Experimental determination of the valid time range for Kerr electro-optic measurements in transformer oil stressed by high-voltage pulses

Knowledge on the electric field distribution in an insulating liquid is helpful for the improvement of its electrical breakdown strength. Kerr electro-optic field mapping measurements are taken in the present work to measure the electric field and space charge distributions between parallel-plate electrodes in transformer oil, a low Kerr constant dielectric liquid. With the use of a high-sensitivity CCD camera, we can directly record the light signal variation due to the electric field induced birefringence, while traditional Kerr electro-optic measurements with transformer oil needed ac modulation and could not work under high voltage pulses. Our experimental setting is a linear polariscope (without wave plates) with crossed polarizers. When there is no electric field in the gap, there should be no transmitted light in principle. However, weak light signal can be registered by the CCD camera with single photon detection capability. By increasing the output power of the pulsed laser, we can make this `leaked´ light intensity (zero-field level) higher than the noise level (environment light fluctuation, camera internal error, etc) and thus distinguish the two. The transmitted light intensity will be even higher when high voltage is applied across the gap. Differentiating the light intensities with and without high voltage, we can evaluate the corresponding electric field at each pixel in the gap. The pulsed laser is triggered with various delays from the start of the high voltage pulses (pulse duration of 20 ms, rise time of 250 μs, and peak voltage of 30 kV). The experiment results and analysis will be presented in this paper. We find that, there exists a time range beyond which the mapped electric field distribution is no more accurate. This is due to the disturbance of flow induced by field stress or current heating. The flow disturbance will break the validity of Kerr electro-optic measurement principle in which no such effect is considered. An extreme - xample is that even when the high voltage pulse has already passed (e.g. 25 ms delay), the detected light intensity is still several times higher than expected.