On the use of finite elements and dynamic programming for prediction of electromagnetic device geometries

The author presents a novel technique for the design of a transformer using a numerical synthesis method which utilizes finite elements and dynamic programming for predicting the core and winding geometries of magnetic circuits. This technique is applicable to both low- and high-frequency designs, as well as in handling the materials´s nonlinearity. The designed transformer was placed in a DC-DC converter circuit used in an energy-producing medical device to verify the numerical design results. This converter was inserted in such a device and was used to boost the voltage from a low-voltage implanted battery to high-voltage energy storage capacitors in a short period of time. The stored energy is then delivered, when needed, through an energy delivery circuit to stimulate of defibrillate the heart. The final transformer design satisfied the performance criteria and provided means for selecting the various converter components. The performance results obtained for the transformer and the complete DC-DC converter circuit were in excellent agreement with laboratory performance tests