SiC MOSFETs are often used in parallel in power modules to increase current capacity, but due to mismatch of circuit parasitic inductances and chip parameters, there is a serious current imbalance between paralleled chips during dynamic processes. Thus, this article proposes a chip screening method based on spectral clustering algorithm which comprehensively considers the differences in parasitic parameters and the differences in chip parameters, aiming to optimize the transient current sharing through the mutual compensation of these two impacts. First, a Spice model of the used SiC MOSFETs is established based on the measured data, and the parasitic parameters of the layout is extracted by Ansys Q3D. After which, using the joint simulation of MATLAB and LTSpice, a data set reflecting the relationship among chip parameters, parasitic inductances and dynamic current sharing can be obtained. Then, the principle and procedures of the chip screening method based on spectral clustering are presented and derived in detail. Finally, the experimental results validate the effectiveness of the proposed method.
