Vous êtes sur la page 1sur 14

Mechanism and Machine Theory 50 (2012) 15–28

Contents lists available at SciVerse ScienceDirect

Mechanism and Machine Theory


journal homepage: www.elsevier.com/locate/mechmt

Review

A novel compliant linear-motion mechanism based on parasitic


motion compensation
Zhao Hongzhe, Bi Shusheng ⁎, Yu Jingjun
Robotics Institute, Beihang University, Beijing 100191, 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, the center shift of the generalized cross-spring pivot is taken advantage of to
Received 9 August 2010 compensate for the inherent parasitic motion of the compliant parallel four-bar mechanism.
Received in revised form 6 November 2011 So the rectilinear motion of a linear-motion mechanism, which is constructed by several leaf
Accepted 17 November 2011
springs or cartwheel flexural pivots, is improved. The compliant linear-motion mechanism is
Available online 16 December 2011
firstly modeled by making use of the building block method. Then the compensation condition
for the parasitic motion is obtained. Meanwhile, the approach to large stroke and small para-
Keywords: sitic error is proposed, for the sake of achieving a high performance rectilinear mechanism.
Linear-motion mechanism
Furthermore, the validity and effectiveness are verified by finite element analysis (FEA). Final-
Parallel four-bar mechanism
ly, both the qualitative and quantitative analyses show that the advantages of this linear-
Flexural building block
Parasitic motion compensation motion mechanism are remarkable, and the application prospect in precision engineering is
promising.
© 2011 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2. The compensation principle for parasitic motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3. The center shift model for a generalized cross-spring pivot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4. The model for compliant linear-motion mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.1. Parameter transformation for the pivot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2. The model of a single kinematic chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.3. The model of the linear-motion mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5. Analysis of the compliant linear-motion mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
5.1. The compensation condition for parasitic motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
5.2. The characteristics of parasitic motion and primary motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.3. The comparison of parasitic motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
6. FEA verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
7. The compliant double parallel four-bar mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
8. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

⁎ Corresponding author. Tel.: + 86 10 82338926.


E-mail addresses: hongzhezhao@gmail.com (H. Zhao), bishusheng@gmail.com (S. Bi).

0094-114X/$ – see front matter © 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mechmachtheory.2011.11.009
16 H. Zhao et al. / Mechanism and Machine Theory 50 (2012) 15–28

