I want to build a better Partial Discharge (PD) detector for events that of ten Oil filled Transformers. THese events are the same as ESD finger discha rges but internal to the transformer. I have measured about 2 Watt peak pul ses with < 1ns rise time and < 10ns pulse duration. Typically these are det ected with capacitive or inductive conducted current sensors or radiated in ductive sensors or antenna. Some have used optical sensors to detect the p ulse of light which I believe includes resonant bands for hydrogen and meth ane in IR and other bands in optical and UV.
My method is a loop CT aka loop antenna around the high voltage insulator t o a grounded coax cable on the outside.
Now I would like to try optical CCD or CMOS sensing of PD activity for imag e capturing of these flashes. Repetition rates are low from 1pps to 10 pps to 100 pps to 1kpps according to excitation currents and voltage and conta mination levels.
The impedance of each pulse is low (50 ohm or less) and inverse to size of discharge (pC). When done in a cup of oil with electrodes with up to 60kV t he voltage is 10V into 50 Ohms or 2 Watts but only for 10ns. But when immer sed inside 5000 liters Of transformer oil, the signal gets shunted by the s urrounding dielectric (in 3D), unless it takes a direct path to tank wall g round. Normally it is an interior conductor and hard to locate.
Doppler Imaging
1D methods.. Time of arrival using ultrasound pulse,with RF pulse and wavel ets using DSP 2D and 3D methods using multiple sensors have limited sensitivity.Better methods?
Any experience? I am sure it is similar to Plasma Physics detectors, but no t my expertise.
background
------------ Pure oil only shows PD activity just before Breakdown thresholds, whereas contaminated oil can be drastically reduce PDIV to 50% of the breakdown spa rk voltage. It doesn't take much and since they are