Improved technique for fault detection sensitivity in transformer impulse test

Insulation of transformer windings can be damaged by impulse voltages of high amplitude produced by lightning or switching transients. Testing of the windings insulation is performed by using the standard impulse test, generally applying the fast Fourier transform (FFT) algorithm to analyze the transformer `fingerprint´ in the frequency domain. However, FFT analysis is limited by the spectral leakage and degradation at spectral peaks. To obtain a better signature analysis, this paper investigates the use of power spectral density (PSD) and wavelets analysis techniques. Experimental results are shown from different tests with windings under normal and turn-to-turn short circuit conditions in a single phase transformer