Issue link: https://iconnect007.uberflip.com/i/1233021
16 DESIGN007 MAGAZINE I APRIL 2020 kle hundreds of decaps on a design, whether it needs them or not. Young: I've seen people do that. I am typically aggressive with decoupling caps because most of my designs end up being in environments where the idea of an interference or signal in- tegrity issue will be very upsetting to the cus- tomer. I don't have an excuse because many of my customers require that my designs do not interfere with other equipment. My decision of when and how to use decoupling capacitors is requirements-driven. It may sound cruel, but often, these design habits are a lack of disci- pline in the process. Shaughnessy: Since you look at the whole pro- cess, and you're talking about design through assembly, where are some of the other spots outside of PCB design that you think design- ers can get more profitability out of the board? Young: Let me walk you through what I would typically do for a project. First, I ask, "What are we trying to do? What are the scope and purpose? Is this commercial, industrial, avion- ics, military, or government?" Those are five industries, large entity industries that have dif- ferent economic factors that affect them. Then, I think, "What's the purpose? Are we trying to design something new or work on product de- velopment? Are we working on sustaining, or are we trying to redesign, fix something in the field, or make something more profitable? Are we doing something for test or research?" A lot of times, research gets left off. Some companies still do it, but I've seen companies blow their entire budget on research and nev- er get out of the door with products. Once you get those factors put into place, then you con- sider your timeline. When do I need to do this? Do I need this now, or do I have time? Is it one week, one month, or one year? The design process that's going to be followed will some- times be defined by the timeline. Is it informal or requirements driven? This means are you going to put something together and hope for the best or follow a defined process. I've developed projects in the avionics indus- try, and some people in the commercial realm might think that because you have a year and a half to two years that you have all the time in the world. But the reality is that it can be an extremely tight schedule. All the in-process testing and the certification testing that needs to take place can cost as much or more than the design effort. People forget this sometimes, and that's how they blow the budget. As an ex- ample, they get to the point where they have a $10 million budget that ignores/underesti- mates the testing involved in the project. In the end, they realize, "We need $10 million ad- ditional worth of test equipment and develop- ment." I've observed that happen on that mag- nitude. Then, they talk about budget overruns where the project is behind. It was supposed to take a year and a half, but it took three years. What happened? They ignored part of what needs to happen and did not talk about design for test and design for manufacturability. You need to design a board so that it's manufacturable. Attention to industry standards is important. In some industries, there might be standards around design for test, but I think it still needs more work to be solidified into more of a stan- dards approach. Another point to consider is paying a premi- um for expediting PCB fabrication and board assembly. A quantity of 20 boards that cost $50 a piece normally for a two-week turn could turn into $150 a board with a 2–3-day turn. That puts you in the area of $3,000 versus $1,000 for PCBs. The assembly of a board with 100 parts (20 ICs, 80 passives) may normal- ly cost around $700–800 with standard assem- bly times (3–4 weeks) could turn into $3,000 for 2–3-day rush. These costs are real because you're paying for overtime, getting ahead in My decision of when and how to use decoupling capacitors is requirements-driven.