Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I worked in the R&D labs of a mobile phone manufacturer recently, and I witnessed the rework lead change BGA chips off of PCBs. She used a special machine to line everything up, turned on the heat, and used a vacuum pump to pull the chip off of the board once the solder melted.

I believe installation is simply the reverse.
 
I worked in the R&D labs of a mobile phone manufacturer recently, and I witnessed the rework lead change BGA chips off of PCBs. She used a special machine to line everything up, turned on the heat, and used a vacuum pump to pull the chip off of the board once the solder melted.

I believe installation is simply the reverse.

Works great for R&D, but for the larger higher pin count chips with fine resolution, the nature of CPUs and GPUs (aka extreme heat on the solder connections and mechanical stresses) tend to limit a lot of this work to R&D labs ... since it could decrease the life of a swap drastically in the field.
 
I believe installation is simply the reverse.

I wish it were that simple. On one hand you have to make sure the contacts are properly heated so that the solder will flow and cure properly without forming cold joints. On the other hand you have to be careful not to damage the chip, board, or other heat sensitive components. Rework stations use computer controlled thermal profiles to apply heat incrementally in specific areas. The "warm down" process is equally as important. Otherwise your chip will only work for awhile before the heat cycles cause micro fractures in the joints. This is basically what a temperature profile looks like: http://www.allegromicro.com/en/Corporate/Environment/images/pbReflowProfile.gif There are 100 different factors that influence how this profile is made. I'm just trying to say that (unfortunately) it's not as easy as placing the chip, then throwing it in an ezbake oven for 2 minutes.

I redact any statement I may have made about this requiring education. It does require some, but what it ultimately requires is experience. I applaud anyones efforts to educate themselves about this or to experiment with it. I don't do rework anymore but I still find myself fascinated just thinking about it. Please just don't go desoldering the cpu off your mini just quite yet though ;)
 
Maybe it's more complex for chips having more than ~60 points of contact (that's all I ever saw the rework lead do) -- but she made it seem easy as 3.14 with the little rework station she had.
 
Maybe it's more complex for chips having more than ~60 points of contact (that's all I ever saw the rework lead do) -- but she made it seem easy as 3.14 with the little rework station she had.

A bit more complex for the bigger chips and when the chips us high pressure heatsinks, the alignment is likely a bit more critical due to the extra mechanical stress on the chips.

Here is a more complex chip procedure with a fine resolution on the pins.

Rework Procedure for NECEL's Flip Chip Ball Grid Array Packages.pdf
 
You are correct. If you have the experience, know the proper temperature profile, and have access to the equipment, it's a piece of cake. The big problem that we hit head on with this isn't really finding someone who can do it. It's the lack of any chip to solder back onto the motherboard. Intel doesn't make a micro-fcbga quad core (or anything terribly interesting in a dual core for that matter). Intel has to make one of these chips before anyone even considers this project.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.