Testing the reflow procedure
My plan is to apply generous quantities of my Amtech flux to the board, seat the FPGA on the footprint and then to run a standard SnPb reflow profile. In my opinion there’s plenty of solder in the balls themselves and no need for me to go to the considerable extra expense of a stencil and solder paste just to put even more solder onto the pads. Flux will be key here. Good wetting of the pads will ensure a reliable joint. Too little and I’ll get dry joints. Too much and the whole package can float off across the board as the flux boils underneath the part which can be amusing when it happens to someone else.
Looking at simple economics I have boards that cost about a fiver, FPGAs that cost about 2 quid and a bunch of peripheral parts that together cost about 20 quid. It makes sense to reflow an FPGA on its own as a test. If the process works then I go ahead and do a full board with all parts on board.
Well, it worked. There were no dramas during the process and I think it’s fully reflowed but without some kind of X-ray analysis I’ll never know for sure.
The best I can do is this high resolution photograph using a macro lens and a set of extension tubes. I can see that the balls at all four corners of the package have gone flat at the bottom where they meet the pad. That should mean that all the central balls have done the same but I can never be completely sure. I decided to go ahead and do a full board.
Reflow and testing
I reflowed all the major parts on the top of the board at the same time, from the FPGA and the memory ICs right down to the 0402 termination resistors. It all went fine. Then I needed to add sufficient parts to the board to be able to validate what I’d done with a simple FPGA design. That’s when I realised that I’d made my job a lot harder than it need be.
An 0402 capacitor next to a safety pin
I needed to mount all the 0402 decoupling capacitors on the bottom of the board and I did so using a hot air gun and a pair of tweezers and boy was there much swearing involved. I know there are people out there that claim to be able to hand solder 0402s in the dark using a plumbers iron and a pair of pliers for tweezers. Well that’s not me. It must be an age thing because I just can’t focus properly on the damn things any more. It took a while, and I got most of them down on the board and at least the same number somewhere on my carpet or in my hair.
With the benefit of hindsight I should have reflowed the bottom side in the reflow oven, kapton taped them in place and then done the top side. Much pain would have been avoided. Afterwards I rushed off and bought one of these on ebay.
Revelation! The large working distance, big lens and LED ring light work really well. Now I can see those little 0402s just like they were 0603s. I highly recommend this cheap piece of kit.
Minimal assembly prior to testing
I’ve added just enough components to be able to do a test, so it’s over to the Xilinx ISE software for some VHDL fun.
