Comparative measurements of two partial discharge measurements used for form wound coils/bars as quality assurance tests

Several utilities both in Canada and in the USA have been using commercial on-line partial discharge (PD) systems as on-line monitoring tools. Partial discharge tests have been used historically to characterize bars/coils that have been sent to Powertech Laboratories for voltage endurance and thermal cycling tests so that the bars that displayed the highest PD quantities were subjected to voltage endurance tests. Utilities are now beginning to specify the limits for PD magnitudes and other related PD quantities as a quality assurance measure in their technical specifications for large motors and hydrogenerators. There appears to be confusion as to what are the reasonable limits for these quantities and whether or not these limits make practical sense. Since Powertech has had the unique opportunity to examine several insulation designs over the last decade this paper is an early attempt to document the current levels of PD quantities that are being measured on bars/coils being manufactured with modern insulation systems. No attempt has been made to correlate these quantities with its performance on voltage endurance (VE) test times but this work is ongoing. Several bars from different manufacturers were subjected to PD measurements at their normal line-to-ground voltage. These PD measurements were made using Powertech's classical PD system as well as a commercial system. One significant difference was that the PHA system uses pico-coulomb to represent charge magnitude while PDA system uses millivolt. Laboratory calibration was done to correlate these discharge magnitudes. Results showed that if the discharge magnitudes and the number of PD pulses were small both the classical and the commercial system gave similar PD responses. On the other hand if the discharge rate (number of pulses per second) was large the classical PD measurement system tended to saturate. Historically, when such saturation occurred during measurements with the classical PD system, measurements were repeated at a lower test voltage (for example 0.8 pu). Subsequent dissections of samples that exhibited such high PD showed the presence of a large number of voids in the groundwall insulation.