You are on page 1of 6

Design and Experiment of a 3DOF Parallel Micro-Mechanism Utilizing Flexure Hinges

Byung-Ju Yi 1, Heung-Youl Na 2, Goo Bong Chung1, Whee Kuk Kim3, Il Hong Suh1
1

School of Electrical Engineering and Computer Science, Hanyang Univ. Korea 2 Micro-Inspection co. Ansan, Kyungki-do 425-791, Korea 3 Department of Control and Instrumentation Eng., Korea Univ. Korea bj@hanyang.ac.kr that must be specified in order to locate the elements of an object relative to another. It is described by [7]
M = N ( L 1) ( N Fi ) ,
i =1 j

Abstract-- Flexure hinge has been commonly used as a substitute for mechanical joints in the design of micropositioning mechanisms. However, inaccurate modeling of flexure hinges deteriorates the positioning accuracy. In this paper, a planar 3DOF parallel-type micro-positioning mechanism is designed with the intention of accurate flexure hinge modeling. For this, a preliminary kinematic analysis that includes inverse kinematics, internal kinematics, and analytic stiffness modeling referenced to the task coordinate is presented. First, the revolute type of 1DOF flexure hinge is considered. The simulation result based on FEM, however, is not coincident to the analytic result. This is due to the minor axial elongation along the link direction that keeps the mechanism from precise positioning. To cope with this problem, a 2DOF flexure hinge model that includes this additional motion degree as a prismatic joint is employed in part. On the basis of this model, the positional accuracy is ensured. The effectiveness of this accurate model is shown through both simulation and experimentation This work emphasizes that the precise modeling of a flexure hinge is significant to guarantee the positional accuracy of parallel micro-mechanisms using flexure hinge.
I.

(1)

where, M, N, L, j, and Fi denote the mobility, the degree-offreedom of the operational (or task) space, the number of links, the number of joints, and the motion degree-of-freedom of the ith joint, respectively. Mobility also represents the minimum number of the system actuator. When M is greater than N , the system is called a kinematically redundant system.

INTRODUCTION Fig.1. Schematic of the 3DOF Micro-parallel manipulator

Micro positioning mechanism is a key and essential technology in many fields, such as Scanning Electron Microscopy (SEM), x-ray lithography, mask alignment and micro machining. Recently, there have been quite a number of studies on the analysis and design of micro positioning mechanisms with flexure hinges. Most of previous studies (Moriyama, et al. [1], Taniguchi, et al. [2], Tomita, et al. [3], Ryu, et al. [4], Chang, et al. [5], Peng, et al. [6], etc [10]-[13]) modeled the flexure hinge by using only 1DOF (Degree of Freedom) for a revolute hinge and only 3DOF for a spherical hinge. However, the flexure hinge also has translational motion even though the amount is small. Some investigators considered an additional degree of freedom of flexure hinges [6]. However, they omitted mobility analysis in their modeling procedure and also exact stiffness model of micromechanisms employing flexure hinges has not been suggested. This led to the unsuccessful implementation of micro-systems. Therefore, this work will focus on the accurate modeling of flexure hinges that are based on mobility analysis. Mobility is defined as the number of independent variables

Fig.2. Schematic of the Flexure Hinge Figure 1 shows a 3DOF micro positioning mechanism with flexure hinges. Usually, the flexure hinges in this system can be modeled as revolute joints. The shape of the flexure hinge

