Adaptive Stray Inductance Extraction Algorithm Using Linear Regression for Power Module With High Noise Immunity and Accuracy

Stray inductance has great impacts on characteristics of power module and how to extract the inductance accurately is a significant challenge to guide the layout design and application of power module. The inductance during rapid gradient of current poses threaten to power module so the suitable stages for inductance extraction are recommended on account of dynamic characteristics. The existing method extracts inductance by two measure points and accuracy is low, which is influenced by measurement error significantly, so this paper proposes an adaptive stray inductance extraction algorithm, the essence of which is the least square method. The proposed algorithm applies least square method to extract stray inductance by fitting numerous sample points by double pulse test and eliminates the influence of measurement error because the residual obeys normal distribution same with measurement error. As a result, the proposed algorithm is with high noise immunity and accuracy. Finally, a multi-chips IGBT power module is tested under various conditions to verify the effectiveness of proposed method. Consequently, the error of extracted inductance is less than 5% and consistency is great.