A real-time algorithm is developed for the detection of series dc arc faults in a grid-tie solar photovoltaic (PV) installation. The sensed dc bus current, which is sampled using an analog-to-digital converter with Galvanic isolation, is filtered using a wavelet-based, two-level filter bank. The filter bank, referred to as the post-processing filter, improves the robustness of the algorithm to any false tripping by rejecting power converter harmonics that are added to the dc bus current. To determine if a fault has occurred, the algorithm calculates the variance of the filter bank output and sees if the calculated variance exceeds an upper threshold value. If the upper threshold is exceeded, and the dc bus voltage falls below a predefined lower limit for a set number of instances, the algorithm trips. The algorithm can detect a series arc fault in under two seconds and does not rely on machine learning techniques to process the sensed signal. The detection algorithm is implemented on a commercial microcontroller using C code, and the filter bank convolutions are implemented using 32-bit floating point variables. / Master of Science / A device is developed for the detection of series dc arc faults in solar photovoltaic installations. Dc arc faults that result from loose connections or worn cable insulation can go unnoticed by most conventional fault detectors. Once it has ignited, the series arc can generate considerable amounts of heat and poses a significant fire risk. By contributing to the development of a dc arc fault detection system, the intention is that dc renewable energy distribution systems, most notably solar photovoltaic installations, can gain even more widespread adoption. This would make a significant impact towards decarbonizing the energy sector and tackling the threat to society posed by climate change.