Assessment of novel power electronic converters for drives applications

Abstract

In the last twenty years, industrial and academic research has produced over one hundred new converter topologies for drives applications. Regrettably, most of the published work has been directed towards a single topology, giving an overall impression of a large number of unconnected, competing techniques. To provide insight into this wide ranging subject area, an overview of converter topologies is presented. Each topology is classified according to its mode of operation and a family tree is derived encompassing all converter types. Selected converters in each class are analysed, simulated and key operational characteristics identified. Issues associated with the practical implementation of analysed topologies are discussed in detail. Of all AC-AC conversion techniques, it is concluded that softswitching converter topologies offer the most attractive alternative to the standard hard switched converter in the power range up to 100kW because of their high performance to cost ratio. Of the softswitching converters, resonant dc-link topologies are shown to produce the poorest output performance although they offer the cheapest solution. Auxiliary pole commutated inverters, on the other hand, can achieve levels of performance approaching those of the hard switched topology while retaining the benefits of softswitching. It is concluded that the auxiliary commutated resonant pole inverter (ACPI) topology offers the greatest potential for exploitation in spite of its relatively high capital cost. Experimental results are presented for a 20kW hard switched inverter and an equivalent 20kW ACPI. In each case the converter controller is implanted using a digital signal processor. For the ACPI, a new control scheme, which eliminates the need for switch current and voltage sensors, is implemented. Results show that the ACPI produces lower overall losses when compared to its hardswitching counterpart. In addition, device voltage stress, output dv/dt and levels of high frequency output harmonics are all reduced. Finally, it is concluded that modularisation of the active devices, optimisation of semiconductor design and a reduction in the number of additional sensors through the use of novel control methods, such as those presented, will all play a part in the realisation of an economically viable system.