Characterization of inductors in partial saturation for SMPS applications

Abstract

The exploitation of partially saturated inductors in Switching Mode Power Supply (SMPS) is a key strategy for reducing the size of SMPS and increasing their power density [1]. However an accurate characterization and modellization of each inductor is needed for a correct design and control of the SMPS [2]. Most of inductors’ datasheets only indicate the nominal inductance at low current. Anyway, when the inductor is subjected to a higher current, the value of the inductance is lessened. Moreover, inductance meters often actuate their measurements at no or moderate DC bias. In this work we describe the realization of an automated test bench (Fig. 1) for characterizing an inductor under variable current levels, moving its operating point form the linear region to the partial saturation region [3]. The core of the measurement setup is a simple buck converter, used for applying a square wave voltage to the inductor under test (Fig. 1). The current flowing through the inductor is measured by a Hall-effect current transducer. In order to measure the inductance at different values of the DC bias current, a constant current sink is placed after the LC filter as a variable load. The measurement is carried out by exploiting the inductor constitutive equation VL = L dIL/dt , rewritten as: L = VL / (dIL/dt). For small current variations, the value of L can be considered constant and depending only on the average current. The DC bias current, the frequency and duty-cycle of the switching element can be controlled by a virtual instrument developed in LabVIEW, which also takes care of data acquisition and post elaboration (Fig.2). Several inductors have been tested using the described setup. Figure 3 shows a typical plot of the inductance L versus the inductor current IL . The inductor under test has a rated value of 470 μH. The partial saturation of the inductor is quite evident. Reproducibility tests have been performed on the system, by comparing the results of several measurements on the same inductor. Standard deviation bars, indicated in the plot, show the acceptable precision of the method, the results being well within 1%. The absolute accuracy has not been tested due to the lack of calibrated reference inductors, however the inductance values measured at low currents are in good agreement with nominal values.