Multi-resolution modeling of power converter using waveform reconstruction

Computer simulation of switching power converters is complicated by the discontinuous (switching) nature of the converter waveforms. When switching details of the waveform are of interest, detailed simulations requiring extremely small time steps are needed. On the other hand, so-called averaged models allow capture of low-frequency converter dynamics (of interest for example when closing the feedback loop) by computing averaged continuous waveforms. The possibility of using relatively large time steps makes averaged models computationally efficient. A real-time hierarchical approach that combines the advantages of the two methods is proposed. An averaged model and a detailed model are used alternately in successive time intervals. The final state of one model at the end of a time interval is used to calculate the initial conditions for the other model running in the following time interval. The method is illustrated with a buck converter. The combined use of two models with a hierarchical or multilevel approach is shown to provide a powerful simulation tool for analysis and design. The detailed model is used 10% of the time and the averaged model 90% of the time. Results indicate that the detailed behavior is accurately simulated when the detailed model runs and the overall simulation time cost is lowered by running the averaged model 90% of the time. This approach provides a way to simulate the detailed behavior of a specific part in a rather large switching circuit network.