is depicted in Fig. 2. If all the flexure hinges have 1DOF, the mobility of the system is 3. However, previous results based on this assumption [1-6, 10-13] were unsuccessful because of the inaccurate modeling of the flexure hinges. On the other hand, if all the flexure hinges are modeled as having a revolute joint and a prismatic joint (in total, 2DOF), the mobility of the system is 12. Even though both FEM (Finite Element Method) analysis and theoretical analysis may give the 2DOF flexure hinge model more accurate results than the 1DOF flexure hinge model, the 12 actuators to control 3DOF is expensive. Therefore, the flexure hinge should be designed in such a way that it not only minimize the number of actuators, but also ensures positional accuracy and sufficient workspace. In this paper, we design two types of flexure hinge. The first is designed with only 1DOF by fabricating a relatively thick the neck-downed section of the flexure hinge. t denotes the hinge thickness as described in Fig. 2. The second has 2DOF by fabricating a relatively thin neck-downed section. Consequently, the second hinge is modeled as having a 1DOF revolute joint and a 1DOF prismatic joint. If each chain of the system consists of one 2DOF hinge and two 1DOF hinges, the mobility of the system is 6. Therefore, it can be operated by only 6 actuators. This paper progresses as follows. Section 2 describes the system configuration. The kinematic analysis and analytic stiffness modeling of the 3-mobility and 6-mobility systems are performed in Sections 3 and 4, respectively. Section 5 presents the simulation results comparing the positioning accuracies of the 2DOF and 1DOF flexure hinge models and shows the merit of the proposed 6-mobility system. The effectiveness of this accurate model is shown through experimentation. II. SYSTEM CONFIGURATION The proposed 3DOF parallel mechanism consists of a platform and 3 chains, each of which has 3 flexure hinges as shown in Fig. 1. This mechanism employs flexure hinges at all joints and the joints are actuated by piezo-actuators. The three chains are 120 apart from each other. This symmetric structure reduces the effect of the temperature gradient and disturbance. The first and third joints of each chain are antagonistically actuated by pair piezo-actuators, which are attached to the base frame of the mechanism. All actuators push the circular surface of each link for fine sliding and rotational motions. The shape of one flexure hinge is shown in Fig. 2. The rotational displacement ( ) of the flexure hinge is defined as [8] [ M z + Fy R (1 sin )]R cos d = , (2) EI z ( ) where I z ( ) =
b 3 h( ) and h( ) = t 2 R(1 cos ) . 12 The translational displacement (dx) in the X-direction is defined as [8]

dx =

Fx R cos d , EA( )

(3)

where A( ) = bh( ) .
III.

KINEMATIC ANALYSIS FOR THE 3-MOBILITY SYSTEM

In order to show the performance of the 1DOF model of the flexure hinge, the kinematic model for the 3-mobility system is introduced in Fig. 3 (a). Since all flexure hinges are modeled with 1DOF, the mobility of the system is 3. The kinematic analysis of this case is performed in the following.

(a) 3 mobility system

(b) 6 mobility system Fig. 3. Modeling of parallel 3DOF planar mechanism A. First-order Kinematics First-order kinematics relates the output velocity vector to the independent joint velocity vector. In the following, [G ] denotes Jacobian and the left subscript of G denotes the number of serial sub-chain of the parallel manipulator. The superscript and subscript to the right side of G denote the dependent and independent parameters, respectively. Also, [Gu ](, j ) and [Gu ](i ,*) denote the jth column and the ith row of
[Gu ] , respectively. [Gu ](i , j ) denotes the (i, j) element of [Gu ] .

ij denotes the angular velocity of the jth joint of the ith chain.
The velocity relation for each serial chain is described as

u = [i Gu ] i ,

(4)

stiffness matrix can be expressed as


[ K uu ] = [i Gu ]T [i K ][i Gu ] ,
i =1 3

where the output velocity vector of the system and the angular velocity vector of the ith serial chain are defined as u = [ x y ] and i = [ i1 i 2 i 3 ]T , respectively. Since the three serial-chains have the same velocities at the center of platform, we have [1 Gu ] 1 = [ 2 Gu ] 2 (5) and
[1 Gu ] 1 = [3 Gu ] 3 .

