VOGONS


First post, by Socket3

User metadata
Rank Oldbie
Rank
Oldbie

Hey guys. I need help with a couple of Voodoo 2 cards. One is a Viewtop Vulcan 2, and the other is a Gainward Dragon 3000, both have similar issues.

The Vulcan 2 shows blue or red spots on textures, particularly in games that support multitexturing, like quake 2. Spots, not pixels, like stains. The spots change with the scene lighting. They are very visible in areas with shadows, and go away completly in scenes with strong light or when firing a weapon. This texturing or lighting bug or artefact appears in quake 2, but does not manifest in quake 1 whatsoever.

The gainward card has a blue hue over the whole picture, witch is also influenced by lighting, but on the gainward card the "blueing" is always present - in both quake 1 and 2. Sometimes the image goes back to normal for a second, then becomes blue shifted again. The image also changes to normal (no more blue shift/blueing) when firing a weapon or in very bright scenes, just like on the Gainward card.

The image coming from the 2d card is normal, so I don't think it's a d-sub connector issue - that and the occasional 1-2 seconds when the image goes back to normal makes me think the Gainward has a similar issue as the Vulcan card - a bad solder joint someware - but where do I even begin to look? I put both cards under my microscope, and found cracked solder joints on memory modules.

The Vulcan card had cracked joints at the edge of one memory chip, and the chip's packaging is also damaged. I re-soldered the legs, but there was no change. Here are some pics of the Vulcan 2 card:

frnt.jpg
Filename
frnt.jpg
File size
577.02 KiB
Views
212 views
File license
CC-BY-4.0
rear.jpg
Filename
rear.jpg
File size
582.47 KiB
Views
212 views
File license
CC-BY-4.0
chip.jpg
Filename
chip.jpg
File size
292.53 KiB
Views
212 views
File license
CC-BY-4.0
chip2.jpg
Filename
chip2.jpg
File size
289.54 KiB
Views
212 views
File license
CC-BY-4.0

And here's a pic of how the "artefacting" manifests on the Gainward card:

bug.jpg
Filename
bug.jpg
File size
202.91 KiB
Views
212 views
File license
CC-BY-4.0

I know it's not very easy to observe in the picture, so here's a short video: https://www.youtube.com/watch?v=BhrwKFmEJXQ

I'll post pics of the colored spots the Vulcan 2 shows tomorrow.

What could be causing this? I think it's bad solder joints, but where? I checked all memory chips on the Gainward card, and found 3 chips with cracked joints, witch I re-soldered and verified the joints with my microscope - no change.. I also checked the legs on both TMUs, none are bent, and it none seem to be loose... Same with the Vulcan 2 - found 4 crecked solder joints on the chip with the damaged package, but everything else is fine...

Please help

Reply 1 of 2, by Tevian

User metadata
Rank Newbie
Rank
Newbie

Well... Not sure but the damaged RAM chip in the pic looks pretty bad. It's hard to see, but it looks like the pins are "flared" a bit. Meaning some force caused them to move in a way that might suggest that side of the chip casing is cracked. If that's the case, the RAM chip might have internal damage.

You mentioned cracked joints, but you might need to do specific continuity testing from the pin to the end point of the trace where ever it goes.

With so much damage, you probably have a broken trace you simply can't see. I'd start doing the continuity testing or scoping the pins (if you have access) while they're live and compare. It might stand out...

Reply 2 of 2, by Socket3

User metadata
Rank Oldbie
Rank
Oldbie
Tevian wrote on 2024-03-29, 00:14:

Well... Not sure but the damaged RAM chip in the pic looks pretty bad. It's hard to see, but it looks like the pins are "flared" a bit. Meaning some force caused them to move in a way that might suggest that side of the chip casing is cracked. If that's the case, the RAM chip might have internal damage.

You mentioned cracked joints, but you might need to do specific continuity testing from the pin to the end point of the trace where ever it goes.

With so much damage, you probably have a broken trace you simply can't see. I'd start doing the continuity testing or scoping the pins (if you have access) while they're live and compare. It might stand out...

Thanks for the tip. The ram chip on the vulcan 2 looks worse in the pics then it does under the microscope, but I'll try to replace it with another one, shouldn't be hard to do.

I'll also take your advice and look for broken traces around the card. So far I haven't found any, but there might be "thinned out" or scratched traces, maybe even dirty pads I've soldered onto, making for poor electrical contact or introducing resistence.

Thanks for pointing me in a direction!