Issue link: https://iconnect007.uberflip.com/i/553274
14 The PCB Design Magazine • August 2015 feature about what's going on in the business. It shows that designers are more aware. This is good news because it is showing that de- signers are taking note of what the manufac- turing industry is finding though the process of "failure analysis" utilizing cross-sections, X-rays, ionic contamination and a lot more. The results are not being ignored. Dack: On the front-end, the cause of many common failures can easily be caught by checking. There are so many failures that can be attributed to poor design checking. Dis - continuity between the schematic and netlist in the layout; failure to edit nodes properly; forgetting to change a solder mask color; for- getting to flood a power plane could cause a huge failure. But a lot of these common problems are easily prevented by a check process. Thompson: I definitely agree. I see that peo- ple are making more progress in this respect. A lot of our customers, for instance, are now supplying an IPC netlist when they didn't in the past. This form of communication helps our war on failure by comparing the design connectivity orders with exported, graphi- cal Gerber data orders prior to making any edits whatsoever. So, let's bring in another military metaphor: friendly fire. A lot of peo- ple, when they're brand new at generating netlists, will do silly things like assign fidu- cials as net points or non-plated holes as net points. Some do not realize that the design may have half-plated pads at the part edge that are creating a connection with a metal post at some point later in its life. Having to in- terpret and resolve design dicontinuities with- out a code talker, is like dodging friendly fire. Not resolving a connectivity is- sue, even though it looks perfect can be like stepping on your own team's landmine. Shaughnessy: We also see a lot of designers who are over-constraining their designs and making them too high-tech. Thompson: Absolutely. Many people rope themselves in with these specifications. I've seen so many drawings that will call out very specific material types, very specific dielectric values and even very specific dielectric con - stants. This is really a problem. If you go to shop X, they may have a flavor of 4101/126 that has a different dielectric constant, a dif- ferent environmental condition, and a differ- ent lamination process and parameter that's different from shop X, shop Y, and shop Z, all the way down the road. If you make a draw- ing and you roped yourself into a specific material type or a specific dielectric constant, you're really doing yourself a disservice. Dack: What Mark is talking about is overspec- ifying. Overspecifying can cause failure at the front-end manufacturing process by limiting sources who can manufacture the part. Thompson: Kelly and I did a tour yesterday and we brought up that very point. We were talking about how sometimes people go on a tangent because it's the new thing to do. I hear customers saying, "We have to have ENEPIG because we have to have EN- EPIG. We don't really care about the surface finish. We don't have any constraints or ap-