You are on page 1of 6

Precision Engineering 34 (2010) 133–138

Contents lists available at ScienceDirect

Precision Engineering
journal homepage: www.elsevier.com/locate/precision

Analysis of parasitic motion in parallelogram compliant mechanism


Zongyue Ni a,∗ , Dawei Zhang a , Yingjun Wu b , Yanling Tian a , Ming Hu c
a
School of Mechanical Engineering, Tianjin University, Tianjin 300072, People’s Republic of China
b
School of Civil Engineering, Tianjin University, Tianjin 300072, People’s Republic of China
c
School of Electronic and Information Engineering, Tianjin University, Tianjin 300072, People’s Republic of China

a r t i c l e i n f o a b s t r a c t

Article history: In this paper, an analytical method for parasitic rotation and displacement calculations of parallelogram
Received 24 November 2008 compliant mechanisms is formulated. Mathematical model is proposed for the geometrical paramet-
Received in revised form 30 April 2009 ric study of parasitic motion, through which the approach to reduce parasitic motion is obtained. The
Accepted 11 May 2009
formulations for calculating the locations of the points on end-effectors with no theoretical parasitic dis-
Available online 18 May 2009
placement are presented. The analytical model predictions are confirmed by the finite element analysis.
Theoretical basis is built for improving the output precision of parallelogram compliant mechanisms.
Keywords:
© 2009 Elsevier Inc. All rights reserved.
Compliant mechanism
Flexure hinge
Parasitic motion
Symmetric structure

1. Introduction eral to a few hundreds of micrometers, and such a small travel range
can be considered as a point relatively to the dimensions of end-
Flexure-based compliant mechanisms are widely utilized in effector. Thus, the parasitic motion of the points on end-effector
ultra precision positioning/scanning systems to achieve nanome- should be calculated, and the point that has the minimum parasitic
ter resolution. To reduce parasitic motion and improve positioning motion will be selected as the best working point.
accuracy, such compliant mechanisms are usually designed as sym- The parallelogram compliant mechanism with beam flexure
metric structures about the motion axes of end-effectors [1–9]. Par- hinges is selected as the research object (Fig. 1), The motion direc-
allelogram compliant mechanisms have relatively smaller volumes tion of the end-effector is parallel with y-axis. The parasitic motion
and simpler structures because of the asymmetry about the motion of any point on the end-effector can be formulated by the dis-
axes (Fig. 1). However, the desired displacement output y in placement ıxD of point D and the displacement ıxE of point E in
y-direction is usually accompanied with parasitic rotation ı in xy- x-direction.
plane and parasitic displacement ıx in x-direction. Thus, the appli- The displacement of point D in x-direction is composed of three
cations of parallelogram mechanisms are limited in the science- parts: the displacement 1 of point B in x-direction caused by the
based engineering field where requires nano-level accuracy. deformation of flexure hinge 1 (Fig. 2(a)); the displacement l3 sin  1
The purpose of this research is to establish the relationship of point D in x-direction caused by the angle  1 which occurs with
between the geometrical parameters and the parasitic motion, and the deformation of flexure hinge 1 (Fig. 2(b)); The displacement 2
thus to reduce the parasitic motion by optimizing the geometrical of point D in x-direction caused by the deformation of flexure hinge
parameters. The methodologies for improving the output precision 2 (Fig. 2(c)).
of parallelogram compliant mechanisms are proposed on the basis The positive direction of x-axis is defined as the positive direc-
of the developed model. tion of displacement, and the clockwise rotation is defined as
the positive direction of angle, the displacement of point D in x-
2. The analytical model direction can be expressed as

The elastic deformation of compliant mechanisms driven by ıxD = 1 − l3 × sin 1 + 2 (1)


piezoelectric actuators is usually tiny in applications. The maximum
travel range of end-effector is generally within the range from sev- The same as point D, the displacement of point E in x-direction is
composed of: the displacement 4 of point G in x-direction caused
by the deformation of flexure hinge 4; the displacement l3 sin  4
∗ Corresponding author. of point E in x-direction caused by the angle  4 which occurs with
E-mail address: mecharts@yahoo.com.cn (Z. Ni). the deformation of flexure hinge 4; the displacement 3 of point

