Atmel’s 128RFA1 and the joy that is QFN
My latest undertaking, and the inaugural project of my new home workshop, is built upon Atmel’s Atmega128RFA1 chip. Being a Microchip follower for the last decade, I was hesitant at first to jump into the Atmel pool. After a week or so with their platform (AVR Studio running on Windows 7 via parallels on a Macbook Pro, GCC a-la WinAVR, and the JTAG ICE MKii) I feel right at home. I can also say that, so far, the 128RFA1 is nothing short of awesome – a great MCU with a 802.15.4 transceiver in a single package. Unfortunately, that package happens to be of the QFN (quad flat no lead) variety making it particularly unsuitable for prototyping. In this post, I’m going to walk through my solution for soldering these little buggers.
I’ve soldered every variety of fine pitch SMT without any issues in the past. If I can get an iron to it, it can be soldered…. and therein lies the problem – there is NO WAY to make contact with QFN pads unless you have an iron tip the size of a sewing needle. I’ve read and watched a few tutorials on people who are doing SMT soldering using stencils and hot plates, but I honestly can’t see that turning out that well. I also don’t want to trash a few boards and chips during the learning curve – especially with the peculiar shortage of 128RFA1’s for the last month or so. So here is my solution:
Step 1. Get a microscope….. preferable a stereoscope. I have the rare fortune of having inherited a pristine Olympus seteroscope, but I think that any scope you could grab on ebay would suffice – you just need to get in close.
Step 2. Get a heat gun. I picked up this Atten rework station on eBay for something like $150. It arrived with a broken PCB like most quality Chinese products and took me an hour or so to fix. In addition, I’m using a trusty Weller iron for normal through hole and SMT such as SOIC, QFP, 805, 603, etc… You can see the JTAG ICE MKii on the right as well.
Step 3: Get some tools. Flux is critical, and Kester makes a nice highlighter style dispenser. I ended up using Chipquik to solder the QFN in place and applying it with a hypodermic needle. Less is more!
Step 4: Line up the QFN on your footprint. Even though you can’t see any of the pads when looking in the microscope, it’s actually not that hard. Just align the chip to have a more or less even “border” of landing pad on all 4 sides. This guarantees that it will be aligned properly. Apply just a bit of the Chipquik (which consists of a million tiny solder balls suspended in some sort of rosin) on one side of the chip. Apply heat from the air gun, keeping the flow on the low side. You’ll know when it is working, as the tiny solder balls start collecting and pooling. A telling transition to look for is when the landing pads or any nearby vias turn to liquid. Remove heat. At this point, the chip should be attached by inspection.
Step 5: Repeat on the other 3 sides. Once you do this, you can angle the board under the microscope to inspect for shorts and opens. I’ve had no issues with shorting so far….. but I use only the smallest amount of Chipquik. You should be able to count the individual solder balls…. on the order of 100. I have had some issues with opens, but this is preferable in my mind. Just add a bit more Chipquik and heat that section.
Once your satisfied, wipe down the PCB with alcohol or flux remover to get rid of any excess Chipquik. Rig up the JTAG or ISP connection and make sure your can read a signature. I’ve done two chips to date, both successfully programmed and transmitting wirelessly. I’ll do a tutorial post on that soon!