(11)

Based on (11), an effective output force vector due to displacement of u is defined as

u = [ K uu ] u ,

(12)

(6)

where it is remarked that [ K uu ] is a diagonal when the system is in its symmetric configuration and the joints at the same position of the three sub-chains have the same stiffness [9]. IV. KINEMATIC ANALYSIS FOR THE 6-MOBILITY SYSTEM When the flexure hinge of each joint is considered as having 2DOF as presented in Fig. 3 (b), the mobility of the system is 6. Note that the first flexure hinge of each chain has an additional prismatic joint according to the elongation along the axial direction. The kinematic analysis of this case is performed in the following. A. First-order Kinematics For each chain, the center position (x, y) and the orientation angle of the platform is, respectively, given as x = Li1 cos(i1 ) + Li 2 cos(i1 + i 3 ) + Li 3 cos(i1 + i 3 + i 4 ) , (13)
y = Li1 sin(i1 ) + Li 2 sin(i1 + i 3 ) + Li 3 sin(i1 + i 3 + i 4 ) , (14)

Considering the three base joints as the independent joints, the velocity relation of the output vector to the independent joint vector is constructed by selecting the first row of each inverse Jacobian of the three serial sub-chains and constructing a matrix relation as below [1 Gu ](1,*) a = [ 2 Gu ](1,*) u = [Gua ]u , (7) [3 Gu ](1,*)
where a = [11 21 31 ]T and [i Gu ] = [i Gu ]1 .

Now, inverting the relation of (7) yields the first-order forward kinematic relation of the system, given by u u = [Ga ]a , (8)
u where [Ga ] = [Gua ]1 .

(15) and the lengths of the virtual links of Fig. 3(b) are denoted as
Li1 =

= i1 + i 3 + i4 ,

B. Stiffness Model Assume that the flexure hinges employed in Fig. 1 have only the rotational displacement about the z-axis and that the displacements in the other directions are ignored. The flexure hinges employed at all joints can be considered as springs, and require force or moment to bend or expand (or compress) them. Therefore, in order to compute the input load to operate this mechanism, we have to obtain the analytic stiffness model referenced to the output coordinate. Let the stiffness matrix for the ith chain be written as [i K ] = diag[i k1 i k2 i k3 ] , (9) where i k j denotes the stiffness element of the j joint of the ith chain. Then, in a state of equilibrium, the potential energy stored in the mechanism can be represented as 1 3 P.E. = {di T [i K ]di } 2 i =1 3 1 (10) = du T {[i Gu ]T [i K ][i Gu ]}du , 2 i =1 1 = du T [ K uu ]du 2 where [i Gu ] represents the inverse Jacobian relating the infinitesimal displacement of i to that of u , and it can be obtained by inverting [i G ] . [ K uu ] denotes the stiffness matrix defined in the operational space. From (10), the output
u

( li1 + li 2 + hi 2 )2 + li32 ,
Li 2 = li 4 2 + li 52 , Li 3 = li 6 ,

where hi 2 denotes the displacement of the prismatic joint of ith chain and lij denotes a fixed length of links. Differentiating the above position equations with respect to time yields the first-order kinematic relation for each serial chain, given by u = [i Gu ] i , (16) where u = [ x y ] , i = [ i1 hi 2 i 3 i 4 ]T . Since the three serial sub-chains have the same output velocity vectors at the center of the platform, we have [1 Gu ] 1 = [ 2 Gu ] 2 (17) and
[1 Gu ] 1 = [3 Gu ] 3 .

th

(18)

Here, we consider the first and third joints of each chain as the independent joints. Thus, there are six independent joints. Then, by rearranging (17) and (18), the relation between the independent joints ( a ) and the dependent joints ( p ) can be obtained as

p = [Gap ]a ,
where

(19)

a = [11 13 21 23 31 33 ]T , p = [h12 14 h22 24 h32 34 ]T .


Embedding (19) into (16) yields u = [1 Gu ] 1 = [1 Gu ][C ]a , where
1 0 0 0 0 0 [Gap ](1;*) [c ] = . 0 1 0 0 0 0 [Gap ](2;*) Finally, the relation of output and independent joint velocity is obtained as u u = [Ga ]a , (21)

In the 6-mobility system, the dimension of a is greater than that of u . Thus, the inverse of (22) is obtained as
u a = [Ga ]+ u ,

(23)

(20)

u where the weighted pseudo-inverse solution of [Ga ] is given by u * u u * u [Ga ]+ = [ K aa ]1[Ga ]T ([Ga ][ K aa ]1[Ga ]T ) 1 . This represents an optimal solution that minimizes the * potential energy of the system subject to (22). [ K aa ] represents an effective stiffness matrix referenced to the six independent joints. It is given by * [ K aa ] = [ K aa ] + [Gap ]T [ K pp ][Gap ] , (24)

where [ K pp ] denotes the stiffness matrix corresponding to the dependent joint set. Finally, the displacement of independent joints and dependant joints are obtained as u ai = ai1 + a = ai1 + [Ga ]+ [ui ui 1 ] (25) and

where
u [Ga ] = [1 Gu ][C ] .