0141-6359/$ – see front matter © 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.precisioneng.2009.05.001
134 Z. Ni et al. / Precision Engineering 34 (2010) 133–138

The displacement in x-direction of any point on the end-


effector can be considered as a composition of the displacement
in x-direction caused by the rotation around point E (or D)
(Fig. 3(a)) and the displacement of point E (or D) in x-direction
(Fig. 3(b)).
The parasitic displacement of any point on the end-effector can
be expressed as
 l 
2
ıx = ı × + ıxE (4)
2+y

where y denotes the y coordinate of the point.


The inertia moments of flexure hinges 1–4 are set as Ii (i = 1, 2,
3, 4), respectively. Assuming the end-effector is cut by plane y = 0,
shearing force X1 , normal force X2 and bending moment X3 exist in
the sections.
Since the end-effector is not really divided into two parts, there
is no relative motion between these two sections.
The relative displacement in x-direction between the two sec-
tions can be expressed as
Fig. 1. Parallelogram compliant mechanisms which are asymmetric structures
1 = ı11 X1 + ı12 X2 + ı13 X3 + 1F = 0
about the motion axes.

The relative displacement along y-direction between the two


E in x-direction caused by the deformation of flexure hinge 3. The sections can be expressed as
displacement of point E in x-direction can be expressed as
2 = ı21 X1 + ı22 X2 + ı23 X3 + 2F = 0
ıxE = −3 + l3 × sin 4 − 4 (2)
The relative rotation angle in xy-plane between the two sections
The parasitic rotation of any point on the end-effector can be can be expressed as
expressed as
3 = ı31 X1 + ı32 X2 + ı33 X3 + 3F = 0
ıxD − ıxE
ı = (3)
l2 where ıij = ıji (i, j = 1, 2, 3).

Fig. 2. The composition of the displacement of point D in x-direction. (a) The displacement caused by the deformation of flexure hinge 1. (b) The displacement caused by the
rotation. (c) The displacement caused by the deformation of flexure hinge 2.
Z. Ni et al. / Precision Engineering 34 (2010) 133–138 135

Table 1
The parameters of loop computing function program.

l1 (mm) l3 (mm) l4 (mm) I1–4 (mm4 )

Start point 60 3 20 0.05


End point 60 8 30 1.05
Step length 0 0.25 0.5 0.05

Based on the computational results, some characteristics can be


found as follows:

a. Set E1∗ as E1 when Ii (i = 1, 2, 3, 4) have the same value, and set


E1 R = E1 /E1∗ . When I2 = I3 , E1 R decreases with increasing ε1 = I1 /I2
or ε2 = I4 /I3 (Fig. 4). When I1 = I4 and I2 = I3 , E1 R decreases with
increasing ε = I1 /I2 = I4 /I3 (Fig. 5).
b. When I2 = / I3 , E1 is invariable with Ii (i = 1, 2, 3, 4) changing in
the same ratio; E1 decreases with increasing I1 or I4 , and with
decreasing Max[I2 , I3 ]. There is usually a point E1 (s) of minimum
E1 during the increase of Min[I2 , I3 ] (Fig. 6). When I1 /Max[I2 ,
I3 ] or I4 /Max[I2 , I3 ] increases, the x-coordinate of point E1 (s)
approaches 1, and the ratio E1 (s)/E1 (1) increases (Fig. 6(a)–(c)).
c. Set E1∗ as E1 when l4 /l3 = 3, and set E1 R = E1 /E1∗ . E1 R decreases
with increasing  = l4 /l3 (Fig. 7).

