Reliability Prediction of Bridgeless AC-DC SEPIC with V-Fill for LED Applications

Abstract

LED luminaires have been one of the lighting technologies that is presently most energy efficient and developing at a fast rate because of its long lifespan and outstanding luminous efficiency. To power the LED, power converters are used as LED drivers. In this paper, the reliability prediction of bridgeless AC-DC SEPIC (Single Ended Primary Inductance Converter) PFC (Power Factor Correction) integrated valley-fill (V -Fill) circuit has been developed for LED applications. The ability of a product to execute the necessary function under circumstances for a predetermined duration of time is referred to as reliability. Reliability is frequently assessed in terms of availability, frequency, and chance of failure. Preventing the occurrence of faults is the fundamental goal of reliability prediction. The choice of AC-DC PFC topology affects the overall reliability of led driver. To anticipate converter reliability, consideration must be given to the operational environment and time. Analyzing the power converter's condition is necessary to prevent converter replacement and downtime. As a result, this paper provides an illustration of the performance degradation analysis, reliability and failure investigation of individual components used in Bridgeless AC-DC SEPIC with V-Fill converter. Based on the MIL-HDBK-217F manual, the Part Stress Analysis (PSA) approach is used to compute every critical component failure rate (λ) and the mean time to failure (MTTF) of the converter. The PSA method requires more amount of information, and it is unable to give an exact value for reliability. Therefore, to predict an exact failure rate of critical components, a Bayesian Belief Network (BBN) method is used, and the BBN graphical structure of proposed converter, the failure rate of critical component, and MTTF is computed using GenIe software. The complex probability issues are efficiently computed via BBN. Therefore, the enhanced BBN technique can be an alternate to PSA approach in the lifetime study of LED drivers. It can produce exact data for designing long-lasting LED drivers, while also considerably increasing design/development efficiency and reducing costs.