High frequency converters based on silicon carbide (SiC) semiconductors are becoming popular, but due to the integration parameters, they are very likely to generate high voltage overshoot and large oscillation, which increase the voltage stress and cause EMC problems. To suppress the voltage overshoot and oscillation, the mechanism is firstly analyzed. Stray inductance is a culprit in causing these problems. To minimize the inductance of power circuit, this paper proposes optimization method including better design of bus-bar structure and parallel connection of snubber capacitors by means of model analysis and formula derivation. To evaluate the method, an inverter model and a prototype were built, several optimization schemes were compared by co-simulation and experimental tests. Experimental results show that the stray inductance of the new bus-bar is 6.4% of the traditional bus-bar. The loop-inductance can be reduced by 46.4% by connecting the capacitors in parallel, and the switching losses can be reduced by 30.8%. This analysis provides guidelines for a full SiC inverter design.
7 April 2026 in IEEE Transactions on Power Electronics
In high-voltage high-power rail transit auxiliary power supply (APS) systems, ensuring zero-voltage switching (ZVS) under discontinuous conduction mode (DCM) is critical for LLC resonant converters (LLC RCs). However, the temperature sensitivity of the turn-off transient in high-voltage field-stop insulated gate bipolar transistors (FS IGBTs) is often neglected, posing a risk of ZVS failure under thermal stress. In response to this challenge, this article proposes a temperature dependent analytical model and an optimal design method. Specifically, an analytical model is derived to capture the characteristic two-stage dVCE/dt of FS IGBTs during the near zero-current turn-off (near-ZCT) transient. Based on this model, a parameter design methodology for LLC RCs is developed, incorporating constraints on junction temperature and turn-off current. Unlike conventional approaches, the proposed method optimizes dead-time selection to ensure ZVS and DCM operation across the full operating temperature range. The approach is validated on a 150-kW prototype, achieving a peak efficiency of 98.97% at full load.