3.2. The method for calculating the location of the point with no
parasitic displacement

Fig. 3. The composition of the displacement in x-direction of any point on the As there is rotation on the end-effector, the parasitic displace-
end-effector. (a) The displacement caused by the rotation around point E. (b) The ment in x-direction of different points is varied. For a specified
displacement caused by ıxE . parallelogram compliant mechanism, the location of the point with
no parasitic displacement can be obtained by using Eq. (4), and fur-
The expressions of ıij and iF (i = 1, 2, 3) can be obtained by ther the parasitic displacement of any point on the end-effector can
graphical solution, and thus the expressions of Xi (i = 1, 2, 3) can be be calculated.
obtained with ıij and iF from these three equations.
The moments at points A–H can be expressed as follows, respec-
tively:

MA = −X1 (l4 − l3 ) + X2 l1 + X3 − Fl1


MB = −X1 l4 + X2 l1 + X3 − Fl1
MC = X1 l4 + X3
MD = −X1 (l4 − l3 ) + X3
ME = X1 (l4 − l3 ) + X3
MF = X1 l4 + X3
MG = X1 l4 + X2 l1 + X3
MH = X1 (l4 − l3 ) + X2 l1 + X3

The expressions of i (i = 1, 2, 3),  1 and  4 can be obtained (see


Appendix) by graphical solution.
Fig. 4. The relationship between inertia moments and parasitic rotation.
3. Parasitic motion analysis

3.1. The geometrical parametric study of parasitic rotation

The parasitic rotation of the end-effector is measured with the


relative rotation error E1 = |ı/y|(y denotes the output displace-
ment). It is easy to find from the expression of E1 (see Appendix)
that E1 decreases with increasing l1 or l4 , and decreasing l3 . How-
ever, the partial derivatives ˛E1 /˛Ii (i = 1, 2, 3, 4) are multivariate
nonlinear equations, a loop computing function is carried out to
obtain the relationship between E1 and Ii (i = 1, 2, 3, 4). Because the
parameter l1 has no influence on the sign of ˛E1 /˛Ii (i = 1, 2, 3, 4),
and the influence of Ii (i = 1, 2, 3, 4) on E1 may be different when li
(i = 3, 4) change, 6-loop computing function of li (i = 3, 4) and Ii (i = 1,
2, 3, 4) should be carried out.
The parameters of computer program are shown in Table 1, the
program totally calculate 216 = 85,766,121 times. Fig. 5. The relationship between inertia moments and parasitic rotation.
136 Z. Ni et al. / Precision Engineering 34 (2010) 133–138

Fig. 8. The relationship between inertia moments and parasitic rotation based on
the finite element analysis.

ering the machining process, there should be small fillets at the


roots of the flexure hinges, which are formed by the wire of EDM.
The fillet radius r is set as 0.15 mm.
Set E1∗ as E1 when Ii (i = 1, 2, 3, 4) have the same value, and set
E1 R = E1 /E1∗ . Marking the positions of different values of E1 R on
squared paper when ε changes, the fitting function can be obtained
by using least square method, E1 R (ε) = 0.3371 + 0.6874/ε (Fig. 8).

4.2. The relationship between the ratio l4 /l3 and parasitic rotation

The parameters of FEM are set as: l1 = 55 mm,  = l4 /l3 , l2 = 2


(l4 − l3 ), Ii (i = 1, 2, 3, 4) = 0.229 mm4 , model thickness is 8 mm, fil-
let radius r = 0.15 mm, Young’s modulus E = 70 GPa, Poisson’s ratio
 = 0.33, input position  = 0.5 (Fig. 1), input displacement: −40 nm
in y-axis.
Set E1∗ as E1 when l4 /l3 = 3, and set E1 R = E1 /E1∗ . Marking the posi-
Fig. 6. The relationship between inertia moments and E1 (s). (a) I1 (I1 = I4 )/Max[I2 ,
I3 ] = 0.5. (b) I1 (I1 = I4 )/Max[I2 , I3 ] = 1. (c) I1 (I1 = I4 )/Max[I2 , I3 ] = 2. tions of different values of E1 R on squared paper when  changes,
the fitting function can be obtained by using least square method,
E1 R () = 0.0832 + 2.8925/ (Fig. 9).
4. Finite element analysis Comparing Figs. 5 with 8 and Figs. 7 with 9, respectively, it is
noted that the analytical results are in good agreement with the
To validate the analytical solution, finite element analysis is con- finite element results.
ducted. The finite element software is patran/nastran 2005, and the
element type is Hex8.
5. Error analysis

