| Abstract |
Progress in material technology in wide-bandgap semiconductors, such as Silicon Carbide (SiC) enables increasing the operation frequency of high-voltage (HV) generators and using less series connected devices for the same voltage rating of the HV-generators, e.g. for x-ray applications. These generators are frequently equipped with HV multipliers, such as the Cockcroft-Walton cascade, a resonant inverter and a high voltage transformer. Power conversion at high frequency usually helps reducing the system size and cost and allows better efficiency at smaller size and lower weight. At high frequency and high voltage, however, the influence of parasitic effects of the multiplier, namely parasitic capacitance, becomes highly relevant to the feeding circuit. Experience from existing circuits indicates, that the total parasitic capacitance at secondary side is not only related to the HV transformer and the junction capacitance of the diodes, but also to the extension of the electric AC field in the rectifier circuit, which reflects the structural parasitic capacitance. It can happen that a substantial amount of driving current is required only for inverting the voltage across the parasitic capacitance, which puts a limit on the maximum useful frequency. In general it is to be expected that parasitic capacitance will increase with denser packaging of the components. In this paper an approach for the determination of the equivalent structural parasitic capacitance is given that is based on a 3D finite element (FE) analysis of the energy in the electric AC fields in the cascade circuit. The method is subsequently used to investigate effect of material technologies on the equivalent structural parasitical capacitance. These investigated technologies are: 1) the effect of the increased blocking voltage and associated reduced number of series connected devices, 2) the effect of the permittivity of isolating media and 3) the effect of a grounded box around the rectifier. The capacitance obtained by field simulation is validated by small signal measurements. |
