Issue link: https://iconnect007.uberflip.com/i/1472851
14 DESIGN007 MAGAZINE I JULY 2022 for modeling large sections of boards where good, but not ultimate, accuracy is required, so they're oen used for power-aware simu- lations and PDN analysis. Hybrid solvers are sometimes called 2.5D solvers, but that's really a misnomer. Hargin: No matter which field solver you're using, providing an accurate description of the structure to be modeled is essential. ere's no point of doing a detailed 3D simulation if the structure doesn't match what you will build. We're well past the point where we can get away with guesstimating important parts of the design like the stackup, and leaving the details to the manufacturing folks to figure out later. Traditionally, the NPI engineer took direc- tion from SI experts on what trace impedances were required, then worked with the fab house to create a stackup that would meet those tar- gets. Most fab houses used 2D field solvers like the Polar 2D solver to develop the stackup and document the resulting impedances. Key to this process was an accurate library of PCB materials and their electrical properties, and each fab house maintained their own. With Z-planner Enterprise, we've gathered that material data and made it available to the system designer, so they can generate accurate stackups themselves. at means the design simulations they perform are more accurate, because they reflect a real stackup, instead of a hypothetical one. Designers can also use it to share stackups back and forth with their fab house and collaborate on poten- tial changes. Z-planner is integrated with HyperLynx for SI and PI simulation, so designers can use accurate data for their design and verification work. People commonly com- pare results from the Hyper- Lynx 2D field solver to the corresponding ANSYS and Polar tools. ese three provide very similar results and I think that provides a high degree of confidence in their accuracy. If you're using a different 2D field solver, it would be wise to compare it to one of these tools. As tempting as it is to say simulated results should always be verified against measurement, in cases like this, it's better to compare results to known good field solver results, because bench measurements introduce other issues. For instance, when I measure a board, am I really measuring what I thought I was measur- ing? Let's say I'm measuring impedance and I want to compare two field solvers to a bench measurement. How do I know what the dielec- tric constant (Dk) of the laminate used in my bench measurement really is? ere's typically a ±10% variance on Dk values, so if I'm going to correlate a field solver to measurement, I must figure out the exact material properties for the particular board I'm measuring first. In this case, comparing to another known good field solver is a better way to validate how good a field solver is. Westerhoff: Field solvers and simulators assume structures can be manufactured with precise material properties and physical dimensions, which they can't. If you want to predict how a