Detection and Identification of Power Switch Failures Using Discrete Fourier Transform for DC-DC Flying Capacitor Buck Converters

Prompt detection of power switch failure is crucial for fault-tolerant operation of multilevel converters. This article presents a new technique of detecting, identifying, and locating a device fault in a multilevel dc-dc flying capacitor (FC) buck converter (FCBC). This is realized by continuously analyzing the magnetic component (inductor or transformer) voltage harmonics for failure detection and subsequently identifying failure type and location with a look up table method. The harmonics analysis is performed by the existing FPGA controller using discrete Fourier transform (DFT). This new approach uses an auxiliary winding in the existing inductor core without adding any new hardware and can identify the location and fault type with good accuracy and speed. The experimental results validate the proposed concept on a three-level GaN-based FCBC prototype.

Automated summary

This article introduces a new technique for detecting and locating device faults in multilevel converters. The method utilizes voltage harmonics of magnetic components for fault detection, and identifies fault type and location using a look up table method. Experimental results validate the effectiveness and accuracy of this approach in practical applications.