Issue link: https://iconnect007.uberflip.com/i/1114420
12 DESIGN007 MAGAZINE I MAY 2019 ing supply chain. They may be different than the physical layout rules to pin escape. All de- sign rules should be added as early on in the layout cycle as possible. Then, it's correct by construction. Shaughnessy: Scott, can you take us through your process? Scott McCurdy: I ap- proach the design in- dustry coming from my background of running a PCB fabrication com- pany, where half the things that came to us had to be put on hold because something was wrong in the design. Sometim es, there was just no adult supervision at the other end. At Freedom CAD, we are a PCB design service bureau and create hundreds of designs each year, so I'm also including Jay Carbone, one of our top designers who lives for this stuff and is one of our better "interactors" with engi - neers, in this conversation. We deal with a lot of different personalities and skill levels. I also brought Rich Kluever, who was at Solectron for many years at the CM level running Valor, identifying design problems at the factory that was going to have to assemble this stuff. Rich has tried to bake into our process and people more of a manufacturing perspective because every one of our designs goes through Valor twice. Jay Carbone: I can identify with the "design- er's triangle" concept because that's a constant battle. A schematic is a perfect world; as soon as we put a piece of etch down on a board, we've compromised that perfect circuit. With RF circuitry, assembly rules go out the win- dow. Sometimes, RF engineers want to pig- gyback caps or share component pads. Vari- ous technologies dictate different sets of rules; there's no one-size-fits-all solution. We start with a kickoff meeting with the electrical engineer to go over the circuitry and technology. From there we start putting in the rules. We like to identify potential signal integ- rity issues up front. As far as spacing and line widths, this might be a little bit further down the road as you handle the placement and be- come more familiar with the components, etc. For example, maybe they can get away with using five or four mil traces for ease of manu- facturing, or maybe they're down to a 3.5 for impedance reasons. Creeden: You can only put so many rules in today's modern CAD tools. The eyeball check that designers do so often is understated and undervalued. The other thing I wanted to ad- dress is an idea for a universal output. Scott and I had this discussion earlier. We have a universal output, such as ODB++ or IPC-2581, which are intelligent data formats for output. As an industry, we do not have a universal in- telligent data input. I've talked to a couple of the CAD manufacturers, and they're all think- ing about it, but they want to make it their own proprietary blend. Instead, I would en- courage them to have a universal input that any fabricator could use; then, it could be used by whatever CAD tool the designer uses. Nolan Johnson: Scott, I was wondering if you'd be willing to drill down on that for a little bit. A universal input is a great idea. What do you envision that looking like? McCurdy: At some point, it needs to be a data file, and I envision it being the same thing as an ODB++ or IPC-2581 where it's universally portable into any CAD tool. It would show the manufacturer's process constraints and capabilities matrix as well as the stackup in the usage, material, and copper weights. All of these things exist in little islands of auto - mation, but I would like to see a fabricator create this input where any CAD tool could read it. Johnson: What do you see as a mechanism for bringing better control over design rules down to the designers who work toward the bottom half of the bell curve? Scott McCurdy