ARTICLE IN PRESS

Mechanical Systems and Signal Processing

Mechanical Systems and Signal Processing ] (]]]]) ]]] ]]]
www.elsevier.com/locate/jnlabr/ymssp

Method for generating and measuring the micro-Newton level forces

Yusaku Fujii
Department of Electronic Engineering, Faculty of Engineering, 1-5-1 Tenjin-cho, Kiryu, Gunma 376-8515, Japan Received 19 October 2004; received in revised form 5 January 2005; accepted 11 January 2005

Abstract A method for generating and measuring the micro-Newton level forces is proposed. In the method, the force is generated and measured as the inertial force acting on a mass. A pneumatic linear bearing is used to realize a linear motion with sufciently small friction acting on the mass, i.e. the moving part of the linear bearing. The inertial force acting on the mass is calculated from the velocity of the mass, and the velocity is determined highly accurately by means of measuring the Doppler shift frequency of the laser light beam reecting on the mass using an optical interferometer. To demonstrate the performance of the method, the mechanical properties of a piece of paper have determined with the resolution of 0.5 mN. r 2005 Elsevier Ltd. All rights reserved.
Keywords: Inertial force; Inertial mass; Dynamic force; Pneumatic bearing; Air bearing; Optical interferometer

1. Introduction Recently, the requirements for evaluating small force in the range of 1 mN1 N have increased in various industrial and research applications. However, it is sometimes quite difcult to generate and evaluate small force, properly.
Tel.: +81 277 30 1756; fax: +81 277 30 1707.

E-mail address: fujii@el.gunma-u.ac.jp. 0888-3270/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.ymssp.2005.01.001

ARTICLE IN PRESS
2 Y. Fujii / Mechanical Systems and Signal Processing ] (]]]]) ]]] ]]]

The difculties in measuring small force mainly come from the following facts: (1) Some methods for supporting a direct realisation of micro-Newton level forces linked to the International System of Units (SI) below 1 N [14] are currently developed by some national standard laboratories. However, any method for measuring micro-Newton level forces has not been established even for static forces. (2) Small force to be generated and/or measured is usually a varying force and any dynamic calibration technique for force sensors has not been established [5]. In other words, this fact means that the uncertainty evaluations both for the measured value of the small force and for the time of the measurement are very difcult. Force can be dened as the product of mass and acceleration as F Ma, where F is the force acting on an object, M is the mass of the object, and a is the acceleration of the centre of the gravity of the object. This means that a well-dened acceleration is required to generate forces directly based on the above denition. Acceleration due to gravity, g; is convenient and usually used for generating and/or measuring constant force. Constant force can be accurately compared using a conventional balance with a knife-edge or a hinge. There is an attempt [1,2] for developing a method for generating and measuring static micro-Newton level forces directly linked to SI using the electrostatic force measured in terms of the quantum standard of electrical quantities, in other words, in terms of the recommended value of the Planck constant. The method is a modication of the Watt Balance Method [6], in which the Planck constant is measured by means of comparing the electric power measured in terms of the Josephson effect and the quantum Hall effect with the mechanical power measured in terms of the mass of the international prototype of kilogram and the acceleration of gravity, g: The method [1,2] will be very useful for generating and measuring the static micro-Newton level forces if the unknown uncertainty sources come form the balance mechanism, such as friction and irregularity of alignment of the electrode, are sufciently eliminated. There is another attempt for developing a method for generating and measuring micro-constant force, in which a combination of an electric balance and a lever mechanism is used as a scale [4]. However there are no working methods for calibrating force transducers under dynamic conditions. Only static methods, in which transducers are calibrated by static weighting under static conditions, are widely available at present. Methods for the dynamic calibration of force transducers are important for fullling these requirements. Therefore, it is very difcult to determine the uncertainty in measuring a varying force or dynamic force using force transducers. These force transducers are widely used without any knowledge on the difference between the dynamic characteristics against the static characteristics, as usual. Usually it is widely believed that force transducers with higher resonant frequency are suitable for measuring the dynamic force with higher frequency. However, that is irrational, of course. Although the methods of dynamic calibration of force transducers are not yet well established, there have been a number of trials aiming at establishing the dynamic calibration methods for force transducers.

