Issue link: https://iconnect007.uberflip.com/i/1359517
26 DESIGN007 MAGAZINE I APRIL 2021 company we talked to was making small boards for consumer applications, but spending over $150,000 a year on prototypes due to problems that were readily detectable with simulation. ere was a clear argument for saving money, even with a tool purchase and the ramp-up cost of training a new user. 3. Costs and delays associated with out- sourced analysis pile up to a breaking point. If I use someone else to do my SI analysis, that takes time and money—even if the SI consul- tant is internal. If I outsource enough work, it becomes a problem. It doesn't have to be about cost; if lots of people rely on the same SI con- sultant, they're all dependent on what other jobs the consultant has already committed to. Time is money. 4. Reference designs and rules of thumb run out of gas, and/or overdesigning drives up manufacturing costs to a breaking point. Com- modity consumer products will be especially sensitive to something like this. 5. Organizations are proactively seeking to improve their design processes. Shaughnessy: Wait a minute. You're saying that process improvement ranks last in terms of reasons why people start simulating? Westerhoff: I am because I'm being practical. Companies don't make big investments to solve problems that aren't costing them a lot of money. ey always have other things to worry about. Simulation is a lot like having good habits: we all agree we should. We all agree that we should floss, eat more vegetables, and exer- cise more. e question is, how do you lower the barrier to helping someone get started and keeping the process going? Here's a paradox: If most of what we talk about are the state-of-the-art problems in sig- nal integrity, then we're making the barriers to entry greater, not lesser. We're experienc- ing an "expert crunch" already and focusing almost exclusively on technology makes it worse. Johnson: e automotive press is similar: ere's lot of coverage on the Corvette and it sells magazines, but when most readers go to the dealership, they buy a Toyota. Westerhoff: So the question becomes, what does it take to make someone productive with SI and PI tools? I propose that there are three main components. e first is domain expertise. If you don't know what characteristic imped- ance is, I'm not going to show you what to do with a signal integrity tool, because you won't understand the problem to be solved. e sec- ond component is tool training, and that's actu- ally the easy one. e third component goes back to our fundamental question: Will my design work? If I give you a simulation tool and you run a simulation, then what? How are you going to determine if the design passes or fails, and by how much? You need a complete ana- lytical methodology for verifying your design's behavior against a spec to determine operating margin. at methodology will be protocol- specific, and sometimes component-specific. So you need knowledge, you need tools, and you need an analytical methodology that maps to your interface's specific requirements. People who run SI simulations for a living will put the pieces together themselves, but system designers and layout designers need solutions that work right out of the box. at's what we're trying to do. So the question becomes, what does it take to make someone productive with SI and PI tools?