Issue link: https://iconnect007.uberflip.com/i/1472851
16 DESIGN007 MAGAZINE I JULY 2022 structure will behave in production, you need to add another process around the solver itself, so you can predict a real-world distribution of design behavior. at's tricky with a 2D solver, where the simulations run fast, and really tough when you start using 3D solvers, because the run times are much longer and the number of experiments you can run is limited. We think an important (but under-discussed) issue with field solvers is accessibility to every- day users. Field solvers have traditionally been expert tools and there aren't enough SI experts to go around. e "expert crunch" has been a problem for years, and it isn't getting better. We've reached the point where even main- stream designs require the use of field solvers, and our SI experts only have time to spend on the most demanding designs. at means most designs won't get expert SI attention, so what are the rest of us supposed to do? Shaughnessy: Signal integrity skills are sort of like a pyramid, aren't they? You have the experts at the top, a small number of people who are full-time SI engineers. ose are the people using field solvers today. You've got a bigger group in the middle who are skilled system designers, who'd like to be able to run analyses themselves but can't, because they don't have the right tools and processes. Finally, you've got the largest group at the bot- tom, who don't have access to simulation tools or enough understanding of SI principles to know how to apply them. Westerhoff: at's especially true for 3D field solvers. ere are lots of technical details associated with setting up and running a 3D simulation that produces meaningful results. e key word here is "meaningful." 3D solv- ers are complicated and it can be tough to get a simulation to just run in the first place. We can fall prey to the temptation to assume every result is valid simply because the solver itself is "known good." at's not true; correct simula- tion setup is critical to producing a good result and setting up 3D simulations is an expert user skill. Here's a recent example of real-world "gar- bage in, garbage out." A customer was com- paring our 3D solver to one they considered "known good." e two simulators weren't producing similar results, and the user assumed the problem had to be the HyperLynx 3D sim- ulator. ey contacted our AE and asked what was wrong with HyperLynx. Our AE realized