ARTICLE IN PRESS
Y. Fujii / Mechanical Systems and Signal Processing ] (]]]]) ]]] ]]] 3

One method was proposed by the author and has been under development [4,7,8]. This method, that is the Levitation Mass Method, was rst proposed [4] as an impulse response evaluation method for force transducers; a mass is made to collide with a force transducer and the impulse, i.e. the time integration of the impact force, is measured highly accurately as a change in momentum of the mass. To realize linear motion with sufciently small friction acting on the mass, a pneumatic linear bearing [7] is used, and the velocity of the mass, i.e. the moving part of the bearing, is measured using an optical interferometer. Although the methods are under development, some information has been obtained. For example, as for a piezoelectric force transducer with a high resonance frequency of approximately 200 kHz, calibrated with a relative standard uncertainty of approximately 1% and widely believed for suitable for dynamic force measurements, a difference of a few percent levels between the static and dynamic characteristics was detected under the condition of a steep impact force with a half-value width of approximately 1 ms [8]. The other method, which was proposed and has been developed by Kumme, uses the inertial force of the attached mass generated by a shaker [9]. In this method, dynamic force of a single frequency is generated and applied to a force transducer. This method is effective for evaluating the characteristics of force transducers under the conditions in which calibration is conducted, such as continuous vibration at a single frequency. Park et al. have used this method for the dynamic investigation of multi-component force-moment sensors [10,11]. Summarising the present status of dynamic calibration methods, there are many categories in dynamic force calibration, such as impact force calibration, oscillation force calibration and step force calibration, and they are still under development. In fact, it is still very difcult even for the specialists researching for the dynamic calibration to calibrate force transducers dynamically. The Levitation Mass Method has been applied to the eld of the material testing, such as the dynamic three-point bending test [12], the viscoelasticity test [13], the friction test [14] and the strength test [15]. In this paper, a novel method for directly generating and measuring small dynamic force has been proposed by modifying the Levitation Mass Method. In the method, the small varying force is measured without use of any force transducer.

2. Experimental set-up Fig. 1 shows a schematic diagram of the experimental set-up for generating and evaluating the small impact force applied to a material under test. Fig. 2 shows a photograph around the test section. The material under test is attached to the base. A pneumatic linear bearing is used to realise linear motion with sufciently small friction acting on the mass, i.e., the moving part of the bearing. Impact force is generated and applied to the material by colliding the mass. An initial velocity is given to the moving part manually. A cube-corner prism (CC) for interferometer and a metal block with a round-shaped tip for adjusting the collision position are attached to the moving part made in aluminum with square pole shape; its total mass, M ; is approximately 21.17 g. The inertial force acting on the mass is measured highly accurately using an optical interferometer. A piece of paper is used as the material for test. The piece of paper with adhesive

ARTICLE IN PRESS
4 Y. Fujii / Mechanical Systems and Signal Processing ] (]]]]) ]]] ]]]

Fig. 1. Experimental set-up.

Fig. 2. Photograph of test section.

area Post-it, 7.5 mm in width and 29.0 mm in length, is attached to the base by its adhesion. Moderate curvature is approximately given to the paper as seen in Fig. 2. The force acting on the material from the moving part is equal to the inertial force of the moving part, F inertial Ma according to the law of inertia if other forces, such as the frictional force inside the bearing, can be ignored. In this condition, the force acting on the moving part from the material is the product of mass and acceleration of the moving part, i.e. F Ma: The

ARTICLE IN PRESS
Y. Fujii / Mechanical Systems and Signal Processing ] (]]]]) ]]] ]]] 5

acceleration is calculated from the velocity of the moving part. The velocity is calculated from the measured value of the Doppler shift frequency of the signal beam of a laser interferometer, f Doppler ; which can be expressed as v lair f Doppler =2, f Doppler f beat f rest , where lair is the wavelength of the signal beam under the experimental conditions, f beat is the beat frequency, i.e., the frequency difference between the signal beam and the reference beam, f beat is the rest frequency which is the value of f beat when the moving part is at a standstill, and the direction of the coordinate system for the velocity, the acceleration and the force acting on the moving part is towards the right in Fig. 1. A Zeeman-type two-frequency HeNe laser is used as the light source. The frequency difference between the signal beam and the reference beam, i.e. the beat frequency, f beat ; is measured from an interference fringe which appears at the output port of the interferometer; it varies around f beat ; approximately 2.8 MHz, depending on the velocity of movement. An electric frequency counter (model: R5363; manufactured by Advantest Corp., Japan) continuously measures and records the beat frequency, f beat ; 14,000 times with a sampling interval of T 40000=f beat ; and stores the values in memory. This counter continuously measures the interval time of every 40,000 periods without dead time. The sampling period of the counter is approximately 14 ms at a frequency of 2.8 MHz. Another same-model electric counter measures the rest frequency, f rest ; using the electric signal supplied by a photodiode embedded inside the HeNe laser. The pneumatic linear bearing, GLS08A50/25-2571 (NSK Co., Ltd., Japan), is attached to an adjustable tilting stage. The maximum weight of the moving part is approximately 1 kg the stroke of the movement is approximately 25 mm and the thickness of the air lm is approximately 10 mm. The tilting angle of the tilting stage is roughly adjusted horizontally with the uncertainty of approximately 0.1 mrad. The slope component of the gravitational force acting on the moving part of approximately 200 mN is approximately 0.02 mN at the slope angle of approximately 0.1 mrad. Measurements using the two electric counters are triggered by means of a sharp trigger signal generated using a digital to analog converter. This signal is initiated by means of a light switch, a combination of a laser-diode and a photodiode. In the experiment, one set of collision measurement with three collisions is conducted.

