Evaluation of steel shafts for magnetostrictive torque sensors (abstract)

Abstract

Based on the magnetostrictive effect in steel, a robust, noncontacting shaft torque sensor can be obtained. A fundamental problem in this scheme is compatibility between mechanical strength required for a shaft and a magnetic one needed for a torque sensor. In order to find shaft material accommodating these two requirements, we investigated basic characteristics, such as hysteresis, linearity, and zero-level fluctuation associated with shaft rotation, of the magnetostrictive torque sensor for various nickel chromium molybdenum steel shafts (SNCM in Japanese Industrial Standard) subjected to case hardening. We prepared three kinds of shafts of 25 mm in diameter: SNCM 420 (Ni=1.69%, C=0.2%), SNCM 616 (Ni=2.91%, C=0.15%), and SNCM 447 (Ni=1.67%, C=0.49%). Shafts of the first two materials were carburized, whereas those of the last one were quenched. We used a magnetic head-type torque sensor consisting of a pair of cross-coupled figure-eight coils (14 turn), which has been already proposed by us.1 The hysteresis in the input–output relationship was measured for the excitation current from 0.1 to 1.0 A at the excitation frequency of 60 kHz. The hysteresis of the SNCM 420 shaft changes sign from negative to positive with the increase in excitation current and that of the SNCM 616 shaft decreases monotonically but never reaches zero, whereas that of the SNCM 447 shaft exhibits minimum. The smallest values obtained are nearly zero for the SNCM 420 shaft at 0.3 A, 1.5%/(full scale (FS)=400 Nm) for the SNCM 616 shaft at 1.0 A and 0.7%/FS for the SNCM 447 shaft at 0.8 A, respectively. The linearity measured for the SNCM 420 shaft, which has the smallest hysteresis of the three, at 0.3 A and 60 kHz was virtually straight for the applied torque range −400–400 Nm and 0.8% of nonlinearity error for the range −1000–1000 Nm. The zero-level fluctuation was measured for the SNCM 420 shaft by rotating the shaft without applying torque. The level of the fluctuation was as small as ±1.8%/FS (FS=1000 Nm). It is interesting to consider the effect of carburization on the performance of the shaft seen from the torque sensor. The sensitivity of the as-ground SNCM 420 shaft is approximately ten times larger than that of carburized shafts, however, the input–output relationship measured for the as-ground shaft was saturated completely at the applied torque of 400 Nm. The zero-level fluctuation for the as-ground shaft was about the same in terms of %/FS as that of carburized shaft. Considering that the reduction of the sensitivity by carburization is a factor of 10, magnetic inhomogeneities at the surface are well improved by the carburizing process. We measured the two-dimensional pattern of magnetic anisotropy distribution by scanning the shaft surface using a 6.5 mm square eddy current probe that is sensitive to magnetic anisotropy.2 We found that the as-ground shaft has a peculiar magnetic anisotropy pattern, which is probably induced during manufacturing steel bars. The pattern was erased by the heat treatment at 920 °C for 4 h, which is a main course of carburization treatment.