Load characterization in high-frequency IPT systems using class EF switching waveforms

Abstract

Large magnetic field volumes associated with core-less high-frequency inductive power transfer (HF-IPT) systems allow multiple receivers to be powered from one transmitter, but also provide greater probability for foreign objects to couple to the system. Knowledge of the types of objects (legitimate receivers or otherwise) that are coupled to the transmitter is critical. Such knowledge on the transmit side would allow the system to be deactivated in the presence of foreign objects, and to determine the exact state of tuning of the receivers so that it may adjust itself accordingly to optimize system performance. This article introduces a technique to calculate the induced voltage generated by coupled receivers and foreign objects on the transmit coil in real time. Changes in the position or electrical quantities of the receivers, and foreign objects, alter the induced voltage on the transmit coil, and with it the trajectory of the switching waveforms of the inverter driving the transmit coil. From the shape of these waveforms, information on the phase and amplitude of the induced voltage can be extracted, thus enabling the induced voltage on the primary to be estimated with a single, easy to access, voltage measurement, which is easier than estimating the induced voltage from measurements of coil current and total coil voltage. We used a support-vector-machine (SVM) to perform regression analysis on the drain voltage data. The experimental setup uses a 100 W, 13.56 MHz Class EF inverter, and the model was generated from a large number of samples of the drain voltage waveforms operating at different known loads. These were generated from our in-house HF-IPT test load, which uses a Class EF synchronous rectifier. The results allow the induced voltage on the transmit coil to be estimated in real time from the drain voltage waveform alone, with a normalized root mean square error of 1.1% for the real part (reflected resistance) and 1.2% for the imaginary part (reflected reactance). This article is accompanied by a video file demonstrating the experiments.