3. Results Fig. 3 shows the data processing procedure in a collision experiment. In the gure, only the rst collision out of three collisions is shown. During the experiment, only the beat frequency, f beat ; and the rest frequency, f rest ; are measured highly accurately using an optical interferometer. The Doppler shit frequency is measured as the difference between the beat frequency and the rest frequency. The velocity, the position, the acceleration and the inertial force of the mass are calculated from the frequency afterward. In the collision experiment shown in Fig. 3, the

ARTICLE IN PRESS
6 Y. Fujii / Mechanical Systems and Signal Processing ] (]]]]) ]]] ]]]

Fig. 3. Data processing procedure: calculation of velocity, position, acceleration and force from frequency.

maximum value of the impact force, F max ; is approximately 1.8 mN. The origin of the position axis is set to be the position where the reaction force from the material under test is detected. Figs. 47 show all the three collisions of the whole set of the collision measurement. Fig. 4 shows the change in force acting on the mass from the material, F Ma; against time. The force acting on the material from the mass is expressed as F according to the law of action and reaction. As shown in Fig. 4, the moving part collides with the piece of paper 3 times during the measurement. The moving part moves back and forth between the material under test and the lefthand side-face of the bearing holder. The steep negative peaks in the gure represent its collision to the bearing holder. The maximum values of the impact force in the rst, second and third collision are 1.80, 0.89 and 0.23 mN, respectively. The force has some values even during the free

ARTICLE IN PRESS
Y. Fujii / Mechanical Systems and Signal Processing ] (]]]]) ]]] ]]] 7

Fig. 4. Measured force against time.

Fig. 5. Measured velocity against position.

sliding motion of the moving part, x40: The variation of force during the free sliding motion is thought to mainly come from the dynamic frictional force acting inside the bearing. In the experiment, the component of the gravitational force acting on the moving part of approximately 0.02 mN due to the uncertainty of approximately 0.1 mrad in adjusting the tilting stage to the horizontal might exist as explained before.

ARTICLE IN PRESS
8 Y. Fujii / Mechanical Systems and Signal Processing ] (]]]]) ]]] ]]]

Fig. 6. Measured force against velocity.

Fig. 7. Measured force against position.

Fig. 5 shows the change in the measured velocity against the position. During the free motion of the moving part, the effect of the dynamic frictional force acting on the moving part during the free travel is clearly observed. In the experiment, the dynamic frictional force acting inside the bearing is not negligible compared to the inertial force acting during the collision. Fig. 6 shows the change in the measured force against the velocity. The lead of the force against velocity, which is caused by the viscosity of the material under test, is observed. The velocity

ARTICLE IN PRESS
Y. Fujii / Mechanical Systems and Signal Processing ] (]]]]) ]]] ]]] 9

where the force has its maximum value of approximately 1.798 mN is approximately 0.0019 m s1. Fig. 7 shows the change in the force acting on the mass from the material, F Ma; against position. The spring constant, or inclination of the curve, is almost constant during the experiment with three collisions. The elastic hysteresis, which is caused by the viscosity of the material, is clearly observed. The R work done by the moving part is expressed as the integral along the trajectory of motion, W F dx: The work, W ; for the rst, second and third collision measurements, is calculated to be approximately 0.815, 0.257 and 0.023 mJ, respectively. The absolute value of this work is equal to the area bounded by the curve shown in Fig. 7. This value agrees well with the reduction of the kinetic energy of approximately 0.815, 0.256 and 0.023 mJ, calculated using the velocity before and after the collision, v1 and v2 : The energy dissipation ratio, W =E 1 W =1=2mv2 1 ; for the rst, second and third collision measurements, is approximately 0.22 (22%), 0.23 (23%) and 0.25 (25%), respectively.

