ps3specialist wrote:
the first picture show the area of the logic board under the graphics chip that show clearly the reason for the graphics failure the second picture show the same logic board after removing the solder, clean the dirt and refresh all the soldering pads , only the blinds don't see the reality and understand that it is not a design or manufacturing issue or a defective graphics chip,
The photo's you posted in above comment don't really prove anything.
You used flux and heated it up above liquid so the balls were soldered again.
Photo should be taken before any maniuplation.
Only (3D) rontgen will show probably.
You also wrote:
" not a design or manufacturing issue "
Before you wrote:
ps3specialist wrote:
The dust that you see is very fine dust granules , they don't just stay around the chips and they don't just go under the chips, you won't beleive me if I say they go under the solder, when I clean the solder from the motherboard after removing the graphics chip I usually find a brown layer of that dust that got under the solder all the way to the soldering pads and that what causes the bad connection and that is why reflow usually doesn't work because that layer of the very fine dust will stay under the solder as it is and that what makes reballing necessary to really fix the bad connection on the graphics chip side, sometimes I feel like the CPU should be also removed and reballed and I am getting ready to start doing that specially in cases like the one I show pictures for.
If it was not a manufactruring fault it should be soldered correctly?
Think we can agree a "correct" solder joint has an intermetallic bond leaving no space to get anything under it? Even dust granius?
Still sitck with my previous thoughts on this: bad soldering process/profile..
D3us wrote:
Imo, the problem is the lead free solder. Or at least the used soldering process/profile.
Lead free solder doesn't wet (solder) as well as lead solder, needs higher preheat and soldering temps.
This temp has to be reached a specific soldering profile.
Higher temps = more chance for oxidation = more chance for errors, which can arise later, etc...
That AMD GPU has over 900 solder balls on it.
Much chance for some going wrong.
Still stand by what I wrote before:
It's probably not an internal GPU fault.
The reason is bad soldering of the BGA most likely.
Not all balls got fully liquidus or long enough TAL, not giving a 100% soldered connection.
It makes contact but is not really soldered, doesn't have a real intermetallic bond.
More "glued" instead of soldered.
Same bad soldering process can also result in "brittle" solder.
The mechanical stress caused by heating/cooling cycles, making it expand and shrink, breaks the "glued" or brittle connections.
Other possibilties are "head in pillow", can break later due mecanical stress too.
Tin wiskers, as mentioned, but less likely in this case.
And others...