Modelling and Optimisation of Power Electronics using Fast Switching Devices and Compact Passive Filters

Name: Jeroen Zwysen

Johan Driesen

Introduction / Objective
The demand for and demands on power electronics have increased over the recent years. Simultaneously, research and developments on the active switches and passive filter components used in these converters have increased. This expands the possible design choices and objectives, making the search for optimal solutions more challenging. In this work, a design methodology for passive inductive components is presented. A combination of innovative simulation techniques and optimization strategy is used. Additionally, a methodology for modelling, simulating and validating the switching behavior of new upcoming active switches is developed and demonstrated.

Research Methodology
In the simulation of the inductive components of passive filters, a big challenge lies in the 3D magnetic field modelling, making the simulation complex and often slow. In this work, a reluctance model is made, breaking down the problem in multiple, simple problems, which can be solved independently using various fast techniques. The resulting simulation model is used in a multi-objective optimization problem, based on genetic algorithms. The result of the calculations is a Pareto front of optimal designs.
For the active switches of the converter, an analytical model which only uses datasheet parameters is presented. This can be used to select optimal candidates for a specific converter design. For the next step of the design, a method to generate a SPICE model is presented. This is used in the design of a 3kW converter to compare simulation and measurements and identify parasitics in the converter.

Results & Conclusions
The presented methodology allows for the automatic generation of optimal designs for inductive components, quantifying the impact of design choices, such as material and cooling choices, in the form of objectives such as power loss and volume.
Using both the analytical and SPICE models developed in this work, the impact of the choice of a specific active switch on the design of a converter can be quantified. Even when only datasheet parameters are available. The combination of simulation and measurements can be used to increase the accuracy of the models, while at the same time identifying specific problems in the design of the converter.