DYNAMIC MODELING OF THE SERIES RESONANT DC-DC CONVERTER

The series resonant DC-DC converter is used to process power from one voltage to another with input-output isolation. Frequently negative feedback is applied to stabilize the output voltage. As a switching converter it has the advantages of very high power efficiency and natural commutation of its switches. It would be useful to dynamically model the converter; however, conventional methods of computing the dynamic response of a linear circuit fail because of the switches which periodically reconfigure the converter's internal elements. Presently ad hoc methods are used in designing the feedback; therefore there is a need for a dynamic model of this converter.;These results were experimentally verified in three ways: The discrete system simulation was compared with a breadboard converter using large disturbances, the steady state solution was compared with previously published solutions, and the small signal frequency response was also verified with the breadboard converter. Excellent agreement was obtained in each case.;Using these results, a feedback system involving a series resonant converter can be designed for local stability about a steady state operating point. Equally useful, a proposed converter could be simulated under a variety of conditions using a simple, fast-running computer program. In view of the high power levels at which converters are being built (100KW), these results provide very useful design tools.;This problem was approached by solving each of the switching intervals for its endpoint values in terms of its initial conditions. Because the endpoint values of an interval will be the initial conditions of the next, the converter can be naturally described as a nonlinear discrete system. A unique definition of the state variables resulted in a simple nonlinear discrete state space formulation of the converter's dynamics. This discrete system provides an accurate large signal simulation of the converter's long term dynamic behavior, and it uses very few computational steps, conserving computer time. This discrete system was also solved for the steady state solution, and was converted to a linear transfer matrix (using a small signal approximation.) The transfer matrix was then converted to steady state frequency responses.