B. Inverse Kinematics Given the center position (x, y) and the orientation angle of the platform, all of the joint positions of the parallel chain are obtained. An equivalent relation of (21) for infinitesimal motion is given by u u = [Ga ]a . (22)

pi = pi1 + p = pi1 + [Gap ]+ a .

(26)

C. Stiffness Model The stiffness matrices of the independent joints and the dependent joints can be written as

(a)X-directional displace ment (b)Y-directional error (c) -directional error Fig. 4. Error analysis for X-directional input

(a)X-directional displacement (b)Y-directional error Fig. 5. Error analysis for Y-directional input

(c) -directional error

(a) -directional displacement (b)X-directional error Fig. 6. Error analysis for -directional input
[ K aa ] = diag[1 k1 1k3 2 k1 2 k3 3 k1 3 k3 ]

(c)Y-directional error

(27) VI. EXPERIMENTS FOR STIFFNESS ANALYSIS (28)


th th

and
[ K pp ] = diag[1 k2 1k4 2 k2 2 k4 3 k2 3 k4 ] ,

respectively, where i k j is the stiffness of the j joint of i

chain. The output compliance matrix equivalent to the stiffness matrix given in (24) is represented as [9] * u u [Cuu ] = [Ga ][Caa ][Ga ]T , (29)
* where [Caa ] = [ K aa ]1 . Consequently, the effective output force vector due to u is defined as u = [ K uu ] u , (30)

The developed micro-manipulator consisting of 9 flexure hinges was made by a wire-cutting machine. Six sets of antagonistically driven piezo-actuators are employed. The stroke of the piezo-actuator is 17.4um . This allows the motion range 0 ~ 100 m in the x- and y-directions, and
0 ~ 0.1 in the direction. In this section, we perform experiment to corroborate the effectiveness of the simulation results. The objective of the experiment is to verify the trend and values of the operational stiffness of the developed mechanism by comparing it with the stiffness calculated through simulation. Fig. 8 shows the experimental setup. The detailed description appears in Fig. 9. The experiment to measure the operational stiffness is executed as follows. A cylindrical pole is attached to the center of the 6-mobility system. A horizontal bar is attached to the top of the pole to measure the rotational stiffness. The center of the pole is driven by a PICOMOTOR and the displacements of the end point are measured by LVDT. Since the motor is rooted in F/T sensor attached at the end-point of a manipulator, the stiffness of the x- and y-directions can be calculated by dividing the measured force data by the reaction forces measured in the F/T sensor. The rotational stiffness is measured by pushing the horizontal bar and measuring the moment resulting from the

where [ K uu ] = [Cuu ]1 . V. SIMULATION Finite Element Method (FEM) is employed to verify the feasibility of the analytic stiffness model of the system. Given the virtual, infinitesimal displacement ( u ) of the platform, the operational force vector ( u ) required to move the system can be calculated by (12) or (30). Then, the force vector ( u ) is applied to a virtual model in the FEM environment. Finally, a comparison of the output displacement vector of the FEM model with the initial displacement given in the analytic model can test the feasibility of the analytic stiffness model based on mobility analysis. The workspace of the system is given 0 ~ 100 m in the xand y-directions, and 0 ~ 0.1 in the direction. The simulation results of FEM are shown in Figures 4 through 6. The micro-mechanism with mobility 3 shows severe errors in the commanding and all the other directions. On the other hand, the micro-mechanism with mobility 6 shows negligible errors in the other directions and a minor error of 510% in the commanding direction. It is conformed, therefore, that that the system with mobility 6 has better precision than the system with mobility 3.

