Arc faults have always been a concern for PV systems. Emerging safety standards will soon require PV system to include arc fault detection equipment. Arc faults can cause fires, shock hazard, and system failure. Existing techniques that rely on pattern recognition in the time domain, or amplitude detection in the frequency domain by using a Fourier Transform, don't work well for arcs because the signal to noise ratio is low and the arc signal is not periodic. Instead, wavelet transform analysis provides a time and frequency approach to analyze target signals with multiple resolutions. In this paper, a new approach using wavelet transform for arc fault detection in DC systems is proposed. Simulation models are established to study the theoretical results of the proposed methodology and traditional FFT analysis of arcing faults. Experimental data from an arcing AC system shows the efficacy is not limited to only dc power systems.