A prognostic approach for non-punch through and field stop IGBTs

doi:10.1016/j.microrel.2011.10.017

Microelectronics Reliability

Volume 52, Issue 3, March 2012, Pages 482-488

https://doi.org/10.1016/j.microrel.2011.10.017 Get rights and content

Abstract

Development of prognostic approaches for insulated gate bipolar transistors (IGBTs) is of interest in order to improve availability, reduce downtime, and prevent failures of power electronics. In this study, a prognostic approach was developed to identify anomalous behavior in non-punch through (NPT) and field stop (FS) IGBTs and predict their remaining useful life. NPT and FS IGBTs were subjected to electrical–thermal stresses until their failure. X-ray analysis performed before and after the stress tests revealed degradation in the die attach. The gate–emitter voltage (V_GE), collector–emitter voltage (V_CE), collector–emitter current (I_CE), and case temperature were monitored in situ during the experiment. The on-state collector–emitter voltage (V_CE(ON)) increased and the on-state collector–emitter current (I_CE(ON)) decreased during the test. A Mahalanobis distance (MD) approach was implemented using the V_CE(ON) and I_CE(ON) parameters for anomaly detection. Upon anomaly detection, the particle filter algorithm was triggered to predict the remaining useful life of the IGBT. The system model for the particle filter was obtained by a least squares regression of the V_CE(ON) at the mean test temperature. The failure threshold was defined as a 20% increase in V_CE(ON). The particle filter approach, developed using the system model based on the V_CE(ON), was demonstrated to provide mean time to failure estimates of IGBT remaining useful life with an error of approximately 20% at the time of anomaly detection.

Introduction

Insulated gate bipolar transistors (IGBTs) are the devices of choice for medium and high power, low frequency applications such as traction motors, high power switch mode power supplies, and variable speed drives [1]. Development of diagnostic and prognostic approaches has been motivated by reports of IGBT failures [2], [3]. There have been several studies reported on IGBT anomaly detection. Xiong et al. [4] proposed an online diagnostic system to predict the failure of an automotive IGBT power module using a look-up table for the collector–emitter voltage. Ginart et al. [5] developed an online ringing characterization technique to diagnose IGBT faults in power drives. Oukaour et al. [6] developed an approach to determine defective IGBTs from healthy IGBTs using neural nets. Lu et al. [7] used a physics-based strain model to determine the remaining life of an IGBT module. Saha et al. [8] implemented the particle filter algorithm for the prediction of the remaining useful life of a punch through (PT) IGBT, using the trend of the collector–emitter current at turn-off.

Although several approaches have been investigated for specific applications, there is a need for a comprehensive approach that enables both the detection of anomalous behavior and the prediction of remaining useful life (RUL). In this study, a prognostic framework for IGBTs is proposed and implemented. This framework involves the use of the Mahalanobis distance to detect anomalies in the IGBT and the particle filter algorithm to predict RUL. Non-punch through (NPT) and field stop (FS) IGBTs were subjected to electrical–thermal stress by power cycling and several device and load parameters were monitored. Features from the monitored data were used to compute the Mahalanobis distance (MD). The MD was transformed by normalization using Box–Cox transformation and used with an appropriate threshold to detect anomalous behavior. Upon anomaly detection, the particle filter algorithm was implemented to predict the remaining useful life.

This paper is organized in five sections. Following this introductory section; in Section 2, the experimental approach and testing procedure are described. In Section 3, the prognostic framework is discussed. In Section 4, the results obtained by implementation of the prognostics framework are presented. In Section 5, the results of this study are summarized and future work is discussed.

Section snippets

Experimental approach and degradation analysis

Electrical–thermal stress tests were performed on 600 V rated non-punch through (NPT) and field stop (FS) IGBTs manufactured by International Rectifier. The experimental setup used for power cycling of IGBTs under a resistive load is shown in Fig. 1.

The experimental setup consists of a gate driver board, a main board, and a power conditioner board. The main board houses the terminal block for the IGBT device and BNC output ports connected to the terminal block, one port each for the gate–emitter

Prognostics framework

The prognostics framework developed and implemented in this study is shown in Fig. 7. This framework involves the measurement of I_CE(ON) and V_CE(ON) at a constant temperature using appropriate current and voltage sensors. Mahalanobis distance (MD) is used as a diagnostic parameter with a defined trigger for detection of an anomaly. The MD calculation is made using the V_CE(ON) and the I_CE(ON). Once an anomaly is detected by the MD approach, the particle filter (PF) process is initiated for time

Results and discussion

Using the MD approach, anomalies were identified in the NPT and FS IGBTs. In Fig. 8, the anomaly detection results are shown for an NPT IGBT. The transformed MD threshold for anomaly detection was 2.7, the anomaly was detected at 4.4 h from the beginning of the test, and the time for parametric failure, which is the increase in V_CE(ON) by 20%