Fig. 8. Experimental setup

rotation of the 3DOF mechanism. Fig. 10 represents the experimental results. The slopes denote the stiffness of the 3DOF mechanism in each direction. It is remarked that the stiffness in the x- and y-directions are identical, which corroborates the analytic result reported previously [9]. The operational stiffness obtained from experimentation and simulation is listed in Table 1, where, K xx , K yy , and K denotes the diagonal terms [ K uu ] . As shown in Table 1, there is about a 30% of nominal values error rate between the simulation and experiment results. It is not perfect, but it is satisfactory since micro systems are very sensitive to the design parameters. This error rate is partly due to manufacturing errors and partly due to measurement errors. It is believed that this error can be recovered through design improvement and exact calibration process in real design.

TABLE I STIFFNESS VALUES

K xx (N/m)

K yy

(N/m) about 6712 about 9500

(Nm/deg) about 0. 067 about 0.048

Simulation (Mobility 6) experiment

about 6712 about 9500

(a)X-directional value

(b)Y-directional value Fig. 10. Stiffness obtained from experiments


[2]

(c) -directional value

VII. CONCLUSION Inaccurate modeling of the flexure hinge does not assure the precise motion of the micro-mechanism. In this paper, a planar 3DOF parallel-type micro-positioning mechanism is proposed in consideration of accurate flexure hinge modeling. Initially, we have shown through simulation that a planar 3DOF micro-mechanism having mobility 3 cannot assure the position accuracy of the system. On the other hand, the design having mobility 6 exhibits good position accuracy. We performed the experiment to corroborate the effectiveness of the simulation results. We conclude that accurate modeling enhances the position accuracy in the design of micromechanism. It is expected that this device can be usefully employed for scanning Electron Microscopy (SEM), x-ray lithography, mask alignment, micro-machining, and other similar technologies. REFERENCES
[1] S. Moriyama, T. Harada, and Takanashi, "Precision X-Y stage with a Piezo-driven Fine-table," Bull. Japan Soc. of Prec. Eng., Vol. 22, No. 1, 1988, pp. 13-17.

[3] [4]

[5]

M. Taniguchi, M. Ikeda, A. Inagaki, and R. Funatsu, "Ultra Precision Wafer Positioning by Six-axis Micro-motion Mechanism," Int. J. Japan Soc. Prec. Eng., Vol. 26, No. 1, 1992, pp. 35-40. Y. Tomita, F. Sato, K. Ito, and Y. Koyanagawa, "Decoupling Method of Ultraprecision Stage using Parallel Linkage Mechanism," Int. J. Japan Soc. Prec. Eng., Vol. 26, No. 1, 1992, pp. 35-40. J.W. Ryu, S.-Q. Lee, D.-G. Gweon, and K.S. Moon, "Inverse Kinematic modeling of a coupled flexure hinge mechanism," Mechatronics Vol. 9, 1999, pp. 657-674. S.H. Chang, C. Kai, and H.C. Chien, "An Ultra-Precision XY z PiezoMicropositioner Part : Design and Analysis," IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, Vol. 46, No. 4, July 1999, pp. 897-905. P. Gao, S.-M. Swei, and Z. Yuan, "A new piezo-driven precision micropositioning stage utilizing flexure hinges," Nanotechnology Vol. 10, 1999, pp. 394-398. A.G. Erdman and G.N. Sandor, "Mechanical Design: Analysis and Synthesis, Vol. 1, 2nd ed. Prentice Hall 1991. J.M. Paros and L. Weisbord, "How to design flexure hinge," Machine Design, Vol. 37, 1965, pp.151-157. W.K. Kim, B.-J. Yi, and W. Cho, "RCC Characteristics of Planar/Spherical Three Degree-of- Freedom Parallel Mechanisms With Joint Compliances," Transactions of the ASME, Journal of Mechanical Design, Vol. 122, March 2000, pp. 10-16.

[6] [7] [8] [9]

You might also like