1. Introduction

The flexural pivots have been widely exploited in precision engineering, such as scientific instruments, aerospace mechanisms,
optical alignment mechanisms, sensors and actuators, because of their high performance and low cost: ease of assembly,
maintenance-free, no backlash, diminished friction, infinitesimal resolution, and simplified manufacturing process [1,2]. However,
the drawbacks restrict its application scope for the flexural pivots, and the main disadvantages are the limited range of motion
and undesirable parasitic motion [3,4]. The leaf-type flexural pivots can increase the motion range via distributed compliance,
but it fulfills at the cost of worsened parasitic motions [5]. Accordingly, it has been generally acknowledged that the design of
flexural pivot faced with performance tradeoff.
On the other hand, the complex flexural pivot and compliant mechanism are investigated through the building block method.
Although the challenge is posed in design, the performance can be improved by the combination of several flexural modules. For
instance, the cartwheel flexural pivot, which is combined by two triangle flexural pivot in series, represents a factor of five improve-
ment in center shift over the cross-spring pivot [1]. A high precision butterfly flexible pivot was exploited in pointing and scanning
space mechanisms, because both large motion range and small center shift were achieved by means of ingenious block design [6].
In addition, the approach for passive compliance was investigated, and some parallel mechanisms can be designed [7].
As far as the compliant linear-motion mechanisms are concerned, the parallel four-bar mechanism combined by two leaves
(Fig. 1(a)) is a typical example. In order to increase the motion range and diminish the parasitic error, the compliant double par-
allel four-bar mechanism is researched [5,8]. Furthermore, Spanoudakis and Henein utilized a ratio-control lever mechanism to
reduce the errors originating from manufacturing and assembly tolerances [9,10]. In addition, a higher precision rectilinear mech-
anism was constructed by mirroring the compliant double parallel four-bar mechanism [8,11]. Except for the compliant parallel-
ogram and its derivatives, some novel compliant linear-motion mechanisms were studied [12–15]. A set of compliant Roberts
mechanisms was combined in series and parallel to compensate for the parasitic motion [12,13]. Compliant Scott-Russel mecha-
nism was also taken advantage of to change the motion direction and amplify the motion range [14,15].
Moreover, the conventional joints in parallel four-bar mechanism were replaced with the cartwheel flexural pivots (Fig. 1(b))
[16,17]. Furthermore, a linear-motion machine for x-ray interferometry was symmetrically arranged by four of these flexural
modules [16]. In the initial direction, the rigid link will becomes shorter during the rotation. However, the instantaneous center
of the cartwheel flexural pivot moves towards the fixed stage, so the distance between the moving stage and fixed stage
decreases. Consequently, the parasitic motion of this compliant four-bar mechanism becomes worse due to the adoption of the
cartwheel flexural modules.
In our previous work [18,19], a non-dimensional model has been developed to predict the properties of stiffness, stress, and
accuracy for the generalized cross-spring pivot. A useful characteristic is captured: comparing with the initial center, the instan-
taneous center of this pivot will move away from the fixed stage during the rotation, if the position of intersection point on the
leaf is in the range (12.7322%, 87.2678%). Accordingly, the parasitic motion of the parallel four-bar mechanism can be compen-
sated for by adopting the generalized cross-spring pivot, as illustrated in Fig. 1(c). Thus several of these linear-motion mecha-
nisms will be combined to further diminish the parasitic error.
The objective of the present paper is to compensate for the parasitic motion of the compliant linear-motion mechanism by
exploiting the center shift of the generalized cross-spring pivot. The compensation condition for parasitic motion is developed,
and the method for choosing the geometric parameters is given to achieve the large motion range and small parasitic error. Com-
paring with the linear-motion mechanism combined by several leaf springs or cartwheel flexural pivots, the distinct advantages
are shown by finite element analysis (FEA).

2. The compensation principle for parasitic motion

Drawing inspiration from the replacement of conventional joint with a flexural one, the compliant linear-motion mechanism is
essentially a parallel four-bar mechanism. In order to explain the compensation principle for parasitic motion, a single kinematic
chain of the compliant four-bar mechanism is taken into account, and an equivalent rigid body model is given.

a) leaf building block b) cartwheel flexural building block c) generalized cross-spring building block

Fig. 1. The compliant linear-motion mechanism combined by flexural building blocks.


H. Zhao et al. / Mechanism and Machine Theory 50 (2012) 15–28 17

As depicted in Fig. 2, the link length is the initial distance between the two pivot points. Obviously, the projection of the link in
the y direction becomes smaller with the rotation of these two pivots, which is denoted by dyl. On the other hand, for a flexural
pivot, the center shift can be treated as two prismatic joints (the detail will be described in Section 3). As shown in Fig. 2, when
the center shift is in the direction of the arrow (the prismatic joint in the x direction is not illustrated), it is possible that the free
end of the kinematic chain dose not move in the y direction. As aforementioned, the center shift of the cartwheel flexural pivot
can not satisfy the desirable direction. But for the generalized cross-spring pivot, the center shift makes the pivot to be a suitable
candidate, if the geometric parameter λ is in the range (0.127322, 0.872678).
As far as the magnitude of the center shift is concerned, the component in the y direction is of the order of θ 2, according to the
accuracy characteristics of the generalized cross-spring pivot [19]. On the other hand, it is a remarkable fact that the shortening of
the link, dyl, is also of the order of θ 2. Consequently, utilizing the center shift of the two flexural pivots, the inherent parasitic error
may be compensated for.
Although it is possible to compensate for the parasitic motion both in direction and magnitude, the quantitative analysis needs
to be implemented to provide the parametric insight. So the characteristics of the generalized cross-spring pivot are reviewed in
the subsequent section, and the model of the compliant linear-motion mechanism is developed.

