Measurement System for a Magnetostrictive Torque Sensor

Measurement system for a magnetostrictive torque sensor A. Ktena, C. Manasis, C. Papadopoulos, D. Bargiotas, O. Ladoukakis, K. Ziatakis, I. Valsamis, F. Magkafas Department of Electrical Engineering TEI of Chalkida Psachna, Evia, GR34400, GREECE {aktena, manasis, cpap, bargiotas}@teihal.gr Abstract— The experimental set up for a magnetostrictive sensor made of NiFe thin film electrodeposited on a cylindrical non magnetic 316L stainless steel rod is presented. The sensor operates inside an AC magnetic field generated by an excitation coil. An amplifier is used to ensure saturation levels in the samples' magnetization. The resulting magnetization is picked up by a sensing coil wound around it in the form of a voltage waveform which is in turn being processed by our LabView® based software. The torque is applied by a specially designed apparatus allowing up to 150 Nm. Preliminary results are presented. Keywords- magnetostriction; NiFe thin films; magnetic torque measurement; electrodeposition

I.

INTRODUCTION

Torque measurement in rotating shafts has many industrial applications as in the automobile industry, aircrafts, machinery, spring testing as well as in biodynamic measurements. Standards for quality management systems require torque measurements during the manufacturing process, especially for parts such as screws and other assembly components [1]. Torque measurement is also a means to determine the output power and vibration levels of an engine, a turbine or other rotating devices. tension (increasing permeability)

applied torque

compression (decreasing permeability)

Figure 1. Shaft under torque

Available commercial torque sensors fall under two major categories: mechanical and magnetic. The surface of a rotating shaft is subjected to two opposing mechanical stresses, compression and tension, as shown on Fig.

John Petrou, Chris Petridis Laboratory of Physical Metallurgy National Technical University of Athens Zografou Campus, Athens 15780, Greece

1. A mechanical torque sensor typically employs a pair of strain gauges positioned on the rotating surface in such a way that one measures the elongation along the tension direction and the other one the decrease in length along the compression direction. These sensors can have an accuracy of ±0.2% full scale and are a popular choice for low speed rotating motion. If the space around the shaft and the centrifugal forces allow it, both sensing elements and their electronics are placed on the shaft. Another kind of mechanical torque sensor uses proximity and displacement sensors in order to measure the angular displacement between the two edges of the shaft. The accuracy of this type of sensors can be as high as ±0.1% full scale but are more expensive than strain gauges. This is the technology of choice when a high resolution measurement of the angle of twist is needed, being appropriate for high speed, high temperature applications. A pair of proximity or displacement sensors measures the relative positions of the shaft's edges producing two respective output voltages. The phase difference between the two voltages varies with the applied torque. The magnetic torque sensors are less accurate, with accuracy in the order of ±1%, and fall under two major categories. The first category includes torque sensors that are basically magnetic displacement sensors measuring the small angular displacements, caused by the rotation, employing a magnetic circuit [1]. This kind of arrangement is quite complex and requires space which is not always available. The second type of magnetic torque sensors are based on the effect of torque on the magnetic properties of the rotating shaft's material, such as its magnetic permeability, μ [2-4]. A magnetostrictive sensing core along with excitation and sensing coils comprise the sensing element. The sensing core practically consists of a thin, magnetoelastic ring, being placed either directly on the shaft or on a ring around it. The magnetoelastic torque sensors measure the change in the shaft's permeability by measuring the change in their own magnetic field, due to the tensile/compressive stress effect on the magnetization loop of the used magnetoelastic sensing core. This type of sensor is more complex than the previous one but relatively cheap and appropriate for applications where accuracy of measurement is not very important.

The proposed sensor consists of a magnetostrictive sensing element electrodeposited on the rotating shaft allowing the measurement of the change in magnetic properties not of the shaft but of the sensing element which is subjected to the same torsion as the shaft. II.