Summary and conclusions

In this study, a prognostics framework for IGBTs was developed and implemented for IGBTs that uses the Mahalanobis distance approach for anomaly detection and particle filters for time to failure prediction. The particle filter approach, developed using the system model based on the V_CE(ON), was demonstrated to provide estimates of IGBT remaining useful life with an error of approximately 20% at the time of anomaly detection. The MD based probabilistic threshold approach implemented in this

Acknowledgments

The authors would like to thank the more than 100 companies and organizations that support research activities at the Center for Advanced Life Cycle Engineering at the University of Maryland annually. The authors would also like to thank the members of the Prognostics and Health Management Consortium at CALCE for their support of this work. The authors thank Dr. Jose Celaya for his assistance in the setup of the IGBT experimental test bed and Dr. Bhaskar Saha for his guidance on particle filter

References (13)

X. Perpina et al.
IGBT module failure analysis in railway applications
Microelectron Reliab
(2008)
A. Oukaour et al.
Ageing defect detection on IGBT power modules by artificial training methods based on pattern recognition
Microelectron Reliab
(2011)
H. Lu et al.
Design for reliability of power electronics modules
Microelectron Reliab
(2009)
B. Baliga
Fundamentals of power semiconductor devices
(2008)
Fuchs F. Some diagnosis methods for voltage source inverters in variable speed drives with induction machines – a...
Y. Xiong et al.
Prognostic warning system for power-electronic modules in electric, hybrid electric, and fuel-cell vehicles
IEEE Trans Indust Electron
(2008)

There are more references available in the full text version of this article.

Cited by (82)

