these cards to know any better. Would soldering new ram on a card with those kinda problems fix them.

geforce 2 ultra that has thin Vertical green lines
GTS Pro has Vertical bars.

Last edited by Warlord on 2020-09-25, 18:29. Edited 1 time in total.

Reply 1 of 18, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

You mean something like that? Smeared output on ELSA Synergy II PCI (RIVA TNT2)

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

Reply 2 of 18, by Warlord

User metadata
Rank l33t
Rank
l33t

Similar but the geforce 2 ultra looks more like this..

Reply 3 of 18, by Warlord

User metadata
Rank l33t
Rank
l33t

The 256 looked more like this but it did alternate some patterns of stripes.

Reply 4 of 18, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

Half-dead RAMDAC.

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

Reply 5 of 18, by Warlord

User metadata
Rank l33t
Rank
l33t

Is it fixable? Anyone know. Are you sure thats whats wrong? I thought the Ramdacs were integrated. Also the 256 is a DDR and has DVI, and I thought DVI doesn't use the ramdac becasue its already digital. That card is a dual output and I didn't see any difference on either output

Reply 6 of 18, by Warlord

User metadata
Rank l33t
Rank
l33t

The geforce 256 DDR Pro has a DVI-D connector It doesn't have the pins to connetc a DVI-I to VGA or anything its just a strait digital connector. I'm pretty sure that does not use a RAMDAC at all. And it has the corrupted display. I could be wrong but I think that ramdac isn't the issue.

Reply 7 of 18, by Hanamichi

User metadata
Rank Member
Rank
Member

Can you take a photo of the output? What do the caps look like on the boards? I have a board similar to what you describe.

If you have the issue using the dvi port on the card to a dvi port on the monitor (all digital) then it's not the ramdac I guess.

This one is different:

Trying to save a Quadro FX2000 (FX5800) AGP which artifacts at post. Caps look okay, artifacts mean GPU or VRAM i guess.

The attachment _20200927_003728.JPG is no longer available

2 goes at refllowing with a heatgun no luck yet, I'll probably give it one more shot at high temps.

I use the solder reel to know when I am close to the right temp to melt the solder underneath the chip.

The attachment _20200927_003824.JPG is no longer available

Reply 8 of 18, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

I could be wrong but I think that ramdac isn't the issue.

There're mostly speculations at this point why this happens. If it happens on pure digital signal, then maybe something else is dying in the chip.

Trying to save a Quadro FX2000 (FX5800) AGP which artifacts at post. Caps look okay, artifacts mean GPU or VRAM i guess.

FX5800 had RAM failures mostly.

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

Reply 9 of 18, by Hanamichi

User metadata
Rank Member
Rank
Member

FX5800 had RAM failures mostly.

Guess it's the DDR2, i'll hit the memory hard this time 😀

Reply 10 of 18, by Warlord

User metadata
Rank l33t
Rank
l33t

well my suspicion is bad ram chips, since there is 8 of them on my 256 it's impossible to know which one or ones. I'd have to find a couple donor cards with the same ram chips. Re flowing bad ram i doubt will help.

Normally I wouldn't worry about it, but its a rare card.

Reply 11 of 18, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

well my suspicion is bad ram chips

If problem is visible only on viewport and there's no direct corruptions in 3D graphics, like strange textures, then it's not VRAM.

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

Reply 12 of 18, by shamino

User metadata
Rank l33t
Rank
l33t
Hanamichi wrote on 2020-09-27, 00:43:

FX5800 had RAM failures mostly.

Guess it's the DDR2, i'll hit the memory hard this time 😀

I had a similar problem on an FX5200 PCI. I tried using freeze spray for the first time ever, and it seemed to give me the answer.
I sprayed each of the chips (including the GPU) with freeze spray and found that spraying one particular RAM IC would cause the glitchy image to change. Nothing else I sprayed made a difference.

It seems like this was an effective test, but I haven't fixed it so can't confirm it wasn't a red herring.
I tried reflowing the chip but the problem remained. Next step will be to replace it sometime, whenever I get a donor chip and have the time and motivation to work on it. Unfortunately that's not nearly as feasible with BGA RAM like I think your card uses.

Reply 13 of 18, by Warlord

User metadata
Rank l33t
Rank
l33t

Found this site. based on the screen shots my cards are probably ram related.
http://www.playtool.com/pages/artifacts/artifacts.html

Early geforce cards used normal sdr or ddr the 256 DDR will be easy to source doner ram. I'm not as sure about the geforce gts ultra.

Reply 14 of 18, by Hanamichi

User metadata
Rank Member
Rank
Member
shamino wrote on 2020-09-29, 06:15:
I had a similar problem on an FX5200 PCI. I tried using freeze spray for the first time ever, and it seemed to give me the answ […]
Show full quote
Hanamichi wrote on 2020-09-27, 00:43:

FX5800 had RAM failures mostly.

Guess it's the DDR2, i'll hit the memory hard this time 😀

I had a similar problem on an FX5200 PCI. I tried using freeze spray for the first time ever, and it seemed to give me the answer.
I sprayed each of the chips (including the GPU) with freeze spray and found that spraying one particular RAM IC would cause the glitchy image to change. Nothing else I sprayed made a difference.

It seems like this was an effective test, but I haven't fixed it so can't confirm it wasn't a red herring.
I tried reflowing the chip but the problem remained. Next step will be to replace it sometime, whenever I get a donor chip and have the time and motivation to work on it. Unfortunately that's not nearly as feasible with BGA RAM like I think your card uses.

Shamino, now that's a name associated with good quality products! My EVGA X58 Classified never misses a beat 😁

Interesting freeze spray trick, does that make it worse or better? I guess the dodgy solder contracts and the glitches increase?

