Ageing defect detection on IGBT power modules by artificial training methods based on pattern recognition

• DOI: 10.1016/j.microrel.2010.08.006
• Publication Date: 2010-09-25T08:30:08Z
• Authors: A. Oukaour; B. Tala-Ighil; B. Pouderoux; M. Tounsi; M. Bouarroudj-Ber...; S. Lefebvre; B. Boudart
• Keywords: Insulated-gate bipolar transistor; Power cycling

Abstract The ageing of power insulated gate bipolar transistor (IGBT) modules is mainly related to thermal and thermomechanical constraints applied to the device. This ageing causes degradation of the device performances and defects appearance which can lead to failures. To avoid these failures, the follow-up of the device operation and the detection of an ageing state remain a priority. This paper presents, at first, ageing tests of 1200 V–30 A IGBT module subjected to power cycling with the aim to highlight online and real-time measurable external indicators of ageing. Secondly, these indicators are used to develop a failure diagnosis method. The diagnosis is realized by artificial training methods based on pattern recognition.