3. The center shift model for a generalized cross-spring pivot

The generalized cross-spring pivot can be adopted to design the compliant linear-motion mechanism as a building block. Our
previous work [19] has developed a non-dimensional model to predict the parasitic motion of the pivot, so the center shift model
for this pivot is reviewed so that the results can be utilized to analyze the linear-motion mechanism.
First of all, the geometric parameters λ and α associated with the intersection point are defined as schematically illustrated in
Fig. 3. The instantaneous center oi is fixed on the moving stage, and its initial position o is the intersection point of the leaves.
However, the instantaneous center will move with the rotation of the pivot. The center shift is defined as the distance of ooi.
The components in the x and y directions are DX and DY respectively, and in Fig. 3 they are enlarged for the sake of clarity.
Then, according to the dimensions of the leaf (as shown in Fig. 4), the non-dimensional parameters are defined as the follow-
ing lower case letters

2 2
!2
M p Lp F p Lp P p Lp DY DX Lp
mp ¼ fp ¼ ; pp ¼ ; dyp ¼ ; dxp ¼ ; dp ¼ 12
EI EI EI Lp Lp Tp

where Mp, Fp and Pp denote bending moment, horizontal force, and vertical force applied on the moving stage as the external
loads; dxp and dyp are the center shift of this pivot; Lp, Wp and Tp are the length, width, and thickness of the beam respectively,
and they are named as shape parameters; I is the moment of inertia; dp is a parameter related to the length and thickness; E is
Young's modulus of the material.
Hence, if the geometric parameter λ is in the range [0, 0.5], the center shift can be expressed as

1  
3 2 3
dxp ≈− 108λ −117λ þ 21λ−1 θ
150 cosα
   " #
pp1 þ pp2 1  2

3
pp1 −pp2 1 1  2 
2
þ 45λ þ 75λ−5 θ − þ 9λ −9λ þ 11 θ ð1:1Þ
2 cosα 63000 2 sinα dp 6300
" #
1 tan2 α  2 
2 2  2 
2
dyp ≈ 1þ 9λ −9λ þ 1 θ − 9λ −9λ þ 1 θ
2 cosα 25 15

Fig. 2. The equivalent rigid body model for a single kinematic chain (prismatic joints in the x direction are not shown).
18 H. Zhao et al. / Mechanism and Machine Theory 50 (2012) 15–28

Fig. 3. Deflected configuration of the generalized cross-spring pivot.

1  4 3 2

4
þ 2592λ −3024λ þ 1338λ −241λ þ 2 θ
1500 " #
  12λ−1 12λ−1  2  ð2λ−1Þ tan2 α  2 
2 2
þ pp1 −pp2 cotα þ 9λ −9λ þ 11 θ − 9λ −9λ þ 11 θ θ g ð1:2Þ
10dp 63000 10500
" #
  1 1  
2 2
þ pp1 þ pp2 þ 9λ −9λ þ 11 θ
dp 6300
where pp1 and pp2 are the axial forces at the free end of the two leaves. The relationships between these axial forces and external
forces are
h i
10 ð12λ−1Þpp þ 120 cosα ð2λ−1Þ θ−100f p
pp1 −pp2 ¼   ð2:1Þ
sinα ð12λ−1Þ2 θ2 þ 100
h   i
40 4 sin2 α 9λ2 −9λ þ 1 −3 cos2 α ð12λ−1Þð2λ−1Þ θ2 þ 10f p cosα ð12λ−1Þθ þ 100pp cosα
pp1 þ pp2 ¼   : ð2:2Þ
cos2 α ð12λ−1Þ2 θ2 þ 100

