Issue link: https://iconnect007.uberflip.com/i/1449048
44 DESIGN007 MAGAZINE I FEBRUARY 2022 en, I'd make all my library parts, double and triple check to make sure that they were reasonably close to what the data sheet said. at would be in a milestone. at's a huge task sometimes. I'd place the part on this, match them up with the schematic, place it on the board, have the customer look at it, verify that it's going to meet his needs, the connectors are where he wants him, or this connector is not rotated, so that he's going to have to hook the wires up from on the board rather than off the board. All the little things that you must think about in that regard would be more milestones. en, I'd route the thing. Once it's routed, it goes out again to the customer, look at it, make sure that I haven't routed his 50-watt lines into 10 mils, that sort of stuff. en, off we go, off to the races. Holden: I'm using the term milestone, like what Bob talked about. If your design process is a flow chart of five boxes, then every box is a milestone, but if you take each box apart, let's say, placement, and instead of being one box, it's actually 10 boxes of a complex flow chart. If within that, before you decide on placement, you've asked the question, "Do we have to have design for test?" Do they have to have space to get in there and probe? Are they going to use boundary scan in which they need access to vias, but this is going to be an HDI board in which the vias are going to be under the BGAs and not accessible? ere are several questions and alternatives in terms of what is going to be your placement strategy. Does the component have access from all four sides? Does the component require cer- tain reference planes that have to be matched? It must be paired up with other components that use the same reference planes. It can't be on opposite sides of the board. What I'm talk- ing about is probably much more detailed— how we design a board in terms of 30 or 40 steps. Dack: I've never seen a chart breaking the design process down into what's important for manufacturing, assembly, test, and fabrica- tion, at least none that offer alternatives. What are the consequences of signal integrity, cost, or time? What options do we have? It is as if PCB designers are steering through this obstacle course, and they're making the best decisions they can. It would be nice if you had two or three people on each side of you con- stantly giving you advice—one with expertise in fabrication and assembly, and another with expertise in test and field repair. We don't have that kind of a thing. Now, maybe with AI we will someday have those virtual people looking over our shoulders offering that advice. Happy, isn't that what our smart factories are about? Matties: at's generations away. Holden: Yes. But that's what they're promising. It might be 10 or 20 years, though. Dack: I think the smart factory is going to explode once it catches on. It reminds me when flat screen TVs came out or as we pres- ently transition to electric vehicles. I rolled my eyes about electric vehicles years ago, but now I drive an EV. Shaughnessy: When I first started covering this industry, I asked for a list of milestones Once it's routed, it goes out again to the customer, look at it, make sure that I haven't routed his 50-watt lines into 10 mils, that sort of stuff. Then, off we go, off to the races.