That's the disadvantage of the newer AGP cards, no memory swapping possible with normal tools.
Thanks for the clue, adding freeze spray to the shopping list.

I will try one more time, especially on the vram at high temps.

Reply 15 of 18, by Hanamichi

User metadata
Rank Member
Rank
Member
Warlord wrote on 2020-09-29, 07:08:

Found this site. based on the screen shots my cards are probably ram related.
http://www.playtool.com/pages/artifacts/artifacts.html

Early geforce cards used normal sdr or ddr the 256 DDR will be easy to source doner ram. I'm not as sure about the geforce gts ultra.

The agp bus defects picture looks like an interesting filter 🤣
I lost an X850XT PE back in the day, exactly like the CS Dust2 pic in the article. Just a few dots and the odd glitch to start off, within a week it was dead.

Reply 16 of 18, by shamino

User metadata
Rank l33t
Rank
l33t
Hanamichi wrote on 2020-09-29, 18:43:

Shamino, now that's a name associated with good quality products! My EVGA X58 Classified never misses a beat 😁

I had to look that up to figure out what you're referring to. 😀 But nah, I wasn't thinking of that when I made the username.

Interesting freeze spray trick, does that make it worse or better? I guess the dodgy solder contracts and the glitches increase?

The firmware of the card I was working on displays a bootup screen of it's own, telling you what card it is and such.
It was generally locking up on that screen, and the image was corrupt.

It's hard to remember if the freeze spray made it better or worse, but all I can say for sure is that it made it change. The glitches would even jump around a bit while the RAM was being hit with the cold.
At least in my case I don't think it was an issue of dodgy solder, because going around all the legs with an iron, some flux, and a little fresh solder didn't change anything.

I think the cold was affecting something internal to the RAM IC, or else it's just a distraction. But the fact that it's the only one of 8 chips that reacted this way makes me think it means something.

I did it with the GPU exposed so I could try spraying it also. The FX5200 GPU doesn't heat up very fast so it was fine, but I don't know if there's any danger doing that on more powerful cards.

Reply 17 of 18, by Hanamichi

User metadata
Rank Member
Rank
Member
shamino wrote on 2020-09-29, 19:32:
I had to look that up to figure out what you're referring to. :) But nah, I wasn't thinking of that when I made the username. […]
Show full quote
Hanamichi wrote on 2020-09-29, 18:43:

Shamino, now that's a name associated with good quality products! My EVGA X58 Classified never misses a beat 😁

I had to look that up to figure out what you're referring to. 😀 But nah, I wasn't thinking of that when I made the username.

Interesting freeze spray trick, does that make it worse or better? I guess the dodgy solder contracts and the glitches increase?

The firmware of the card I was working on displays a bootup screen of it's own, telling you what card it is and such.
It was generally locking up on that screen, and the image was corrupt.

It's hard to remember if the freeze spray made it better or worse, but all I can say for sure is that it made it change. The glitches would even jump around a bit while the RAM was being hit with the cold.
At least in my case I don't think it was an issue of dodgy solder, because going around all the legs with an iron, some flux, and a little fresh solder didn't change anything.

I think the cold was affecting something internal to the RAM IC, or else it's just a distraction. But the fact that it's the only one of 8 chips that reacted this way makes me think it means something.

I did it with the GPU exposed so I could try spraying it also. The FX5200 GPU doesn't heat up very fast so it was fine, but I don't know if there's any danger doing that on more powerful cards.

Haha I didn't think so but you never know! I'd guess Ultima actually.

Your reasoning gave me a good clue for a card I managed to save which I'll post about below.

No freeze spray yet but with this issue I pressed on each ram chip, only on the bottom left did pressing hard make a difference to the artifacts shown on screen 👍

Reply 18 of 18, by Hanamichi

User metadata
Rank Member
Rank
Member

@Warlord any luck with the cards so far?

I can report 1:1 fail and success.

For the FX5800/FX2000 with the problems shown above I tried 3 times to cruedly reflow. Final go I really hit with the heat and now it won't post. If it was an actual FX5800 I'd recap etc but for now RIP

Victim #2- TI4600 with this issue:

The attachment _20201011_003238.JPG is no longer available
The attachment _20201011_003202.JPG is no longer available

Tried Shamino's trick kind of and what do you know putting pressure on or around the bottom memory pad not only changed the artifact colour but after a reboot temporarily made it disappear.

Sadly this behavior was short lived first showing a new artifact as below then going back to the same artifact and no pressure later on would change it.

The attachment _20201011_003648.JPG is no longer available

At least there was hope that it was fixible. Took off the heatsink, note to all Ti4600/4400 FX5800/FX5950 users... your TIM pad is basically like dried gum by now and needs replacing. The heat gun and a scraping device was needed to remove it.

The attachment _20201011_003920.JPG is no longer available

Preheated the board with the heatgun moving around the core, then made sure to raise the temp to the point solder melted on top of the core (my temperature test) just a bit + 10secs more heat for good luck. Let it cool for 30 mins at least.

What do you know Ti4600 working happily and all games working fine! Of course no idea how long it will last.

The attachment _20201011_003717.JPG is no longer available

Having experienced the temporary reprieve described earlier and combined with the fact that the artifact is a uniform 1/4 ish of the screen tells me this was VRAM related. However, my guess is that the actual root issue is the cooler, as the TIM dries the core dumps more heat on the PCB (you can really feel the difference by putting your finger on the board behind the core before and after) warping the PCB it as it cools, heats and repeat with use.

Over time some issue between the core and the VRAM chip occurs, in this case maybe a corresponding solder ball at the peripheral of the GPU chip (not sure if pressing around the VRAM would influence this or not). By cruedly reflowing the GPU and fixing the cooler issue it seem this one can live on a bit longer.

Sorry for the essay, wanted to share in case it helps others. 👍