modeling and simulation of non linear magnetic cores at high frequencies using PSpice

摘要

Using PSpice of MiscroSim to simulate power converters sometimes brings awkward complications such as the simulation of real magnetic components. It deals a little bit of work to bring a good agreement between experimental and simulated results. Simulation with linear inductors or transformers does not produce good results. The reason is that real magnetic components present saturation and hysteresis, a phenomenon that characterizes magnetic materials. The hysteresis or B-H loop of a magnetic component, shown in the Fig. below, is highly dependent on frequency, waveform (not very important for ferrite material), temperature, excitation level, etc. Although PSpice includes in its magnetic library a list of non-linear components, rarely these parameters are coincident with our working conditions or simply they are not specified. In all of these cases it is necessary to model the non-linear magnetic cores in our particular application. This paper presents a systematic procedure that permits one to obtain the main PSpice parameters to model magnetic cores using the Jiles and Atherton theory. The procedure is totally general and models both the static B-H loop given by the manufacturer data sheet as well as the one that may be obtained in accordance with the specific operating conditions of an electronic system. This is really important, since it permits the user to define the model parameters for high frequency applications, which are not completely correct in the PSpice library. All this is done in a simple way by using the experimental measurement of the coercivity (H{sub}c) remanence (M{sub}r), saturation magnetization (M{sub}s), maximum differential susceptibility X{sub}(max) differential susceptibility at the remanence point (X{sub}r), the coordinates of the loop tip (H{sub}m, M{sub}m) and the slope at this point (X{sub}m). With this procedure a good agreement between experimental and simulated B-H loop is obtained, as shown in the Figure.