Pizza Sponsor: (https://www.cornucopiatechnicalsales.com/) Venue Sponsor: (https://www.apextestlabs.com/) Kristof P. von Czarnowski Speaker Bio: Kristof P. von Czarnowski is an RF and hardware engineer specializing in high-frequency and mixed-signal system design for automotive applications. At Lear Corporation since 2022, he leads transceiver design, implementation, and validation, while also contributing to low- and high-voltage hardware platforms with a focus on signal integrity, power integrity, and electromagnetic susceptibility and emissions. He serves as an electromagnetic compatibility (EMC) simulation subject matter expert, working to shift EMC from reactive post-test mitigation to proactive design integration. His work aims to embed EMC constraints directly into the development process through design guidelines, automated design rule checks, simulation workflows, and libraries - pushing for earlier issue detection, reducing validation risk, and improving first-pass success rates across schematic design, PCB layout, and compliance validation. Kristof holds undergraduate and graduate degrees in engineering from Oakland University, where he specialized in electromagnetics and wireless systems. His technical interests include practical EMI/EMC simulation workflows, SI/PI co-design, and model validation for RF and power distribution networks. His recent work explores AI-assisted engineering tools and machine learning approaches to automate component characterization and accelerate system-level EMC prediction - ad dressing the industry's missing-model bottleneck that prevents full system-level compliance prediction. Abstract Every electromagnetic compatibility problem begins with Maxwell's equations. Yet the path from fundamental physics to actionable EMC prediction remains opaque to many practicing engineers. This presentation aims to demystify this journey. We start at the foundation: how Maxwell's equations in differential and integral form lead to fundamentally different numerical solver families - volume-based methods (FEM, FDTD) versus surface-based methods (MoM, PEEC). The choice is not arbitrary; it is dictated by your problem's physics, materials, and electrical size. Through practical examples, we demonstrate solver selection for automotive EMC: when to use full-wave 3D versus quasi-static extraction, how to bridge the MCAD/ECAD domain gap, and why near-field antenna geometry matters. A detailed biconical antenna walkthrough demonstrates the complete workflow - from geometry definition exploiting axial and planar symmetries, through material assignment strategies (PEC versus realistic copper), radiation boundary setup for CISPR 25 test distances, port excitation, and convergence. But an antenna model alone does not predict EMI. The critical step is integration: combining the biconical antenna with a device harness in a unified simulation domain to capture mutual coupling, then extracting S-parameters for the complete antenna-harness-LISN system. We demonstrate the EM-to-SPICE handoff - Touchstone curve-fitting with passivity and causality enforcement, integration with nonlinear sources, and time-domain-to-frequency-domain conversion for emission prediction. The practical bottleneck is not Maxwell's equations - the physics is solid. The challenge is missing component models when modeling your DUT (e.g., PCB with SMD components). Vendor libraries are incomplete or physically inconsistent (non-passive, non-causal), and measuring every passive on a real BOM is not scalable. We conclude with a physics-informed machine learning approach that synthesizes broadband capacitor models from part descriptions alone, achieving accuracy sufficient for design comparison without waiting for vendor data. Simulation does not replace measurement. But it enables the critical capability every EMC engineer needs: the ability to rank design alternatives before building hardware. Layout variant A versus B. Filter topology trade-offs. Shielding effectiveness. You don't need absolute dBμV accuracy to pick the better design - you need the ranking to be correct. And when combined with analytical design rule checking and validated component models, computational electromagnetics delivers that predictive capability early in the design cycle, where changes are inexpensive and design freedom is highest. Agenda: 5:30 Pizza and networking 6:00 Presentation 7:30 End Apex Test Labs, 815 N Opdyke Road, Auburn Hills, Michigan, United States, 48326