On the other hand, if the geometric parameter λ is in the range (0.5, 1], the external loads applied on the moving stage can be
transformed into the loads applied on the fixed stage.
8 0
>
> f p ¼ −f p cosθ þ pp sinθ
< 0
pp ¼ pp cosθ þ f p sinθ ð3Þ
>
>
: 0
mp ¼ mp þ f p ½λ cosα cosθ þ ð1−λÞ cosα −pp λ cosα sinθ
where the superscript ′ refers to the pivot whose geometric parameter λ is in the range [0, 0.5]. Thus the result can be fully uti-
lized and the components of center shift dxp and dyp will be derived from dx′p and dy′p
( 0 0
dxp ¼ −dxp cosθ þ dyp sinθ
0 0 : ð4Þ
dyp ¼ dyp cosθ þ dxp sinθ

Fig. 4. A generalized cross-spring pivot.


H. Zhao et al. / Mechanism and Machine Theory 50 (2012) 15–28 19

Fig. 5. Compliant linear-motion mechanism.

Finally, replacing λ′ with 1-λ, the center shift of the pivot whose geometric parameter λ is greater than 0.5 can be obtained.
Thus the applicable scope for the model is extended by the boundary transformation, and the accuracy of the model is kept for
the pivot whose moving stage has a large deflection. In addition, the building block method is preliminarily explored.
It is noteworthy that the dominant term of the component dyp will be positive, if the geometric parameter λ is in the range
(0.127322, 0.872678). Thus the instantaneous center of this pivot will drift away from the fixed stage with the rotation, which
satisfies the desirable direction of parasitic error compensation.

4. The model for compliant linear-motion mechanism

As shown in Fig. 5, the compliant linear-motion mechanism is developed by two kinematic chains, and each chain consists of
two generalized cross-spring pivots as the building blocks. Thus the geometric parameters μ, a, and α (not shown in Fig. 5) are
defined to describe the configuration of these compliant linear-motion mechanisms. The thin lines refer to the leaves, but the
thick lines denote the rigid bodies. In order to differentiate these building blocks from each other, the module numbers are
assigned as 11, 12, 21, and 22.
The mechanism is different from the conventional parallel four-bar mechanism. Because the parasitic motion of the rigid link
will be compensated for, if the reasonable parameters are chosen. So the model is developed using the building block method.

4.1. Parameter transformation for the pivot

Above all, the non-dimensional parameters need to be defined again, because the linear-motion mechanism is normalized by
the length L, which is the distance between the fixed platform and moving platform. Hence the leaf length for the generalized
cross-spring pivot is

Lp ¼ L=ξ ð5Þ

where ξ = cosα/μ. Then the normalized moment is

Mp Lp 1 Mp L mno
mp ¼ ¼ ¼ ð6Þ
EI ξ EI ξ

Fig. 6. The transform model for a single kinematic chain.


20 H. Zhao et al. / Mechanism and Machine Theory 50 (2012) 15–28

where the suffixes no refers to the module number, and it corresponds to the parameters normalized by L but not Lp. Similarly, the
other non-dimensional parameters are

f no pno dno d
fp ¼ ; pp ¼ ; dyp ¼ ξdyno ; dxp ¼ ξdxno ; dp ¼ ¼ : ð7Þ
ξ2 ξ2 ξ2 ξ2

In addition, the relationship between the geometric parameters is

λ ¼ 1−a=μ: ð8Þ

It should be noted that the constraint condition for the geometric parameters μ and a is validated throughout the whole
research

0≤a≤μ≤1: ð9Þ

Then the center shift of the generalized cross-spring pivot can be normalized by L, and they are expressed as the following
form on purpose that the physical attributes are revealed clearly.

