Analysis of DC Arc Characteristics in an Advanced Measurement Setup

Faults in DC systems (e.g. photovoltaic systems), such as serial arcs, pose a significant safety risk. Reliable detection of these faults is therefore critical and requires a thorough understanding of the arc itself. Most literature suggests that such serial arcs exhibit pink noise (1/f behavior in the power spectral density (PSD) of the arc voltage signal. This characteristic is often used as a basis for arc detection. As part of this work, an advanced arc generator was designed and built that is both mechanically very precise and equipped with high-precision measurement technology to verify the claims about arc behavior under a wide range of conditions, such as different electrode materials (tungsten alloy, copper, aluminum, MC4 connector) or different ambient humidity levels. Measurements with tungsten alloy electrodes show that the expected pink noise characteristic is met only with clean elec-trodes and deviates considerably with repeated ignitions. There is also evidence that this characteristic depends on the surface condition of the electrode, such as the material or wear of the electrodes, which presents an additional challenge to the practical application of arc detection algorithms. Further series of measurements with different electrode materials and MC4 connectors commonly used in PV systems show that the decrease of the spectrum with increasing frequency depends on the material as well as on the surface condition (clean vs. worn) and the current level. However, the influence of ambient humidity on the arc characteristics seems to be negligible. The measurements performed clearly show that the widespread assumption that arcs exhibit pink noise behavior in the PSD of the arc voltage is only valid to a very limited extent. This is especially true for arcs that have been generated under more realistic conditions.