Application of machine learning algorithms in prognostics and health monitoring of electronic systems: A review
2023, e-Prime - Advances in Electrical Engineering, Electronics and Energy
In the modern age of digitalization, electronics are fundamental to any engineering system. With the current strong focus on the Internet of Things (IoT), autonomous vehicles and Industry 4.0, reliable electronics are gaining crucial importance. Predicting the health of complex systems is able to avoid catastrophic failures. Prognostic and Health Monitoring (PHM) approaches are an important step toward trustable and reliable electronics. Nowadays, Artificial Intelligence (AI) and machine learning (ML) algorithms are integrated into PHM approaches, enabling complex fault diagnosis. In this contribution, we provide an overview of the application of intelligent algorithms in PHM of electronics in a systematic manner. The challenges of prognostics in electronics are provided and a detailed overview of the available PHM precursors for various electronic components and the associated selection process is given. Based on the literature review conducted, the main research challenges with ML algorithms in PHM are discussed along with performances of each model.
Recursive zonotopic set-membership approach for system-level prognostics with application to linear parameter-varying systems
2023, ISA Transactions
A robust recursive zonotopic set-membership approach for remaining useful life forecasting with application to linear parameter-varying systems is proposed in this paper. The proposed approach addresses systems with degraded components formulated as a system-level prognostics problem. Thus, the critical degraded components of the system are augmented to the states resulting a nonlinear system that is reformulated as a linear parameter-varying model. Hence, joint estimation of states and parameters is adopted in a zonotopic set-membership scheme with an optimal linear matrix inequality-based tuning and assuming unknown-but-bounded noises and uncertainties. As a result, a recursive zonotopic set-membership approach is proposed for remaining useful life forecasting based on the prediction of the failure precursors of degraded systems. Finally, this approach is tested on a DC–DC converter case study with unknown degradation behaviors, and the obtained results show the estimation and the forecasting accuracy of this methodology.
A health state prediction method of traction converter IGBT based on optimized particle filter
2022, Microelectronics Reliability
Citation Excerpt :
Since the degradation trajectory of the IGBT health state characterization parameters is a nonlinear and non-Gaussian noise system, Kalman filter, extended Kalman filter, and unscented Kalman filter cannot provide a better estimation accuracy for the system with non-Gaussian noise, while the particle filter can obtain a better estimation accuracy when the system has nonlinear and non-Gaussian noise. In [23], a particle filter is used to predict the remaining useful life of the IGBT. However, the traditional particle filter method suffered from two problems: particle degradation and particle impoverishment [27].
Insulated gate bipolar transistor (IGBT), as the most critical and vulnerable power electronic component in the field of rail transportation, its reliability has become a research hotspot in recent years. To improve the safe and reliable operation of rail transportation and to reduce the maintenance cost by rationalizing system maintenance, it is critical to monitor the health of the traction converter IGBTs. This paper first reviews the failure mechanism of the IGBT, then analyzes the health state characterization parameters of the IGBT by the accelerated aging test, selects thermal resistance as the health state characterization parameter of the IGBT, and establishes the degradation model of the IGBT. On this basis, a health state prediction method of IGBT based on the genetic algorithm unscented particle filter (GA-UPF) is proposed. This method uses an unscented Kalman filter to generate important proposal density distribution for each particle, estimates, and updates the parameters of the degradation model, alleviates the particle degradation problem, and uses a genetic algorithm to solve the particle impoverishment problem introduced by the traditional resampling method, which improves the particle diversity and prediction accuracy. Mahalanobis distance is used to monitor the abnormal state of the thermal resistance. After the abnormal occurs, the value of the thermal resistance is predicted according to the degradation model and updated parameters. The degradation trend of the thermal resistance is predicted to reflect the health state of the IGBT. Experiments on IGBT degradation data show that the proposed method can provide higher health state prediction accuracy than the traditional particle filter method.
An artificial neural network supported stochastic process for degradation modeling and prediction
2021, Reliability Engineering and System Safety
Citation Excerpt :
Modern manufactured products, such as bipolar transistors [1], seals [2,3], and LED-based lights [4], have seen a growing demand for high reliability.
An artificial neural network supported stochastic process for degradation modeling and prediction is proposed in this paper. An artificial neural network is applied to describe the degradation path due to data fitting flexibility and path description considering degradation path uncertainty. The assumption that the initial degradation is zero in the stochastic process is freed. The artificial neural network supported stochastic process is trained by minimizing the minus log-likelihood offline-based on the run-to-failure degradation data. Considering unit-to-unit variance in population degradation modeling, the process parameters are assumed to be randomly distributed. Here, three common distributions describe the process parameters, and Akaike information criteria applied to select distributions of process parameters. The process parameters are evaluated by Bayesian inference based on the trained path, and distributions of process parameters are based on real-time degradation data for online prediction. The proposed method is verified by a simulation experiment based on a Wiener process, which seemed a true model. Furthermore, an actual degradation dataset is also applied to illustrate the effectiveness of the proposed method. Both the simulation experiment and actual example indicate that the proposed stochastic process is capable of degradation modeling and degradation predicting, even without prior information about the degradation path.
The aging analysis method of IGBT with composite failure mode based on Miner linear fatigue cumulative theory
2021, Microelectronics Reliability
Insulated Gate Bipolar Transistor (IGBT) is the core power switch device in electrical field. IGBT's reliability is the basis of safe and stable operation for electrical system or equipment. At present, the aging analysis of IGBT mostly focuses on a single failure mode. This paper proposes a new aging parameter based on Miner's linear fatigue cumulative theory. Then this parameter is used to characterize the aging state of IGBT when multiple failure modes act simultaneously. The aging analysis method of IGBT considering composite failure mode is proposed. Firstly, the bonding wire and solder layer are analyzed by finite element method. Then analyzed according to the load, the master S-N curve method is used to calculate the fatigue accumulation. The larger value is taken as the overall aging degree of IGBT module. Finally, the aging analysis method of IGBT based on Miner's linear fatigue cumulative theory under composite failure mode is verified by data analysis. This method can not only be used for aging analysis and lifetime prediction of IGBT, but also can predict the final failure mode of IGBT to a certain extent.
Fault diagnosis and failure prognosis of electrical drives
2021, Fault Diagnosis and Prognosis Techniques for Complex Engineering Systems
The interest in reliable operation of electrical drives has increased in recent years. recently an AC electrical machine was typically a synchronous generator or a three-phase uncontrolled motor that would operate connected directly to the power grid, or if a DC machine, its speed would be controlled through a rectifier. Although these applications have been augmented by many others. Many developments have led to this explosion of applications. The ubiquitous use of power electronics and microcontrollers, DSPs, etc., have allowed control of speed, torque etc., and the introduction of new motor types: permanent magnet AC, switched and synchronous reluctance etc. with increased versatility, higher torque and power density. These new power electronics-controlled machines have been finding uses in applications in manufacturing, robotics, vehicle traction and operation, airplane propulsion, wind and wave generators, but also in medical, commercial and consumer electronics that are becoming vital in health and other necessary devices.
Fault diagnosis in electrical drives covers the faults in the power electronics, the DC link capacitor, batteries if present, and electrical machines. In power electronics in particular, the faults can be open or short circuit of switches, and they can be identified from current and voltage measurements. In electrical machines the faults can be in the windings, with either incipient in precipitous degradation, and can be identified and their severity determined using either model- or signal-based techniques. Mechanical faults in bearings are usually detected through the measurement of vibrations, while eccentricity can be detected through the change of flux and inductances.
Most of these faults can be identified accurately, and their development predicted. They can lead then to the estimation of the remaining useful life, and together with partial or full redundancies, can result in increased reliability on the drive.

View all citing articles on Scopus

View full text

A prognostic approach for non-punch through and field stop IGBTs

Abstract

Introduction

Section snippets

Experimental approach and degradation analysis

Prognostics framework

Results and discussion

Summary and conclusions

Acknowledgments

Microelectron Reliab

Microelectron Reliab

Microelectron Reliab

Fundamentals of power semiconductor devices

Prognostic warning system for power-electronic modules in electric, hybrid electric, and fuel-cell vehicles

IEEE Trans Indust Electron