Electronic Thesis and Dissertation Repository


Doctor of Philosophy


Electrical and Computer Engineering


Dr. Robert Sobot

2nd Supervisor

Dr. Tarlochan S. Sidhu

Joint Supervisor


Advancements in relaying algorithms have led to an accurate and robust protection system widely used in power distribution. However, in low power sections of relaying systems, standard voltage and current measurement techniques are still used. These techniques have disadvantages like higher cost, size, electromagnetic interference, resistive losses and measurement errors and hence provide a number of opportunities for improvement and integration. We present a novel microsystem methodology to sense low-power voltage and current signals and detect disturbances in high-voltage power distribution lines. The system employs dual sensor architecture that consists of a piezoelectric transformer in combination with Hall-effect sensor, used to detect the disturbances whose harmonics are in the kHz frequency range.

Our numerical analysis is based on three-dimensional finite element models of the piezoelectric transformer (PT) and the principle of Hall-effect based “Integrated Magnetic Concentrator (IMC)” sensor. This model is verified by using experimental data recorded in the resonant frequency and low frequency regions of operation of PT for voltage sensing. Actual measurements with the commercial IMC sensor too validate the modelling results.

These results describe a characteristic low frequency behaviour of rectangular piezoelectric transformer, which enables it to withstand voltages as high as 150V. In the frequency range of 10Hz to 250Hz, the PT steps down 10-150V input with a linearity of ±1%. The recorded group delay data shows that propagation delay through PT reduces to few microseconds above 1kHz input signal frequency. Similarly, the non-intrusive current sensor detects current with a response time of 8μs and converts the current into corresponding output voltage. These properties, in addition to frequency spectrum of voltage and current input signals, have been used to develop a signal processing and fault detection system for two real-time cases of faults to produce a 6-bit decision logic capable of detecting various types of line disturbances in less than 3ms of delay.