Download PDF

Ieee Transactions On Power Electronics

Publication date: 2019-02-01
Volume: 34 Pages: 1683 - 1695
Publisher: Institute of Electrical and Electronics Engineers

Author:

Rasilo, Paavo
Martinez, Wilmar ; Fujisaki, Keisuke ; Kyyra, Jorma ; Ruderman, Alex

Keywords:

Science & Technology, Technology, Engineering, Electrical & Electronic, Engineering, Eddy currents, hysteresis, inductors, inverters, pulsewidth modulation, LOSS PREDICTION, FREQUENCY, IMPLEMENTATION, CONVERTER, DESIGN, STG/18/004#54786148, 0906 Electrical and Electronic Engineering, Electrical & Electronic Engineering, 4008 Electrical engineering, 4009 Electronics, sensors and digital hardware

Abstract:

© 2018 IEEE. A new implementation of an iron-loss model for laminated magnetic cores in the MATLAB/Simulink environment is proposed in this paper. The model is based on numerically solving a one-dimensional diffusion problem for the eddy currents in the core lamination and applying an accurate hysteresis model as the constitutive law. An excess loss model is also considered. The model is identified merely based on the catalog data provided by the core material manufacturer. The implementation is validated with analytical and finite-element models and experimentally in the case of a toroidal inductor supplied from a GaN FET full-bridge inverter with 5-500 kHz switching frequencies and a deadtime of 300 ns. Despite the simple identification, a good correspondence is observed between the simulated and measured iron losses, the average difference being 3.3% over the wide switching frequency range. It is shown that accounting for the skin effect in the laminations is significant, in order to correctly model the iron losses at different switching frequencies. Some differences between the measured and simulated results at high switching frequencies are also discussed. The model is concluded to be applicable for designing and analyzing laminated magnetic cores in combination with power-electronics circuits. The Simulink models are openly available.