4.1. The relationship between inertia moments and parasitic When Eq. (4) is used to calculate the location of the point with
rotation ıx = 0, there are usually obvious calculating errors. After a further
investigation, it can be found that the calculating errors are mainly
The parameters of FEM are set as: l1 = 50 mm, l2 = 30 mm, caused by the bending deformation of lever 1 and lever 2.
l3 = 5 mm, l4 = 20 mm, I1 = I4 , I2 = I3 , ε = I1 /I2 = I4 /I3 , model thickness An infinitesimal dx in length of the levers is considered as the
is 8 mm, Young’s modulus E = 70 GPa, Poisson’s ratio  = 0.33, input research object (Fig. 10), O1 O2 represents the neutral layer of the
position  = 0.5, input displacement: −35 nm along y-axis. Consid-

Fig. 9. The relationship between the ratio l4 /l3 and parasitic rotation based on the
Fig. 7. The relationship between the ratio l4 /l3 and parasitic rotation. finite element analysis.
Z. Ni et al. / Precision Engineering 34 (2010) 133–138 137

Table 2
The geometry parameters of finite element models.

Model l1 l2 I1,4 /I2,3 l4 /l3 h 

1 100 80 1 7 20 0.220
2 90 50 6 6 10 0.194
3 80 40 5 5 20 0.188
4 70 40 2 5 15 0.179
5 50 30 3 4 12 0.190

Table 3
The locations of ıx = 0 based on Eq. (6) and FEM.

Input (␮m) 0.1 0.5 2 10

Model Eq. (6) (mm) −7.9071


1 FEM (mm) −7.2003 −7.2001 −7.2001 −7.2003
Fig. 10. The infinitesimal dx in length of lever.
Model Eq. (6) (mm) −32.455
2 FEM (mm) −31.302 −31.305 −31.304 −31.302
infinitesimal. When the bending deformation of the infinitesimal Model Eq. (6) (mm) −5.8929
occurs, the neutral layer bends with its length invariable. 3 FEM (mm) −5.3758 −5.3756 −5.3753 −5.3758
Setting as the radius of curvature of O1 O2 , M as the moment in
Model Eq. (6) (mm) −7.0514
the cross-section of the infinitesimal, and I as the inertia moment 4 FEM (mm) −6.1026 −6.1030 −6.1028 −6.1026
of the cross-section, the following equations can be obtained:
Model Eq. (6) (mm) −5.3346
1 M 5 FEM (mm) −5.0893 −5.0892 −5.0890 −5.0893
=
EI

m n − mn h
= In Eq. (6), if the ıx* is set as zero, we can find the expression of
mn 2
y does not have any relation with the input displacement, which
The elongation of the surface on the infinitesimal is indicate that the location without parasitic displacement (ıx* = 0)
Mh on the end-effector is fixed when the input displacement changes.
ı= dx Finite-element simulations are used to verify the above inference.
2EI
Model thickness is 8 mm, Young’s modulus E = 70 GPa, Pois-
The elongation of the surface on lever is son’s ratio  = 0.33 and fillet radius r = 0.15 mm (see Fig. 1)
 l1 /2 (Tables 2 and 3).
h
= M(x)dx
2EI −l1 /2
6. Conclusions
The elongation of the surface BC on lever 1 and surface GF (Fig. 1)
on lever 2 are 1 and 2 , respectively (the expressions are shown The analytical equations of parasitic motion for parallelogram
in Appendix).Eq. (1) is corrected as compliant mechanism have been developed. A geometrical para-
∗ metric investigation was performed based on the established
ıxD = 1 − l3 × sin 1 + 2 + 1
model. Finite element analysis was also conducted to verify the
Eq. (2) is corrected as developed model. The coincidence between the analytical results
and finite element analysis shows the established model is correct.
ıxE∗ = −3 + l3 × sin 4 − 4 + 2 It is found that the parasitic rotation reduces with the increasing
ratio I1,4 /I2,3 , and l4 /l3 , and the increasing length l1 . Based on the
Eq. (3) is corrected as
achieved analytical model, the location of the point without para-
∗ − ıx∗
ıxD sitic displacement can be calculated, and it is fixed when the input
ı ∗ = E
l2 displacement varies, thus this point can be selected as the working
point.
Eq. (4) is corrected as
∗ + ıx∗
ıxD
ıx∗ = E
+ ı ∗ ×y (5) Acknowledgment
2
ıx∗ ıx∗ + ıx∗ This work is supported by the National Nature Science Founda-
y= ∗
− D ∗ E (6) tion of China [NSFC] No. 50275104.
ı 2ı

