Please use this identifier to cite or link to this item:
Title: In-situ Health Monitoring Applied to High-Voltage IGBT Power Modules
Authors: Mandeya, Mabasa-Ashe Richard
Issue Date: 2019
Publisher: Newcastle University
Abstract: This thesis addresses an important issue of identifying insulated gate bi-polar transistor (IGBT) chip failures in multichip IGBT power modules. IGBT power modules are the dominant semiconductor devices of choice in high-voltage (HV) high-power converter applications which include domestic, commercial, automotive, railway, marine, aerospace and industrial applications. Commonly available HV IGBT power modules in the market are rated at 3.3 kV, 4.5 kV and 6.5 kV. These HV IGBT modules comprise several IGBT chips connected in parallel to achieve high-current capability; hence they are also known as multichip IGBT power modules. IGBT power modules are not flawless. The increased complexity of IGBT power module construction and inhomogeneous semiconductor chips make HV power modules less reliable. IGBT chips and electrical and mechanical interface material within the modules wear out and fail due to thermal cycling, operating environment or mishandling. IGBT failures while in application have repercussions on safety and failure costs. Thus the reliability of IGBTs while in their application is crucial especially in HV applications which comprise critical and large loads. To improve the reliability, an in-situ (online) health monitoring interface for HV IGBT power modules is proposed in this thesis. Two distinct advantages of in-situ IGBT health monitoring are that it allows IGBT module replacement prior to complete failure thus reducing safety and reliability risks. The second advantage is that the interval time for IGBT maintenance work can be tailored towards the real degradation rather an obligatory fixed time interval thus reducing maintenance costs. In large power modules, it is common to have IGBT chips as well as anti-parallel diode chips within the power module. This research focusses only on the health monitoring of the IGBT chips and not the diode chips. The main reason is that IGBT chips experience higher thermal stresses compared to diodes hence IGBT chips are more susceptible to failures compared to diode chips. In practice, IGBT chip failures are accompanied by a change in junction temperature. Thus this thesis proposes the use of temperature- sensitive electrical parameters (TSEPs) for in-situ health monitoring of IGBT power modules. Following a comparison of twelve traditional online TSEPs from literature and five new TSEPs proposed in this thesis, this thesis employs a novel TSEP, gate-emitter prethreshold voltage (VGE(pre-th)) as a health-sensitive parameter (HSP) for chip failure detection in multichip HV IGBT power modules. A VGE(pre-th) online chip loss monitoring circuit has been successfully implemented on a commercially available IGBT gate driver. VGE(pre-th) is measured at a fixed pre-determined instant of the gateemitter voltage (VGE) between the VGE zero-crossing (VGE(0)) and threshold voltage (VGE(th)) during IGBT turn-on. VGE(pre-th) requires low hardware with only a voltage sensor and a counter. Since it is based on the low-voltage (LV) gate side rather than the HV collector side of IGBT, VGE(pre-th) does not require HV isolation or HV insulation. Simulation and experimentation of 16-chip 3.3kV 800A DIM800NSM33-F IGBT power modules from Dynex Semiconductor Limited (Ltd) have shown that VGE(pre-th) has a good accuracy and repeatability; a linear sensitivity of 500 mV/chip loss with IGBT chip failures; a linear virtual junction temperature (Tvj) sensitivity of -2.2 mV/°C and tracks the highest chip temperature. It has thus been concluded that VGE(pre-th) can be used for both Tvj and IGBT chip failure monitoring in HV IGBT power modules. VGE(pre-th) can be tested during normal IGBT turn-on operation or during the off-state of the IGBT. In both cases the same information about temperature and loss of chip number can be detected which makes VGE(pre-th) more versatile than any other TSEP or HSP.
Description: PhD Thesis
Appears in Collections:School of Electrical and Electronic Engineering

Files in This Item:
File Description SizeFormat 
Mandeya R 2019.pdfThesis5.29 MBAdobe PDFView/Open
dspacelicence.pdfLicence43.82 kBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.