BGAs are easier than you think.
Oh, BGAs
are easy. They have relatively large pitch (they can afford it, because pin number scales quadratic to linear dimension) so alignment is not that critical. Balls self-align well during reflow. And most importantly, thinking of side view, the empty space around the balls regulate the amount of excess solder, so stencil apertures and printing process are not that critical.
QFN is opposite. It feels like it's somehow "easier" with pins on the edges. But it still doesn't have legs. It shares the limitation of legged parts: pin number increasing only linearly to linear dimension, so higher pitch is needed to enable miniature size, 0.5mm is very typical and even smaller used. And without balls, there is nothing limiting insertion depth physically: when you press it into paste, excess paste squeezes everywhere. Thus balanced amount of solder and insertion in perfect level is important, otherwise some pads do not flow at all (too little solder, and the solder ends up in a small isolated ball under the component).
It was kinda interesting to see how much problems I had with some trivially simple and low-cost 20-pin QFN, yet at the same time soldered hundreds of expensive specialty image sensors - chip scale packaged so "no packaging", and a lot of pins, all of which important for operation. Something which many would be afraid of, but 99% success rate even with a crappy amateurish process. But they had
balls!
The historical reason for avoiding BGA as hobbyist or small player is that you usually need at least 4-layer board with small vias and small trace sizes to fan-out them. Those used to cost arm and leg, but for past 10-15 years this hasn't been a problem anymore. But there is nothing
difficult either in design phase or prototype manufacturing either. Now if you had to manufacture your own PCBs then things would be different.
Home
Help
Search
Login
Register
