Issue link: https://iconnect007.uberflip.com/i/241303
feature New Year's Resolution: Validate Your Model Data by Todd Westerhoff SiSoft SUMMARY: As we begin 2014, remember: Signal integrity simulations are only as good as the models that go into them. But how thoroughly do we really validate simulation models before we use them? Sure, you'll have plenty to worry about in the upcoming year. But here's an idea to consider that might save you some time and money in 2014: Do you test and correlate simulation models, or do you just take someone else's word for it? When we do take someone else's word for it, is there written evidence of that testing and correlation, and do we take the time to review 10 The PCB Design Magazine • January 2014 it? Let's be honest—we all want fast, accurate simulation models with plug-and-play simplicity. Testing simulation models is difficult, timeconsuming and a general pain in the neck. We know that if we really take a close look at most of the models we use, we'll likely raise questions that take time to resolve—time that most of us don't have in our project schedules. There are two types of models involved in SI simulation: interconnect and semiconductor models. Interconnect models include transmission lines, vias, connectors and passive components. They may be represented in the simulation as individual elements or they may be combined together and provided as S-parameter data. Semiconductor SI models can be a type of SPICE (there are multiple variants), IBIS, IBISAMI, or a model that is designed to run in a par-