THE PROPOSED SENSOR

Magnetostriction is a ferromagnetic material property where the material expands or contracts in the direction of an applied magnetic field due to the non-180o magnetic domain wall displacement or the rotation of magnetic domains. Inversely, the mechanical expansion or contraction may give rise to magnetization change that can be sensed by a sensing coil as voltage. Related to magnetostriction is the Wiedemann effect [5] where, when an ac-current carrying wire is placed inside an axial magnetic field, a helicoidal magnetization is picked-up as output voltage at the ends of a sensing coil, simulating the effect of torsion to the wire. The inverse (IWE) has been also utilized in sensing arrangements [6] where the ac-current carrying wire is under torsion giving rise to helicoidal susceptibility at the surface of the wire. Torque sensors based on the torque magnetostrictive wires or ribbons have been employing the Wiedemann effect in one way or the other are being described or overviewed in [6, 7]. The magnetostrictive effect causes the permeability of the material to increase under tensile stress and decrease under compressive stress (Fig. 1). In the sensor proposed in this work, the torque on a rotating shaft is expected to affect the effective anisotropy and the magnetic permeability of the magnetostrictive thin film deposited on it. This change can be detected as a change at the output voltage peak at the ends of a sensing coil. The degree of permeability change depends among other things on the so called magnetoelastic coupling factor of the used sensing core. Magnetostrictive sensing materials include transition metals such as iron, nickel, and cobalt and their alloys. The material of choice in our study is NiFe because of the excellent magnetostrictive properties of Ni [8] and the considerable magnetic moment of Fe. Also NiFe can be easily electrodeposited which is the selected method for the growing of a thin film on the shaft [9-10].

For a torque measurement, the structural uniformity of the NiFe film is very important [11]. Conventional electrodeposition devices typically result in non uniformities and microcracks on the surface of the thin film due to tensions built in during the electrodeposistion process [8-9]. In order to attain better control on the sample preparation process, the following parameters must be controlled: the current density, the bath's temperature for higher solubility and the velocity of revolution of the bath for higher uniformity in the NiFe deposition. For this, a new electrodeposition device has been designed and constructed consisting of: i) a cell that can hold 250ml of solution and is portable, light, waterproof, and can be easily dismantled to allow for cleaning and maintenance and ii) an electrical circuit that provides for the heating of the solution via a heating resistor, offers temperature control via a thermocouple, and rotation and speed control via an inverter controlled motor [12]. In order to select the optimal NiFe composition for the torque measurement, twenty four samples of varying stoichiometry, with Fe content varying from 0% to 100%, have been prepared. Each stoichiometry has been used to prepare samples with 3 and 4 min deposition time in order to compare between different thicknesses for the same stoichiometry. IV.

EXPERIMENTAL SET UP

To measure the magnetic properties of the samples we constructed an inductive ac magnetometer mounted in such a way as to allow for measurements with and without the application of torque. The arrangement shown on Fig. 2 consists of an excitation coil inside which the steel rod with the NiFe film on it is inserted. The pick up coil is wound around a plastic cylinder and placed around the sample inside the excitation coil in such a way as to sense the sample's magnetization at its center where the flux lines are more uniform. Using a NI-6251 multifunction card, the output signal is sent to a PC for further processing using LabView®. The excitation coil consists of 1500 turns of 1 mm diameter enameled copper wire wound on a 20 cm long plastic cylinder of 4 cm diameter. It can generate a magnetic field in the order of 100 kA/m at its center.

A straightforward advantage of this arrangement is that it can be used to measure torque even on non magnetic shafts which simplifies the design. Also, electrodeposition is a low cost industrial process and can be easily integrated in the manufacturing process. III.

SAMPLE PREPARATION

Austenitic stainless steel 316L (USA AISI) rods of 10mm diameter 30 cm long were chosen as substrates for the sample preparation. This material is non magnetic and has a high tolerance to corrosion by chlorine solutions like the one used in the electrodeposition process. The electrodeposited NiFe thin films are 10 cm long and have a thickness in the order of several microns.

