Comparison of Control Signal Oversampling Techniques for Real-Time Capable Simulation Models

Power electronics simulation in a Hardware-in-the-Loop (HIL) setup represents a challenging task because of the fast-switching modern semiconductor devices causing rapid structural changes of the electrical circuit. The demand for compact and lightweight parts, particularly in the automotive industry, results in extremely high switching frequencies, reaching up to the megahertz range. For real-time-capable simulation, models with a simulation step size in the nanosecond range are required. Achieving the smallest possible step sizes alone does not guarantee the best simulation results. If gate signals are measured once in the model simulation step, inaccurate duty cycle detection can lead to aliasing oscillations. Oversampling of the PWM control signals is a well-established approach of minimizing such effects. Further processing of the on-ratio can be performed in various ways. In this article, Multi-Model Averaging (MMA) is compared to Sub-cycle Averaging (SCA) and Four-period integration oversampling methods (4PIOM). Results of the closed-loop real-time simulation of a DC-DC converter operated with 1 MHz switching frequency as well as the consumption of needed FPGA resources for the realization will complete the paper.