We carried out a study on partial discharge (PD) phenomena in SF6 gas and tried to construct a diagnostic system capable of separating noise from observed PD signals for assessing the condition of insulation performance in power apparatus. We have constructed a computer-aided P D measuring systein which permits a time-sequential data acquisition and phaseresolved pulse height analysis. Using the P D measurement system, we measured 60 He ac PD in SF6 gas and in air at' atmospheric pressure for needle-plane electrodes with three different radii at the tip of the needle, 3, 30 and 500 pm. We ptoposed particular PD statistical parameters, unbalance ratios U, and U, of the pulse repetition rate and average of charge, respectively, between positive and negative PD pulses occurring per cycle; these parameters proved t o be good indicator? to distinguish PD occurring in SF6 from that in air. The mechanism of PD inception was also discussed.
INTRObUCTlONFe gas has been used widely as an insulating mate-S rial because of its superb properties. It meets the social demand by making pbwer apparatus much smaller in siee, more efficient, larger in capacity, and environmentally clean [l]. With respect to power apparatus, monitoring and diagnostic systems have been used extensively to prevent failures of apparatus. The conventional sensing system for diagnosis, however, is not capable to identify the cause and to determine the location of the failures. Accordingly, improving the technology of sensing and diagnostics becomes a crucial issue.Recently, Baumgiirtner et d. [2,3] have tried to assess the insulation performance and to identify the cause of PD using phase-resolved PD pulse height analysis; also several reports have dealt with the PD inception mechanism in SF6 gas [4-71. Fundamentals of the PD occurrence phase distribution have not been fully understood. It is also a crucial issue now to improve the reliability of diagnosis of insulation performance. At the same time, since in a field test the electromagnetic noise arises from PD in air, it is necessary for assessing the condition of insulation performance to separate the noise from the PD signal.We have attempted to construct a computer-aided PD measuring and diagnostic system to study PD phenomena in SF6 gas; the investigation must provide a basis of the assessment made by the system. The computeraided PD measuring system was developed to permit time-sequential data acquisition and phase-resolved pulse