The broadband oscillation in IGBT modules

with numerous types

wide frequency ranges

and complex oscillation mechanisms

causes existing research to be faced with the problems of unclear characterization and mechanism

which makes it difficult to analyze the internal broadband oscillation problems of the module. Therefore

this paper proposes a complete characterization and analysis method for the broadband oscillation of high-power multi-chip power modules

and realizes the online time-frequency and spatial feature localization of broadband oscillation based on the near-magnetic field radiation signal. Firstly

this paper analyze the generation mechanism of the near-magnetic field of the power module

and the spatial near-magnetic field signal is selected to characterize the broadband oscillation phenomenon inside the power module. This paper systematically expounds the implementation of the characterization platform architecture based on the double pulse test

online time-frequency and spatial feature localization

and offline working condition reconstruction algorithm. The experimental results show that the proposed characterization platform can effectively characterize the broadband oscillation problem in the power module

clearly identify the stage and device of the broadband oscillation

and predict the frequency corresponding to the spectral peak under the actual operating condition of the converter based on the test results under discrete testing conditions. The prediction errors of oscillation peaks at low and high frequencies are no more than 3 and 5 dBμV respectively

and the Pearson similarity coefficient of the envelope on the spectrum is 0.814

indicating that the overall trend is consistent. This study provides indicators and experimental tools for the modeling and analysis of broadband oscillation.