The procedures and results are described for an experiment to seek electrical current signals during water treeing, and to characterize any that exist. The apparatus was described in the companion paper Part 1, where the sensitivity, characteristics, and noise effects were assessed carefully because such signals had not been detected before. In this Part 2 the use of the several modes of the apparatus is described for various conditions of signal, noise, and phase resolution. Methods of identifying and eliminating noise artifacts are described. A method is described for resolving the HV phase existing when a fast event occurs. Four types of pulsed currents are studied; two are noise (1-channel and 2-channel noise) and two are signal (fast and slow pulses). Each has unique characteristics as to the channel(s) in which it occurs, amplitude, shape, and phase of the 60 Hz HV when it occurs. It is observed that two types of pulsed signals originate only in the region where water treeing is occurring; any conclusion that they are produced by the water treeing process is inferred from that observation. The 'fast pulse' signal lasts only a few days, is < 25 ns wide, contains 1 to 4 fC charge, and occurs only at the water treeing region and only when the 60 Hz HV waveform is zero and falling. The 'slow pulse' signal appears after a few days, is 35 to 60 mu s wide, has 70 to 125 nA peak, and occurs only at the water treeing region and only when the 60 Hz HV waveform is at the negative minimum.
