Issue link: https://iconnect007.uberflip.com/i/1491842
14 DESIGN007 MAGAZINE I FEBRUARY 2023 is shorter than the distance that the signal travels during that switching edge. You don't have significant reflection problems because you are actively driving the signal you are actively putting energy into, or taking energy out of, that transmission line. at active driv- ing will overcome any reflections or crosstalk, rise time, and all that. Well, again at A1 nano- second edge rate and a dielectric constant of 4.0, that gives us about a 6-inch trace and that is fairly long on a circuit board at one-quarter of a nanosecond. You now have to start dealing with termination techniques, crosstalk avoid- ance, and all these technical problems, just because the switching edge got faster because the die has shrunk. Shaughnessy: So, it sounds like the "typical" PCB designer will need to be aware of all these potential pitfalls far before the design cycle starts. What can designers do to get ahead of this? Moyer: Step one: Assume that if you have digital chips on your board, you have signal integrity issues and must apply the proper sig- nal integrity and design analysis methodolo- gies. Based on your stackup, determine your velocity of propagation and your transitional electrical length, and put in proper series ter- mination. is is going to be a collaboration between the board engineer and the circuit engineer. If your company doesn't have a dedicated signal integrity analysis engineer, the electri- cal engineer should be able to do the signal integrity analysis to determine the appropri- ate series termination resistor values. What are the parallelism requirements? Assume you're going to have those up front before you lay your first trace on the board. Have all your design rules created and defined, assuming sig- nal integrity-related issues. Nolan Johnson: As you say, it's not really black magic; it's all right there. Moyer: There's a great textbook about this topic, written by Dr. Howard Johnson, called High-Speed Digital Design: A Handbook of Black Magic. He literally named his book "black magic," because in the early days, this was all not well understood. What do these structures look like? One of the things that our designers really need to understand is that we need to stop thinking about our board struc- tures in two dimensions. We need to start con- sidering them as 3D structures, considering the Z-axis for separation to planes. What about the fact that we have fringing capacitance com- ing off the edges of our traces? ere is a verti- cal edge to our traces; it's not just a flat trace, even though that's how we draw it. You know, in our CAD tools we really need to start con- sidering all the structures on our boards as 3D structures. What are the physics of those 3D structures and how are they going to behave at these high frequencies? Shaughnessy: Who are the great instructors in this area? Who has written books on this? Moyer: ere are four main gurus here as far as authorship is concerned. I mentioned Howard Johnson already. Lee Ritchey of Speeding Edge is an outstanding guru and signal integrity instructor and author. Doug Brooks has writ- You now have to start dealing with termination techniques, crosstalk avoidance, and all these technical problems, just because the switching edge got faster because the die has shrunk.