Issue link: https://iconnect007.uberflip.com/i/1401769
14 PCB007 MAGAZINE I AUGUST 2021 what they're talking about." And to your point, there's no such thing as Chinese in China— there are two basic languages, and then there are 21+ dialects on top of that. Thompson: Case in point: app notes. I'm sure you've seen some crazy app notes. Some things that have been translated three or four differ- ent times and you read them and you're like, "at doesn't make any sense at all. Even the instruction doesn't make any sense." Holden: And IPC-2581 is standardized so that we wouldn't have all these extra spreadsheets or Word documents. We could just refer to some- thing in IPC-2581 that de- fines it for us. Korf: We actually have a sub-team part of the 2581 consortium looking at fab- rication notes now. Which ones can we make intelli- gent? Or, what addition- al attributes do we need to add to get rid of these non- intelligent notes? Because if you look at a fabrication print, to Mark's point ear- lier, the board outline a lot of times doesn't match the board outline in the Gerber file. And so you send a TQ back to the customer asking which one they want us to use. e extracted dimension doesn't match the fabrication print dimension. ey say, "You use the one you extracted from the Gerber file." It's not uncommon to have issues like this. To Andy's question earlier, my goal is to get rid of the drawing as an intelligent method of transferring the data and treat it as a report. Instead, send the information using attributes wherever it is in the design file. e drawing could be extracted from the data. ere's so- ware out there that can create a drawing if they want to have a picture or hard copy. So, why can't we send specs electronically too? ere's no reason they must be retyped again. Thompson: One of my favorite things that cus- tomers would say, they'd have a Class 3 job, and they'd say, "Well, go ahead and generate a net-list. At what point would I ever have a mis- match?" I don't think we really want to do that. Shaughnessy: But it's interesting to talk about having too much data because one of the things we heard when we were doing our fab notes issue was that a lot of times the CAM o p e r a t o r d o e s n ' t g e t enough data. So is it feast or famine? Korf: And/or it's conflict- ing. Thompson: at's where the fab engineer comes into play. at's the per- son that must call and say, "Hey, what did you mean by this? What did you real- ly mean by that note?" Shaughnessy: Right. One thing that came up in our survey is there's a disagreement among people as to whether DFM is a philosophy or a pro- cess. For some companies, DFM has become a philosophy. Korf: I think it's a little bit of both. My opinion is, from a philosophical standpoint, everyone's assuming that the person receiving the data is going to run a DFM to check it. e whole in- dustry is based on that premise. So, in that re- spect, it's a philosophy. If the philosophy was that they're going to build it the exact same way I send it to them, you don't need DFM. But the process of performing a DFM is defi- nitely a process. To me, it's both. Happy Holden