A while back I asked for advice on BGA soldering. After a serious amount of screwing around in DipTrace I finally got something to work and I am quite proud of this, even though it was maybe easier than I first expected. This is an ATMega48A in a 32 ball BGA. It's very, very small. 4x4mm total. It has a ball pitch of 0.5mm. And you can hand-solder it. All I did was flux the heck out of it and melt it on my board using a SparkFun hot air station gun. Very unprofessional, but it worked. First try, the programmer saw it right away, but after testing, pins PD5 and PD7 were joined (the board would jump from 10mA to 50mA when one was low and the other high). To remedy that, I just fluxed it again and re-flowed the chip and it took care of it. Now all the pins I have assigned seem to be fine. My testing methodology is simple, I make them all outputs, set them all low except for one, test all pins, and then repeat for all pins. So with all low except for one, that will certainly show if there is a join on any of these pins and they are all good. The programming pins are all good. The VCC and ground pins must be good. The 9 pins that I could not break out at all, who knows, but it's not causing problems. Because this was a 2 layer board and because of design rules and minimum via and trace sizes, I had to abandon 9 pins. On a 4-layer board I am sure I could have routed them. Anyway, if I can do this, you can do this. As long as you have a hot-air gun.
Could you describe the process in a bit more detail? How did you put the solder paste on? Were you able to do the chip separately from the other smd parts?
bill2009:
Could you describe the process in a bit more detail? How did you put the solder paste on? Were you able to do the chip separately from the other smd parts?
Sure. First of all, I didn't use paste. My understanding is that you don't with BGA parts. The little balls on the part that are in the images I uploaded are solder. So all I did was flux the board (see board for footprint), flux the bottom of the part, made sure ball 1 was in the right corner, placed it on the board, held it down lightly with the point of my tweezers, and put my hot air gun (this tool: https://www.sparkfun.com/products/10706) on the part, rapidly moving it back and forth so as to heated evenly. I had the gun at 400C. This is hotter than what you use in reflow ovens but it has been working great for me so far with all surface mount parts. You don't need to apply this amount of heat for long. Anyway, when my intuition told me the part was probably nice and melty (very scientific!), I pointed the gun away and nudge the part very, very slightly. The part then returned to its pads based on surface tension. This tells you with most surface mount parts that you got it right. Anyway, like I said in the original post, it wasn't quite right and two pins were joined. So second time I just fluxed the top of the chip to help spread the heat and hit it again, again nudging it a bit. This time, all pins behave correctly. For all the other surface mount parts on this board what I did was flux the pads, apply a bit of solder to the pads from fine diameter solder (Kester), flux it again, hit it with the hot air gun until it melts, and then apply the part with tweezers. Then I usually hit it again and nudge it to see how it behaves, usually it lines up very well with the pads. The through-hole stuff I did in the normal fashion.