A contribution in partial discharge detection

Electric discharges that not completely bridge the electrodes are named Partial Discharges. Although the amplitude of such discharges is usually small, if compared with nominal conditions, they can cause progressive deterioration and may lead to ultimate breakdown. The apparent charge is the charge displacement in the leads of the sample subsequent a PD event. The measure of apparent charge is very important for two reasons: 1. The apparent charge ''q'' flowing through terminals of a sample is related to the energy of the PD; 2. The apparent charge is related to the size of defect. The measurement of apparent charge may be achieved using very different defection techniques. In classical systems, there is a lack of very important information: the value of the applied voltage which induces the discharge and the time instant in which if occurs. The proposed system allows the measurement of: 1. Apparent charge by means of direct analog integration of the current induced by the PD event,. 2. The corresponding value of the applied voltage ''V''; 3. The time instant in which the Partial Discharge occurs ''t(PD)'': This particular instrumentation, devoted to the measurement of partial discharges, is able to detect, measure and store in a large size memory the three measured data. If is so possible to track the real time evolution of the partial discharges. In order to obtain a good evaluation of the discharge characteristics a high-sensitivity and high accuracy detector has been developed using a mixed analog and digital signal processing technique. The system uses a high Q-notch filter to stop industrial frequency and the first odd order harmonics. integration of signal is then used to achieve apparent charge with lower spectrum distortion of the pulse than classical techniques. A block performs an analog to digital conversion of signals and stores them in memory for subsequent mathematical treatment. All the steps of measurement are supervised by microprocessor. This work presents an analysis which validates the previous results. In particular are discussed the practical difficulties to achieve an integration circuit for very sharp pulses in nanosecond range with very quick restoration for PD detection, minimizing errors and noise.