Nunoalex wrote on 2024-11-04, 18:58:
Indeed my friend
so many times dead boards come back to life after a few power cycles and a few reseats
I have no scientific explanation but something inside that silicon must be sleeping and needs time to wake up 😀
Never had that experience. My stuff usually either always works or it doesn't at all. I've had a few boards work "shortly" and then die... but those are outlier cases. Also, I tend to recap most stuff I get if it's of any value (and sometimes even if it isn't.) So I suppose that could be why too. 😀
Wes1262 wrote on 2024-11-05, 07:03:
x850xtpe
"perfect conditions tested!!!!!!!!!!!!!!!!!!!!11111111111111111111111111oneoneoneoneoneoneoneoneoneone"
:roll eyes:
Damn!
This is why I rarely buy "tested" stuff on eBay anymore, except from sellers that have proven feedback with selling old electronics and hardware. If it's some fly-by ebayer or a "thrift-like" store seller that has all kinds of stuff (not just electronics), then good chance they have no idea if the item they are selling really is working or not.
A long while back, I bought a Diamond Radeon HD3870 that was supposed to be working. Of course, as soon as I got it, I noticed it was missing a large logic gate IC on the back (some 74HC-- IC), so I didn't even try to power it on. Was going to email the seller... but then with a bit of research online, I found out what the missing IC should be and found the same exact one on a scrap motherboard. So I just fixed the card and left it at that. Didn't even bother to tell the seller. They had all kinds of clothes and other items, so exactly like an online thrift store. Good chance they just didn't handle/store the card properly, hence how the IC got knocked out. But whatever, I fixed it, the card worked and I couldn't be bothered any further. What's funny is I paid quite a bit (at that time) to get that card. A year later, I acquired something like 2 or 3 more HD3870's, mostly untested... but they all worked and were much much cheaper.
Trashbytes wrote on 2024-11-05, 08:01:
looks like a memory issue, might be temporarily fixable by reflowing the memory ICs.
Memory issue it is... but it won't be the memory IC's that are faulty here (they very rarely are.) It's more than likely the GPU data lines that go to the memory ICs that are going bad... so bad GPU, in short. Thus I'd say reflow the GPU chip first. If that doesn't bring it back, reflowing the memory chips won't do squat. The myth that the BGA solder under the memory ICs is going bad really needs to go away.
Trashbytes wrote on 2024-11-05, 08:52:
Give it a reflow with a hot air gun on the memory ICs, use Kapton tape around them if you have it so SMDs dont get heated up.
That's another big NO-NO I've seen in so-called "repair" videos online.
Do not mask off anything. Masking areas can cause "cold" spots in the PCB, and that can cause warping and uneven pressure. It's better to let everything heat up. Only things you might want to mask off is any electrolytic capacitors and plastic connectors that may be close by the hot air blast radius (so that the former doesn't overheat and the latter doesn't melt.)
Trashbytes wrote on 2024-11-05, 08:52:
If this improves the artifacting then it might be worth reballing
Reballing isn't going to fix a bad GPU or a dying memory IC.
It's 2024... people need to stop believing this myth/crap.
I've worked in a console "repair" shop back in the day and did my fair share of reballs under the direction of the owner. Half the time, they didn't do jack. The other half, they worked... only for a while - just the same as the reflown chips. We came to the conclusion that it was just the multiple heat cycles that made the reballs (slightly) more likely to work than the reflows. This was even more so evident with PS3, where just about neither worked. Unfortunately, neither I nor the owner knew about the bad NEC/Tokin caps going bad... so we just went reflowing and reballing PS3's like madmen, only to get very few working. No surprise, though - PS3 uses a variation of the GeForce 7900 series GPU, which are among the worst in the "bumpgate era".
Trashbytes wrote on 2024-11-05, 10:33:
reflowing may help it may not at this point it cant hurt. I doubt itll last long even if it does help
That's about the only part I agree with.
And in case that a reflow does "fix" anything, the only way you *might* prolong the "repair" is by keeping everything much much cooler than it would be stock. Often times that means gutting the stock cooler and replacing it with something much bigger. Ideally, temperatures on reflown stuff (and actually, even on "virgin" silicon too) should be kept under 55C at all times. I used to say 60C, but even that's a bit too high in some cases. 50-55C under 100% load can become challenging on the higher-end GPUs, especially when the TDP goes above 40-50W and the stock cooler is only a single-slot. It just won't work. There's a reason why so many of these retro GPUs (and even modern ones) die: it's because they never had a proper / adequately-picked cooler for the TDP of the GPU to begin with. 70C under load is NOT OK. Neither is even 60C. Of course, all manufacturers even way back in the day would spit lies all day long, telling everyone those temperatures were "within spec" - yes, within spec if you're OK with a 5 year expected life cycle (on average.)
