FBGA, TQFN, and other four letters to know when building circuit boards
Miniaturization of product and devices is the designer’s eternal struggle. “Smaller, faster, cheaper” has been and is the mantra of not only the consumer device world, but design in general. More and more applications demand more features from less space using less energy.
None of this is a surprise. This has been the modus operandi of the industry since the days of the monolith. One complication this introduces, however, is the increased difficulty in quickly assembling prototype equipment utilizing a certain class of integrated circuit packages without the need to turn to manufacturing and assembly in-house.
Those integrated circuit packages – the leadless packages. These packages help to conserve board space for routing, in many cases are the most cost effective physical package, and in some cases are the only option you really have (think high-speed SDRAM, wireless ICs, and feature-rich SoC devices).
Of course, there are always manufacturer development boards, and many times that is sufficient for verifying that “the thing does what it says on the tin” and is suitable for your application. But there are times when it is prudent to do further testing within your custom design before having a third party involved in assembly. Sometimes you need that flexibility that hand assembly gives you during design implementation, or perhaps you can’t deal with an assembly lead time or cost. Sometimes, the bridge between your first wire wrapped mess of a prototype on perfboard and your shiny, polished, DFM’d PCB is an array of prototype PCBs with all manner of trace cuts and wire jumps.
With that in mind, there are a few things you can do with your PCB layout to help deal with small run prototype PCBs involving some small, leadless parts.
First and foremost: you are not hand placing a BGA part. While it is feasible to align and reflow a BGA part with a low pin count (i.e. less than 37 pins), I advise against such a practice unless you’ve got boards, parts, and time to spare. If you are going to make such an attempt, I can only recommend a lot of flux and spare parts. It would be best to have a third party assembler place the difficult components such as the BGA part for you.
To increase success with either an attempted manual hand placement, or with placement through a third party, keep in mind the drill size limitations that your PCB fabricator has and pick the smallest drill size for your pin fanout vias. A short between a BGA pad and a via is a very real possibility, and you don’t need oversized holes making the problem worse. That being said, if it is possible to comfortably fan out all BGA pins, I would recommend it. It will allow you to more easily check for assembly errors due to shorts or cold/non-existent solder joints.
For particularly dense BGA packages, the amount of room you have between pads becomes very small, and in this case you’re likely looking at microvias for your fanout. In those scenarios, hand assembly is really not going to be a valid option as the chances for slight misaligned placement causing a non-functional part become quite high. In a circumstance like this, I would recommend locating a breakout board from the manufacturer and placing that into your design to validate the rest of your board build prior to dealing with the BGA route in the first place.
If you do need to make a prototype board with BGA components, I recommend that for your first pass of a route, start with two more additional routing layers than you think you need. Pick a relatively simple route strategy and use those extra two layers in your BGA area only. Once you’ve got your route complete, see if you can rework the BGA area to eliminate the extra layers.
Working with other leadless packages such as QFN or other package that does not have conductive signal lines underneath the middle of the package is quite a bit easier. For your prototype level boards, you’ll want to oversize your pads for the footprint. Your goal is to make sure that you have enough metal exposed on your pad when you place your part down as to be able to fit a soldering iron tip. Strive to have the leads of the edges of your QFN part land on the middle of your pad, with at least 0.005” exposed. If your part does not have a thermal pad on the bottom for heat sinking purposes, you can give yourself some extra wiggle room by extending your pad another 0.005” towards the center of the part. If there is a thermal pad you should avoid this additional overage, as that increases the chances of an unintended short.
These are just some quick guidelines for doing short run, hand assembled prototypes. Once you start moving closer to production-ready hardware, you’ll want to rework some of your footprints. In the case of a BGA part, you probably have additional pins fanned out that do not need to actually route. Eliminate those, but other than that, you are likely not going to change any pad sizes, trace widths, or hole sizes. In the case of a QFN part, you’ll want to shrink your pad sizes back down to more closely match the actual pins of your part.
One thing to keep in mind: small parts are great, and the routing real estate savings with leadless parts are even better, but sometimes, it is just unnecessary to utilize those components. There may be no actual cost savings; there may be no need to save on the real estate. Evaluate your project/product, your options, and make sure the possible increase in prototype/development complexity is worth it.