High fidelity “replay” arc fault detection testbed

Abstract

Arc faults in photovoltaic power systems pose safety concerns ranging from localized damage to the equipment, electric shock hazard for humans, and fire that spreads to buildings and beyond the PV systems. Thus, robust and effective detection of initial arcs before they become sustained arc faults is imperative. However, high frequency noise caused by the switching of DC/DC power optimizers or DC/AC inverters can mask the signals produced by an arc, making it difficult or impossible to reliably identify if an arc is occurring. This paper discusses a testbed that was developed for the purpose of testing arc fault detectors in a laboratory environment using precise-reproduction, or replay, of pre-recorded arc signals. The testbed is capable of replaying both the arc signature and the noise from the power electronic circuits at proper amplitude to represent real-world conditions. The testbed is characterized and validated by frequency analysis across the range of frequencies typically associated with an arc fault. Fast Fourier Transform (FFT) analysis of reproduced arc signals further justifies the effectiveness of the testbed and a certified arc fault detector (AFD) is tested using reproduced arc signal. Utilization of such a testbed will facilitate the study of reliable detection algorithms.