Abstract: Drive circuit is the guarantee for the normal operation of Insulated Gate Bipolar Transistor (IGBT). Based on the incipient fault theory of analog circuit, an incipient fault diagnosis method based on Scale Space Empirical Wavelet Transform (SS-EWT) combined with Back Propagation Neural Network (BPNN) for 200 A/1200 V fast IGBT driver circuit based on EXB841 driver module is proposed. Firstly, based on the analysis of the hardware structure and working principle of the drive circuit, the sensitivity analysis of the main components of the drive function part and short-circuit protection function part is carried out to select the test objects. Secondly, the data obtained by fault injection were processed by SS-EWT and constructed into the incipient fault eigenvector. Finally, BPNN is used to train the eigenvector to recognize the incipient fault. In view of the difficulty in distinguishing some fault categories of short-circuit protection elements, retraining is used to optimize the diagnosis method. The simulation results show that the incipient fault diagnosis method based on SS-EWT for driving function elements and short-circuit protection elements can reach the accuracy of 91.04% and 96.05%, respectively.