This item is non-discoverable
Investigating component life and led driver failure prediction method
De Vas Gunawardena, Andravas Patabendi Sachintha, Gayashan
This electronic version is a licensed copy owned by Rensselaer Polytechnic Institute (RPI), Troy, NY. Copyright of original work retained by author.
Today, LED lighting systems are used in many applications. These systems experience various thermal and operating conditions. Although the lifetime of an LED source is long, the lighting system life may be shorter due to the shorter life and failure of the LED driver. Understanding the factors contributing to the failure of LED drivers is important for estimating the lifetime of an LED lighting system and for designing more reliable LED drivers for lighting applications. Most LED drivers operate similar to a constant current switched-mode power supply (SMPS). Presently, a commonly accepted test method for estimating LED driver lifetime does not exist, so the industry uses statistical methods based on reliability handbook data to estimate the lifetime of LED drivers. However, these methods generally fail to account for real-world use patterns and operating environment conditions found with lighting systems in use. Furthermore, no studies to date have investigated how parametric degradation of several electronic components simultaneously affects the LED driver lifetime. This dissertation study investigated factors affecting the parametric degradation of the two most frequently failed components, namely, the electrolytic capacitor and the MOSFET, in LED drivers. Past studies have shown that more than 90% of driver failures were due to failures of these two components. In this study, their time-dependent parameter change patterns were identified under different thermal conditions and on-off switching patterns. A method was developed to estimate LED driver lifetime when the capacitor and the MOSFET in the LED driver degraded simultaneously in a driver circuit. Past studies have mentioned that the primary failure mechanism of an electrolytic capacitor is the evaporation of the electrolyte, which is activated by the high operating temperature of the capacitor. The second chapter of this dissertation describes the capacitor degradation study and its findings. Electrolytic capacitors are subjected to cycled thermal stresses to understand if thermal cycling introduces any additional failure mechanisms. The parametric degradation of capacitors was tested under temperature-cycled test conditions. There was no evidence to suggest that additional failure mechanisms were introduced to the electrolytic capacitor’s parametric degradation due to thermal cycling. Based on the experimental results, a method was developed to predict the time-to-failure of electrolytic capacitors operated under cycled thermal conditions by considering their operating thermal profile, and a quantity named Equivalent Constant Temperature was introduced for these thermal conditions. In an electrolytic capacitor, the parametric degradation results in a reduction of capacitance and an increment of equivalent series resistance (ESR) with time as a function of equivalent constant temperature. To the best of the author’s knowledge, this is the first time this finding has been reported. Similarly, the MOSFET’s failure mechanisms were studied under different thermal and use patterns. The failure mechanisms of MOSFET are either chip-related or package-related. This dissertation focused on package-related failures that increased on-state resistance (Ron) caused by solder failure induced by high temperature and on-off cycling. The third chapter describes the MOSFET degradation study and its findings, which investigated the impacts of different thermal cycle characteristics on the parametric degradation of MOSFETs. The level of degradation was quantified by measuring the Ron of the MOSFETs. It was observed that three factors of a thermal cycle–temperature difference (ΔT), maximum junction temperature (Tm), and thermal cycle on-time–accelerate parametric degradation, namely, the Ron increase of MOSFETs. To the best of the author’s knowledge, this is the first time evidence has been shown for creep stress (caused by the temperature during the on-time of the thermal cycles) accelerating MOSFET degradation, and thus Ron increase. After understanding how the two most frequently failed electronic components, the electrolytic capacitor, and the MOSFET, degrade and their parameters change over time when subjected to different thermal conditions and switching on-off cycles, in the fourth chapter, a method for estimating the useful life of an LED driver was investigated. When one or more components degrade simultaneously, the LED driver controller (a subcomponent in a driver that acts to regulate a constant current to the LED load) will change the operating point of the driver. Since none of the previous studies have investigated LED driver failure with more than one component degrading in a circuit simultaneously, this dissertation analyzed the parametric change of LED driver parameters, namely, output peak-to-peak voltage (Vp-p) and average input current (Iin), when the electrolytic capacitor and MOSFET are working together in an LED driver circuit. Degradation patterns of electrolytic capacitors and MOSFETs under different operating thermal conditions were used in a simulation study to identify how the above-mentioned driver parameters changed with these two component degradations. The driver output peak-to-peak voltage change corresponded to the capacitor degradation, namely, capacitance and ESR change, and the input current change corresponded to the MOSFET degradation, namely, Ron change. By measuring these driver parameters as a function of operating time and developing circuit-specific failure thresholds for an LED driver based on the driver controller stability criterion, a method was developed for LED driver lifetime prediction. Once again, to the best of the author’s knowledge, this is the first time a method to estimate LED diver failure has been proposed when two critical electronic components degrade simultaneously in an LED driver. This study also showed that the single-component failure time estimation for LED driver lifetime is not correct because multiple components can degrade simultaneously and cause the driver to fail faster than predicted by a single-component failure. This methodology can be implemented for estimating LED driver lifetime after obtaining component parameter degradations as functions of time under appropriate stress conditions and operating patterns experienced by the lighting system during application-specific conditions. The method can also be applied for failure prediction by implementing continuous driver parameter monitoring during lighting system application. This method can be extended for more than two components degrading in a driver circuit. The findings of this dissertation can influence a greater understanding of the failure modes and failure mechanisms of components in an LED driver under real-world operating conditions. The findings can also guide LED driver manufacturers, fixture integrators, and lighting professionals in evaluating an LED lighting system's long-term performance and useful lifetime, and guide industry regulators on what factors to consider when developing a standard test method for estimating a LED driver's useful lifetime. The new knowledge contributions to the field of electronic drivers from this dissertation are summarized below. The new knowledge contributions to the field of electronic LED drivers from this dissertation are summarized below. 1. This study showed that thermal cycling does not introduce additional failure mechanisms that affect the parametric degradation of electrolytic capacitors. a. Developed a method based on experimental results to estimate the electrolytic capacitor lifetime determined by degradation under cycled thermal conditions. 2. Analyzed the effects of different thermal cycling profile characteristics experienced by LED drivers on power MOSFETs' package-related failure mechanisms. a. Proposed a modification to the Coffin-Manson equation (an equation that can be used to model crack growth in solder due to repeated thermal cycling) to incorporate the dwell time of a thermal cycle on the number of cycles to failure for a MOSFET. 3. Developed a method to predict LED driver lifetime when two components degrade in the driver circuit simultaneously. a. The developed method can be implemented at the design stage of an LED driver to estimate useful life based on the driver operating environment and use pattern. b. The developed method can also be implemented for practical LED drivers with continuous driver parameter monitoring to predict failure.
School of Architecture
School of Architecture
School of Architecture
Rensselaer Polytechnic Institute, Troy, NY
Rensselaer Theses and Dissertations Online Collection
Users may download and share copies with attribution in accordance with a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 license. No commercial use or derivatives are permitted without the explicit approval of the author.