Marketing is quite often considered to be the nemesis of engineering. I can certainly understand why, as I've been on both sides of that fence. Still, I relate to both.
Today, my day job involves marketing for a prototype assembly house: Screaming Circuits. In past jobs, I've written database software, assembly language driver software for old 8-bit CPUs, and some applications software.
I have less professional hardware experience. Most of my hardware designs have involved my after-hours pursuits of robots for world domination, for which I design about a half-dozen new boards per year.
I did design just a few hardware things professionally back in the day. One device was a highly sensitive, low-noise audio amp designed to help locate termites and other wood-chewing insects. You could actually hear them eating the wood and rustling around inside a wall.
Given my experiences, I can reasonably understand a few things that are important to engineers. However, it would be quite presumptuous of me to believe that I understand what any specific person needs.
That leads me to my basic philosophy of marketing, which is what makes me feel like it's reasonable for me to write stuff for an engineering publication. So here it is: Good marketing is listening and translating. Bad marketing is manipulation.
The difference may seem subtle, but think about it this way; if you speak French and I speak English, you won't understand what I'm saying. Bad marketing would be me moving in closer and speaking louder. Good marketing would be me carefully translating facts into French so you can understand what I'm saying.
Good marketing is translating what customers say so people in the company can understand and then translating what the company has to offer so a customer can make an informed decision based on facts.
I used to say that I'd gone over to the dark side, but now I think of myself as more of a double agent that is hopefully doing some good inside the world of marketing. I don't sit in a vast sea of cubicles as a tiny, nameless, faceless cog in some titanic marketing machine. It's just me and a few other talented folks. What I really like doing is sharing some of the things that frequently whack people upside the head during the prototyping process.
For example, one of the most common trip-ups we see is the dastardly "via in pad." When layout space is tight, it may seem logical to put vias smack dab in the middle of SMT pads. After all, that's how through-hole components work, right?
What happens if you pit vias in pads without capping them is that capillary action will wick the solder down the via. At best, you'll get voids and a poor mechanical connection. At worst, you'll have no electrical or mechanical connection and solder blobbed up on the back side of your board.
For large pads, you can sometimes put vias in the pad, provided you cap them off with solder mask. "A" in the image above is the best option -- a soldermask cap 100-125 microns bigger than the via. But not all board houses can do that.
"B" shows that a bigger soldermask cap can work if you have a lot of pad area, but keep in mind that you'll get some big voids and reduce heat transfer. "C" is never an option. No. No. No. Option "D" can work also, but is the least preferred. You will get voids under the part and sometimes the mask will break, leading to an open via.
For BGAs, the only option, other than to not put a via in the pad, is to have the pads filled at the board fab house, with copper or special epoxy, and plated over with copper. I can't emphasize this enough: Nothing goes on a BGA pad except metal. The above image illustrates copper filled, on the left, and epoxy filled, on the right, with blue soldermask between, creating non-soldermask-defined pads.