Figure 2. Experimental setup (a) sine wave generator, (b) amplifier, (c) magnetometer: (1) exciting coil, (2) receiving coil, (3) test object, (4) NiFe thin film, (d) data acquisition card, (e) computer

The sensing coil is wound on a 2 cm diameter plastic tube using 2000 turns of 0.1mm diameter enameled copper wire. The cylindrical sample is mounted in such a way as to ensure that the measurement is taken at the center of the NiFe thin film and therefore ensure the repeatability of the measurement.

The actual excitation coil with a sample and its mounting board is shown on the photo (Fig. 3).

this setup. Measurements were carried out at various frequencies, between 50Hz and 500 Hz.

Figure 3. The excitation coil and the sample holder

An AC sinusoidal current is fed to the excitation coil by a signal generator connected to a power amplifier. Initially, the characterization of the samples was carried out using just the generator's signal which could be as high as 100 mA. However, such current levels were not enough to saturate most of the samples while several stoichiometries were hardly magnetized. A power amplifier based on the TI OPA580 was built and sample characterization was repeated greatly enhancing the system's response. V.

RESULTS AND DISCUSSION

The AC magnetic field generated by the current through the excitation coil's turns magnetizes the magnetostrictive NiFe film inducing a voltage at the sensing coil. The magnetic induction inside the sensing coil consists of the response from to the input magnetic field H(t) and the resulting magnetization M(t): B(t)=μ0·[H(t)+M(t)], where μ0 is the magnetic permeability of free space. The voltage response according to Faraday's law of induction is given by

V ( t ) = − NA

dB ( t ) dt

⎛ dH ( t ) dM ( t ) ⎞ = −μ 0 NA ⎜ + ⎟ (1) dt ⎠ ⎝ dt

where A is the sensing coil's cross section and N the number of its turns. From Eq. 1, it can be seen that the voltage output consists of one component which is due to the excitation field H(t) and another component which is due to the presence of the magnetic substance with magnetization M(t) inside the excitation coil. Fig. 4 shows a LabView® screen capture for a sample measurement. The top waveform is the output voltage obtained at the sensing coil V(t) which consists of a sinusoid dependent on H(t) modulated by voltage pulse due to M(t). The bottom waveform is obtained by substracting the sinusoid and therefore reflects the voltage due to the sample magnetization M(t). Applied torque affects the shape and the peak of this voltage pulse. First, all samples were measured with zero applied torque in order to classify them as to their magnetization response in

Figure 4. The output voltage detected at the pick-up coils edges (top) and the voltage pulse obtained after the subtraction of the sinusoidal carrier (bottom)

Fig. 5 shows the output voltage peak values for Ni40Fe60 before and after the input signal amplification as a function of frequency. This type of response is typical for all of our samples. Most samples started giving measurable output at frequencies as low as 20 Hz and would go up to 450 Hz. The voltage output was stable for frequencies around 100 ± 20 Hz for all samples with Ni70Fe30 demonstrating a constant output of 0.22 V from 50 to 200 Hz and Ni30Fe70 0.38 V from 70 to 150 Hz.

Figure 5. The voltage peak output in V as a function of frequency for Ni30Fe70

Fig. 6 shows the effect of deposition time on the frequency response for Ni20Fe80 which yielded the highest response. Its response is typical for all the samples that yielded measureable output. The increase in thickness significantly enhances, up to 100%, the magnetization response since there is more magnetic material involved in the measurement. The two curves increase with frequency from 35 Hz up to around 100 Hz where they reach a maximum value or a plateau region and

Figure 7. The voltage peak output in V as a function of % Ni-content

VI.

