A Graph-Based Methodology for the Programmable Construction of ZVT Cells

Zero voltage transition (ZVT) is a soft switching technology renowned for its superior performance. While the judicious selection of topology, parameters, and control modes can notably enhance the efficiency of power converters, the operational performance of diverse ZVT cells exhibits significant variances. This paper proposes a programmable graph-based method for automated ZVT cells construction, capable of generating all feasible topologies with specified component types and counts. The proposed approach eliminates the conventional reliance on researchers' intuition and experience, addressing the inherent uncertainty and incompleteness in manual ZVT topology discovery. The proposed method identifies 18 distinct ZVT cells (including 13 new configurations) for a component set comprising one switch, two diodes, one capacitor, and one inductor, with a total computation time of 396s. Furthermore, among these 18 configurations, a 200 W boost converter prototype incorporating an optimized ZVT cell (operating at 100 kHz) was experimentally validated. It achieves a peak efficiency of 96.7%, which is 1.8% improvement over the non-ZVT configuration, thereby demonstrating the efficacy of the proposed methodology.