Appendix A.
din l3 ((I1 I2 I3 (I2 + I3 ) − I2 I3 (I2 + I3 )I4 + I1 (I22 + l4 I2 I3 + I32 )I4 )l32 + 4I2 I3 (−l2 I1 + I2 + I3 )I4 l3 l4 + 48I1 I2 I3 I4 l42 )
1 =
2l1 ((I1 I2 I3 (I2 + I3 ) + I2 I3 I4 (I2 + I3 ) + I1 (I22 + l4 I2 I3 + I32 )I4 )l32 − 48I1 I2 I3 I4 l3 l4 + 48I1 I2 I3 I4 l42 )

−(din I1 I3 I4 l3 (l3 − 2l4 )(−5I2 l3 + I3 l3 + l2 I2 l4 ))


2 =
l1 ((I1 I2 I3 (I2 + I3 ) + I2 I3 I4 (I2 + I3 ) + I1 (I22 + l4 I2 I3 + I32 )I4 )l32 − 48I1 I2 I3 I4 l3 l4 + 48I1 I2 I3 I4 l42 )
138 Z. Ni et al. / Precision Engineering 34 (2010) 133–138

−(din I1 I2 I4 l3 (l3 − 2l4 )(−5I3 l3 + I2 l3 + l2 I3 l4 ))


3 =
l1 ((I1 I2 I3 (I2 + I3 ) + I2 I3 I4 (I2 + I3 ) + I1 (I22 + l4 I2 I3 + I32 )I4 )l32 − 48I1 I2 I3 I4 l3 l4 + 48I1 I2 I3 I4 l42 )

−(din l3 ((I1 I2 I3 (I2 + I3 ) − (I2 I3 (I2 + I3 ) + I1 (I22 + l4 I2 I3 + I32 ))I4 )l32 − 4I1 I2 I3 (−l2 I4 + I2 + I3 )l3 l4 − 48I1 I2 I3 I4 l42 ))
4 =
2l1 ((I1 I2 I3 (I2 + I3 ) + I2 I3 I4 (I2 + I3 ) + I1 (I22 + l4 I2 I3 + I32 )I4 )l32 − 48I1 I2 I3 I4 l3 l4 + 48I1 I2 I3 I4 l42 )

1,4 = din /(l1 )

−(2(I2 + I3 )(I1 I2 I3 + I2 I3 I4 + I1 (I2 + I3 )I4 )l32 (l3 − l4 ))


E1 =
l1 l2 ((I1 I2 I3 (I2 + I3 ) + I2 I3 I4 (I2 + I3 ) + I1 (I22 + l4 I2 I3 + I32 )I4 )l32 − 48I1 I2 I3 I4 l3 l4 + 48I1 I2 I3 I4 l42 )