CONCLUSION

The experimental and measurement set-up for a magnetostrictive torque sensor has been presented in this paper. NiFe thin film samples were deposited on a stainless steel cylindrical rod using an in-house made electrodeposition device with temperature and rotation control for higher film uniformity. A torque measurement apparatus has also been constructed based on AC magnetometry. All samples were characterized as to their hysteresis loop with respect to film thickness and frequency of the excitation field. Preliminary results indicate that samples must have at least 60% Fe content, the optimum frequency response occurs in the neighborhood of 100 Hz and the output signal is higher than 0.2 V even for samples with 3 minute deposition time. Torque measurements are under way while structural characterization is deemed necessary in order to better understand the effect of the deposition parameters and control the NiFe film’s properties.

decrease down 400 and 450 Hz respectively. ACKNOWLEDGEMENTS Figure 6. The effect of deposition time (film thickness) on the frequency response and the output voltage peak for Ni20Fe80

Figure 7 summarizes the results on the voltage peaks as a function of stoichiometry for 3 and 4 minute deposition times, before and after input signal amplification. As stated already, the longer the deposition time the higher the voltage output but also the trend is similar regardless of the thickness of the NiFe film. The measurements taken with the amplified input signal indicate that the stronger current was indeed necessary in order to saturate the samples. More conclusions on the effect of thickness on the magnetic properties and microstructure of the NiFe film will be drawn once the torque measurements are completed and TEM studies are carried out. As expected, samples with higher than 50% Ni content hardly gave any response since Ni is a very soft magnetic material. Once Fe content became more prominent, the reponse was greatly enhanced yielding output voltages roughly between 0.4 and 1V, a very decent output signal that does not require further processing or amplification. However, the increase in Fe content is expected to reduce the magnetostrictive properties of the samples and therefore its sensitivity as torque sensing material.

This work has been carried out as part of the project ARCHIMEDES II jointly funded by the European Community Fund (75%) and the Hellenic Ministry of Education(25%). REFERENCES [1] [2] [3] [4] [5]

[6]

[7] [8]

[9]

[10]

[11]

[12]

http://www.omega.com/literature/transactions/volume3/force3.html Y. Chen, J.E. Snyder, C.R. Scwichtenberg, K.W. Dennis, R.W. McCallum, D. C. Jiles, IEEE Trans. Magn. 35, 3652-3654 (1999) F. Umbach, H. Acker, J. von Kluge, W. Langheinrich, “Contactless measurement of torque”, Mechatronics, 12, 1023-1033 (2002) H. Wakiwaka, M. Mitamura, “New magnetostrictive type torque sensor for steering shaft”, Sensors & Actuators A, 91, 103-106 (2001) Aphrodite Ktena, Evangelos Hristoforou, Magnetic Effects in Sensing Applications, Encyclopedia of Sensors, ed. C.A.. Grimes et. al., Vol. 10, pp. 1-70 (2006). E. Hristoforou, Magnetostrictive delay lines: Engineering Theory and Sensing Applications, Review Paper, Meas. Sci. & Tech., Meas. Sci. Technol. 14 No 2, p. R15-R47 (2003) M Tejedor et al, Magnetoelastic torque magnetometer using the inverse Wiedemann effect, 1984 J. Phys. E: Sci. Instrum. 17 869-871 I. Giouroudi, C. Orfanidou, E. Hristoforou, “Circumferentially oriented Ni cylindrical thin films for torque sensor applications”, Sensors & Actuators A, 106, 179-182 (2003) I. Giouroudi, A. Ktena, E. Hristoforou, “Microstructural Characterization of Cylindrical Fe1-xNix Thin Films”, Journal of Optoelectronics and Advanced Materials, 6(2), 661-666 (2004) C. Petridis, A. Ktena, E. Hristoforou, “Ni-Fe circumferential thin film covered Cu wires for field sensing applications”, Sens Lett 5 (1), 93-97 (2007) C. Petridis, A. Ktena, E. Hristoforou, “On the magnetic and magnetoelastic uniformity measurements on magnetostrictive ribbons and wires”, JMMM (2007) O. Ladoukakis, Device for controlled electrodeposition on cylindrical rods, Technical Report, ARCHIMEDES II – Subproject 3, TEI of Chalkida, 2007.