Optimization of Sets of Semiconductor Power Modules for Paralleling in High Power Converters

Abstract:
To reach sufficient output power, paralleling of SiC MOSFET modules is mandatory in many high-power applications like for example railway traction converters. First, this work demonstrates the detrimental impact of semiconductor parameter mismatch on the switching behavior of paralleled SiC MOSFETs. For this reason, sets of power modules selected according to specified constraints must be used in parallel operation to assure for example a certain maximum output power derating. To find a maximum number of these sets within a batch of modules, two algorithms based on methods of graph theory are proposed in this work: The first algorithm delivers an exact solution but might need excessive calculation time. The second approach calculates an approximate solution in reasonable time, which is shown to be close to the optimal solution. Shipping inspection data of current 3.3 kV SiC MOSFET devices for railway traction applications is used to compare these approaches to a straightforward reference method. In the studied examples, the proposed methods deliver superior results regarding the number of sets that can be created from a batch of modules, enabling stricter specifications for decreased deratings at the same time.