4. Discussions The resolution of the frequency measurement is 1 Hz and the sampling interval is approximately 14 ms. It corresponds to the resolution of the force measured in the experiment of approximately 0.5 mN. The proposed method can be applied to the dynamic calibration of force transducers with very small capacity and the dynamic material tests with very small load. In the proposed method, only the frequency is measured during the oscillation experiment, and all the other quantities, such as velocity, position, acceleration and force, are numerically calculated afterward. This results in the good synchronisation between the obtained quantities. In addition, force is directly calculated according to its denition, that is the product of mass and acceleration. As for a material test method, the simplicity is the most signicant advantage of the proposed method compared with other conventional methods using a force transducer and a position sensor. The set-up procedure for a new test is very easy in the proposed method. Any object, such as a force transducer, a viscoelastic material or a specimen with complicated structure, can be attached to the base using an appropriate adhesive material or a mechanical holder. Only one side of the material must be xed to the base with an appropriate bonding agent. Then its dynamic response against impact force is evaluated highly accurately by means of measuring the frequency of the laser light. At this moment, the uncertainty evaluation of the measured force has not been done. For this, the frictional characteristics of the air bearing are required. To determine them, the method [7] developed for evaluating the frictional characteristics of the larger size air bearing with the moving part of approximately 2 kg can be used.

5. Conclusions A method for generating and measuring the micro-Newton level forces is proposed. In the method, a mass levitated with sufciently small friction using a pneumatic linear

ARTICLE IN PRESS
10 Y. Fujii / Mechanical Systems and Signal Processing ] (]]]]) ]]] ]]]

bearing is made to collide with a material under test. During the collision measurement, only the Doppler shift frequency of the laser light beam reecting on the mass is measured highly accurately using an optical interferometer. The velocity, the position, the acceleration and the inertial force of the mass are calculated from the frequency afterward. The impact response of the piece of paper is highly accurately determined by means of the proposed method. Any small object, such as a force transducer, a viscoelastic material or a specimen with complicated and weak structure, can be attached to the base using an appropriate adhesive material or a mechanical holder. Only one side of the test specimen must be xed to the base with an appropriate bonding agent. Acknowledgements This work was supported by a research-aid fund of The Asahi Glass Foundation. The author would like to express his gratitude to the reviewer for the fruitful suggestions.

References
[1] D.B. Newell, J.A. Kramar, J.R. Pratt, D.T. Smith, E.R. Williams, The NIST microforce realization and measurement project, IEEE Transactions on Instrumentation and Measurement 52 (2) (2003) 508511. ` me International dUnite s [2] J.R. Pratt, D.T. Smith, D.B. Newell, J.A. Kramar, E. Whitentonrogress, Toward Syste traceable force metrology for nanomechanics, Journal of Material Research 19 (1) (2004) 366379. [3] P.J. Cumpson, J. Hedley, Accurate analytical measurements in the atomic force microscope: a microfabricated spring constant standard potentially traceable to the SI, Nanotechnology 14 (12) (2003) 12791288. [4] Y. Fujii, H. Fujimoto, Proposal for an impulse response evaluation method for force transducers, Measurement Science and Technology 10 (4) (1999) N31N33. [5] E.R. Williams, R.L. Steiner, D.B. Newell, P.T. Olsen, Accurate measurement of the Planck constant, Physical Review Letters 81 (12) (1998) 24042407. [6] S. Niehe, A new force measuring facility for the range of 10 mN10 N, Proceedings of XVII IMEKO World Congress, Dubrovnik, Croatia, June 2003. [7] Y. Fujii, Measurement of force acting on a moving part of a pneumatic linear bearing, Review of Scientic Instruments 74 (2003) 31373141. [8] Y. Fujii, Measurement of steep impulse response of a force transducer, Measurement Science & Technology 14 (2003) 6569. [9] R. Kumme, Investigation of the comparison method for the dynamic calibration of force transducers, Measurement 23 (1998) 239245. [10] Y.-K. Park, R. Kumme, D.-I. Kang, Dynamic investigation of a three-component force-moment sensor, Measurement Science & Technology 13 (2002) 654659. [11] Y.-K. Park, R. Kumme, D.-I. Kang, Dynamic investigation of a binocular six-component force-moment sensor, Measurement Science & Technology 13 (2002) 13111318. [12] Y. Fujii, D. Isobe, S. Saito, H. Fujimoto, Y. Miki, A method for determining the impact force in crash testing, Mechanical Systems and Signal Processing 14 (6) (2000) 959965. [13] Y. Fujii, T. Yamaguchi, Method for evaluating material viscoelasticity, Review of Scientic Instruments 75 (1) (2004) 119123. [14] Y. Fujii, T. Yamaguchi, Optical method for evaluating material friction, Measurement Science and Technology 15 (10) (2004) 19711976. [15] T. Suzuki, Y. Fujii, J.D.R. Valera, Optical method for strength test for general industrial products, Mechanical Systems and Signal Processing, in press.