Skip to Main Content
A new method of transfer function analysis using spectral density estimates in conjunction with a random error and coherence function calculation has shown excellent experimental results in identifying mechanical displacements in the windings of power transformers. The new technique represents an extension of the traditional low voltage impulse method of frequency response analysis (FRA). Traditional low voltage impulse methods have been used to detect mechanical displacements in transformer windings or clamping structures through detection of changes in the transfer function, but have been limited in practical value due to the sensitivity of the transfer function to differences in the time parameters of the applied impulses, as well as sensitivity of the transfer function to variables such as oil condition and temperature. The new technique offers improved repeatability in diagnosing winding movements, and is based on an objective numerical calculation to quantify changes through a weighted normalized difference calculation.