0
\$\begingroup\$

I bought a computer for a person in which, when delivered, the video card (Gigabyte Nvidia GTX 570) never gave output. The only way the display got basic VESA output was when I moved the card into in the PCIe 8x slot (was on PCIe 16x) but of course blue screen on drivers install.

The warranty was void since they detected burned capacitors and excess of solder.

I've noticed in the pics from warranty that some components were burned and others where some solder was melted. You can see the pics here: http://imgur.com/a/CVy3y

How can this damage occur? My guesses are:

  1. Overvoltage: I think I would have noticed any behavior/sound of it; also, the PSU should block this.
  2. Wrong overclock configuration that led to heating the GPU.
  3. Video card was defective on arrival in some way.
\$\endgroup\$
  • 2
    \$\begingroup\$ It's not clear who bought what from whom, but whoever received the dead-on-arrival equipment should return it to the seller for a refund. \$\endgroup\$ – Olin Lathrop Dec 9 '11 at 13:09
  • 2
    \$\begingroup\$ You're pretty unlikely to be able to get a refund, I suppose... since it's likely taken a while to get warranty to look at it and more importantly whoever sold you that is a sleazy bastard. They either did a poor repair job that broke in shipping, or more likely they failed to repair it entirely and just sold the busted card. You can hope they somehow bought it from someone else and didn't know... \$\endgroup\$ – darron Dec 9 '11 at 18:09
  • \$\begingroup\$ Unfortunately I bought online in Mexico; the Gigabyte offices here don't want to make valid the warranty and the person who sold the computer told me "maybe I touch it when was power on"! hard to believe. \$\endgroup\$ – Daniel Guerrero Jan 18 '12 at 18:39
3
\$\begingroup\$

The on each photo, the central part looks hand-soldered to me. On the last two pictures for example the excess amount of burnt flux is visible. Also while the components do no appear to be beautifully soldered, I don't see why they wouldn't work because they all appear to have electrical connections.

If nobody tried to hand-solder the components, it may be possible that someone tried to bake the card and may have added extra flux to the components to make the process go more smoothly. If that was the case, then we're probably not seeing the main damaged area. People usually bake cards when they believe that the solder balls on one of the BGA chips (and that's usually the GPU itself) are cracking and have bad connection. Unfortunately from what I've heard the only relatively cheap way of seeing the damage there is to make x-ray images of the card itself and even then not many people would go through the trouble of actually replacing a BGA component. The damaged capacitors also in my opinion support the baking idea since they can easily be overheated during the process and leak or explode.

If the card really wasn't "repaired" by someone else, then only thing that could in my opinion cause such problems would be really really bad overheating. There are stories of SMD resistors for example desoldering themselves because they heat up too much, but then again you have some problematic capacitors too.

\$\endgroup\$
  • \$\begingroup\$ I agree with this. There's no way that's from a volume production line. Someone tried to either fix or enhance it, and failed. \$\endgroup\$ – darron Dec 9 '11 at 18:04
  • \$\begingroup\$ > There are stories of SMD resistors for example desoldering themselves because they heat up too much - I've never heard of this. Most graphics cards will use lead-free solder, and that's soldered at over 200°C, or almost 400°F. There's no way that a hot graphics chip or capacitor could even approach that temperature and still have a chance of functioning. \$\endgroup\$ – Kevin Vermeer Dec 10 '11 at 18:26
  • \$\begingroup\$ @KevinVermeer Don't dismiss such anecdotes out of hand. I've seen SMT diodes floating on molten solder due to a PCB internal short. Obviously they were still conducting; I don't know for a fact but would be inclined to suspect they still functioned as diodes at least when cool. Components on prototypes routinely survive all sorts of non-production assembly and re-work that easily exceeds the standard "10 seconds at 300C" type data sheet spec. There's a big difference between what something is rated to tolerate, and what it can often survive while remaining somewhat functional. \$\endgroup\$ – Chris Stratton Dec 10 '11 at 19:01
  • \$\begingroup\$ @Kevin Vermeer♦ A quick search gave me a comment on this question and this article. I actually heard the anecdote on this site for the first time and I'm pretty sure that I've heard it more than once but I can't find the post where I read it for the first time at this time. I do know that in that post, the component was a resistor. \$\endgroup\$ – AndrejaKo Dec 10 '11 at 22:43
  • \$\begingroup\$ Thanks for the comment, I think similar things, but I don't have much experience in electronics; just learned basic stuff in college. Unfortunately there were many on the line which could modified \$\endgroup\$ – Daniel Guerrero Jan 18 '12 at 18:41
2
\$\begingroup\$

What company made the card? Those parts could only be that way coming out of the factory. It seems the company should replace the card for you.

\$\endgroup\$
  • 3
    \$\begingroup\$ It really looks like someone tried to bake the card to fix some perceived GPU problem. The fact that they'd then sell it as part of a PC after that obviously didn't work is disgusting. \$\endgroup\$ – darron Dec 9 '11 at 18:02
  • \$\begingroup\$ @Darron Yes, I see what you mean now. From the author's question I supposed it was bought from an online store or such. \$\endgroup\$ – Mark C Dec 11 '11 at 23:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.