Issue link: https://iconnect007.uberflip.com/i/1114420
16 DESIGN007 MAGAZINE I MAY 2019 such a way that it started to flag areas that you wanted to pay attention to. Rather than just straight-up manufacturing violations, which is how everybody seems to look at how they de- fine the rules, you have a philosophically dif- ferent approach, and Happy is touching on that too. When you change to that sort of a thinking model, how did that change using the tools and results? Kluever: It requires you to be a lot more inter- active with the tool in making decisions and not letting the tool make the decisions for you. I'll give you a good example with solder mask clearance. Some people want the solder mask to be designed exactly as they want the end product to be, whether it's two- or three- mil clearance and some mask defined pads. There are folks that roll up all of their solder mask at one-to-one and let the fabricator modify it ac- cordingly. In our world, how do I set a rule for that? I have multiple rules. I set a rule very high so that all of the solder mask fails. Then, I let the tool do the math for me and segregate the re- sults. Now, I can look at it, and say, "All of the solder mask has two-mil clearance." If 90% of it is two-mil clearance, but six parts are one-to- one, I don't like that. So, I focus on those six parts and adjust them to be two-mil clearance so that everything is the same. Creeden: I think what Happy was alluding to was intelligent input for rules. When do you use your rules? When do you apply them? When do you check them? If you're doing them at the end of the design, you're going to find errors, but those are very costly. When you appropriate correct-by-construction meth- odologies supported by your toolset, whatever that may be, you're correct by construction. Matties: Are there any housekeeping rules that we should have for managing design rules? In other words, how do we keep them organized, and what should people think about? Creeden: To me, they're relative to the circuit and fabricator or manufacturing supply chain you're using. Essentially, rules don't stand alone by themselves because there are 1,000 rules out there. I want ones that apply specifi- cally to the circuit I'm doing, and I'm going to implement it to achieve the performance of the circuit as well as how I'm going to build it with reliability and have the circuit perform the way I want it to. Matties: Do you create a new set of rules for ev- ery design that you're starting? Creeden: Absolutely. They're similar, but every circuit is its own entity for rules. I may use a rule set that I used someplace else, but rules have to be assigned to parts, nets, areas, fea- tures, layers etc. I never want to assume that if a rule worked on a previous circuit, it will also work here. Matties: That's great advice because, for a manufacturer, you may fall into, "I've used these rules in the past with this manufactur- ing and it's going to apply here again." Does that happen? Creeden: Yes, because factories do have a ca- pability matrix, and they may be the same, but how they apply them on a four-layer board is different than how they would on a 12-layer board. You have to drill down, and ask, "How does this board work in this factory?" Carbone: Multiple plat- ing cycles or a different stack-up could change the line widths or im- pedance of a proven circuit. There are many different factors for changing stack-ups or moving a circuit to dif - ferent layers of the card. Kluever: You also have to consider the condi- tions of your design and what you're trying to do. I've had cases where people apply a rule, such as a solder mask oversize on a BGA, and say, "Yes, but we've always used this and it Jay Carbone