−(3din (12hI1 I2 I3 (I2 + I3 )I4 (l3 − 2l4 ) + I4 (−(I1 I2 I3 (l3 I2 + I3 ) + I2 I3 I4 (I2 + I3 ) + I1 I4 (I22 + l4 I2 I3 + I32 ))l32 + 24I1 I2 I3 (3I2 + I3 + 2I4 )l3 l4 − 48I1 I2 I3 (2I2 + I3 + I4 )l42 )
+((I2 I3 I42 (I2 + I3 ) + I12 (I2 I3 (I2 + I3 ) + (I22 + l4 I2 I3 + I32 )I4 ) + I1 I4 (l4 I2 I3 (I2 + I3 ) + (I22 + l4 I2 I3 + I32 )I4 ))l32 − 48I1 I2 I3 I4 (I1 + I2 + I3 + I4 )l3 l4 + 48I1 I2 I3 I4 (I1 + I2 + I3 + I4 )l42 )))
1 =
bh2 l3 ((I1 I2 I3 (I2 + I3 ) + I2 I3 I4 (I2 + I3 ) + I1 I4 (I22 + l4 I2 I3 + I32 ))l32 − 48I1 I2 I3 I4 l3 l4 + 48I1 I2 I3 I4 l42 )

3din I4 ((I1 I2 I3 (l3 I2 +I3 )+I2 I3 (I2 +I3 )I4 +I1 (I 2 +l4 I2 I3 +I 2 )I4 )l2 +12hI1 I2 I3 (I2 +I3 )(l3 −2l4 )−24I1 I2 I3 (I2 −I3 +2I4 )l3 l4 −48I1 I2 I3 (I3 −I4 )l2 )
2 = 2 3 3 4
bh2 l3 ((I1 I2 I3 (I2 +I3 )+I2 I3 I4 (I2 +I3 )+I1 I4 (I 2 +l4 I2 I3 +I 2 ))l2 −48I1 I2 I3 I4 l3 l4 +48I1 I2 I3 I4 l2 )
2 3 3 4

din E((I2 I3 I42 (I2 + I3 ) + I12 (I2 I3 (I2 + I3 ) + (I22 + l4 I2 I3 + I32 )I4 ) + I1 I4 (l4 I2 I3 (I2 + I3 ) + (I22 + l4 I2 I3 + I32 )I4 ))l32 − 48I1 I2 I3 I4 (I1 + I2 + I3 + I4 )l3 l4
+48I1 I2 I3 I4 (I1 + I2 + I3 + I4 )l42 )
Fin =
l12 l3 ((I1 I2 I3 (I2 + I3 ) + I2 I3 I4 (I2 + I3 ) + I1 I4 (I22 + l4 I2 I3 + I32 ))l32 − 48I1 I2 I3 I4 l3 l4 + 48I1 I2 I3 I4 l42 ) 2

The parameters b, din and Fin denote model thickness, input dis- [4] Choi SB, Han SS, Han YM, Thompson BS. A magnification device for precision
mechanisms featuring piezoactuators and flexure hinges: design and experi-
placement and input force, respectively.
mental validation. Mechanism and Machine Theory 2007;42:1184–98.
[5] Kim D, Kang D, Shim J, Song I, Gweon D. Optimal design of a flexure hinge-
based XYZ atomic force microscopy scanner for minimizing Abbe errors. Review
References
of Scientific Instrument 2005;76:073706.
[6] Wang YL. Modular design process for a micro motion actuator. Mechatronics
[1] Kim JH, Kim SH, Kwak YK. Development and optimization of 3-D bridge-type 2005;15:793–806.
hinge mechanisms. Sensors and Actuators A 2004;116:530–8. [7] Chu CL, Fan SH. A novel long-travel piezoelectric-driven linear nanopositioning
[2] Lobontiu N, Garcia E. Analytical model of displacement amplification and stiff- stage. Precision Engineering 2006;30:85–95.
ness optimization for a class of flexure-based compliant mechanisms. Computers [8] Ma HW, Yao SM, Wang LQ, Zhong Z. Analysis of the displacement amplifi-
and Structures 2003;81:2797–810. cation ratio of bridge-type flexure hinge. Sensors and Actuators A Physical
[3] Choi KB, Kim DH. Monolithic parallel linear compliant mechanism for two 2006;132(2):730–6.
axes ultraprecision linear motion. Review of Scientific Instrument 2006;77: [9] Gao P, Swei SM, Yuan ZJ. A new piezodriven precision micropositioning stage
065106. utilizing flexure hinges. Nanotechnology 1999;10:394–8.

You might also like