Partial Discharge Testing and Detection under PWM Voltage

Partial Discharge detection and measurement is an important part of electric insulation design. However, in PWM (pulse width modulation) voltage source converter environments, the noise resulting from switching voltage (rise and fall times of tens of nanoseconds) makes detection and extraction of partial discharge difficult. Unique methods of partial discharge detection are needed to address and decouple the noise from partial discharge measurements. PWM voltage can feature high switching speeds and high dv/dt during voltage switching. These PWM voltage behaviors are not found in traditional utility high voltage applications, and the effects that this type of voltage has on insulation and associated partial discharge behavior are not well understood. A good understanding of partial discharge behavior is vital to effective insulation design in high voltage power electronic systems. \\ To better understand the behavior of partial discharge in a PWM voltage source environment, a partial discharge testing platform is designed. This testing platform features a 3kV Si half-bridge converter to apply PWM voltage at varying switching frequency (5-60kHz) and dv/dt (11-16V/ns). An additional full-bridge 10kV SiC converter provides extra voltage capability for partial discharge testing. Converters with wide bandgap semiconductors can feature high switching frequency and dv/dt not currently possible with Si converters. The effect that switching frequency and dv/dt has on partial discharge behavior will become more of an issue as wide bandgap semiconductors are operated at full potential. The output voltage from this converter will be applied to test subjects chosen to model types of partial discharge: surface, internal, and corona discharge. This testing platform requires an effective method of partial discharge detection that is not interfered by voltage switching transient noise. Three electrically isolated methods of detection are explored to determine the most effective methods of partial discharge detection in the presence of PWM voltage switching transient noise: optical detection, electromagnetic detection, and acoustic detection. Post-processing wavelet denoising techniques are applied to remove any remaining noise from partial discharge signals. With effective partial discharge detection and a testing platform with test subjects chosen to model types of discharge, the effects of PWM voltage on partial discharge behavior are demonstrated. Methods of partial discharge classification and identification are demonstrated as well to show methods to effectively identify defects in insulation caused by partial discharge due to PWM voltage.