Issue link: https://iconnect007.uberflip.com/i/1276973
16 DESIGN007 MAGAZINE I AUGUST 2020 rize people that way. "These people want to hear, but these people don't." The ones that don't want to hear issues, you just give them the ones that make it totally unmanufactur- able. Just say, "We have to stop." I call it "law of physics"-type questions. It can't be built unless we resolve the issues. Definitely, in a competitive bidding situation, there are a lot of EMS providers and a lot of people that are buying the boards for assembly that would just be price-based. They don't care. It's just price. If you're the cheapest, you get the order. Companies will learn those customers, too, and tailor their responses to them. There may be a partial high-level DFM being done in the background. We look for any major issues that may cause a yield hit, a rework hit, or a major cost hit. Sometimes, it's done in the quot- ing group, sometimes by another group. Dack: I think it's very valuable to establish the orientation of the documentation. We start every note set with the primary side of the board shown, and we follow the IPC standard declaration of a primary side and a sec- ondary side. The inter- pretation of that is that the primary side is the side that the designer has designated. It got confusing. We all know the story of how the solder side of the board became the solder side, and the com- ponent side became the component side. With complex boards now, it's not as clear. Where we have gone is that the primary side is the side that the designer has designated as such. Korf: That note's real critical for assembly. For fab, we're going to look at a file that says the layer names. What's the file layer name, and how do I match that to the board? The primary side, is it called layer 1? Is it called top? Is it called Jim? I don't really care what it's called. Just tell me what it is. Most fabricators will align boards the way they want to build them. They might not label them the same way as the designer, just to have some standardization in the factory. From a conversation standpoint, it's always good for assembly. For fab, it's not too useful. We just care about matching the data layer names to the actual data provided. Dack: In the context of the design, if you're going to specify Class 2, you'd better be designing Class 2 and not a different class for manufacturing and inspection purposes. What do you think? Korf: That's critical because that guides a lot of default criteria, the IPC class in general. A lot of people will say, "I want Class 2, except for my plated through-holes. My annular ring will be Class 3." One of the consulting projects I've set up to solve someday is this: Every IPC spec has this category called AABUS, or "as agreed between user and supplier." If you look at IPC-6012E, there are 38 sections and tables that have that as a criteria. If everyone technically followed that spec, we would have to have 38 ques- tions every time going back to the customer, saying, "What do you want to do?" No one does that in reality. That's one of the con- flicts in the specs versus reality. No one is going to ask 38 questions every single time just because of the IPC spec, and the committee couldn't agree on an answer. Or, as I call it, some of these AABUS categories should be a tutorial. You should worry about this. I agree that the IPC class note and exceptions to it is critical. Happy Holden: Dana's hitting all of the key points. The big problem I always have is the