3
dxno ¼ C xθ θno þ H xθXno þ C xf Xno f no þ C xpXno pno ð10:1Þ

2
dyno ¼ C yθ θno þ H yθXno þ C yf Xno f no þ C ypXno pno ð10:2Þ

where Hxθ_no and Hyθ_no are the higher order terms; Cxf_no and Cyf_no are the compliances for the horizontal force fno in the x and y
directions respectively; Cxp_no and Cyp_no are the compliances for the vertical force pno in the x and y directions respectively; Cxθ
and Cyθ are the coefficients of the dominant terms, and they remain constant for all the flexural modules. In addition, Cxθ and
Cyθ are given as the same form irrespective of the range of geometric parameter λ.

−1  
2
C xθ ¼ 9λ −9λ þ 1 ð12λ−1Þ ð11:1Þ
150ξ cosα

−1  2 
C yθ ¼ 9λ −9λ þ 1 ð11:2Þ
15ξ cosα

4.2. The model of a single kinematic chain

Then a single kinematic chain is taken into account. A transform model is further developed to parametrically predict the per-
formance of the single chain, by exploiting the result of the generalized cross-spring pivot. As illustrated in Fig. 6, taking the first
kinematic chain as an example, this transform model must satisfy two conditions as follows.

Boundary condition. Both of the two pivots are fixed on a “rotational base”. A local coordinate system that defines loads and
displacements will rotate with the base. This condition defines a virtual base.

Initial condition. The moving stage of pivot 12 must be in parallel with the x axis of the global coordinate system at any final time; at
the initial time it must be in parallel with the x12 axis of the local coordinate system. This condition defines a virtual initial position.

Fig. 7. Parameter space for μ, a, and α.


H. Zhao et al. / Mechanism and Machine Theory 50 (2012) 15–28 21

Fig. 8. The constraint for geometric parameters to achieve large stroke and small parasitic motion.

In essence, the boundary condition is translated into an initial condition and another relaxed boundary condition by the trans-
form model. Accordingly, the external loads on both pivots can be given in the local coordinate system.

8
< f 11 ¼ f 1 cosθ12 −p1 sinθ12
p ¼ p1 cosθ12 þ f 1 sinθ12 ð12:1Þ
: 11
m11 ¼ m1

8
< f 12 ¼ −f 1 cosθ12 þ p1 sinθ12
p12 ¼ p1 cosθ12 þ f 1 sinθ12 ð12:2Þ
:
m12 ¼ m1 þ f 1 f2ðμ−aÞ þ ½1−2ðμ−aÞ cosθ12 g−p1 ½1−2ðμ−aÞ sinθ12

where suffixes 11 and 12 refer to the module number; for the load suffix 1 denotes the chain number. Then the parasitic motion of
the kinematic chain can be obtained.

dy1 ¼ ðdy11 þ dy12 Þ cosθ12 −ðdx11 −dx12 Þ sinθ12 −ðμ−aÞ½1− cosðθ11 þ θ12 Þ−½1−2ðμ−aÞð1− cosθ12 Þ ð13Þ

The primary motion is

dx1 ¼ ðdy11 þ dy12 Þ sinθ12 þ ðdx11 −dx12 Þ cosθ12 þ ðμ−aÞ sinðθ11 þ θ12 Þ þ ½1−2ðμ−aÞ sinθ12 ð14Þ

It should be noted that the external forces f11 and f12 are equal and opposite, but the forces p11 and p12 are entirely identical.
Therefore, the following expressions can be obtained according to Eqs. (10.1) and (10.2).

     
2 2
dy11 þ dy12 ¼ C yθ θ11 þ θ12 þ HyθX11 þ H yθX12 þ C yf 11 −C yf X12 ðf 1 cosθ12 −p1 sinθ12 Þ
 
þ C ypX11 þ C ypX12 ðp1 cosθ12 þ f 1 sinθ12 Þ ð15:1Þ

   
3 3
dx11 −dx12 ¼ C xθ θ11 −θ12 þ ðH xθX11 −HxθX12 Þ þ C xf X11 þ C xf X12 ðf 1 cosθ12 −p1 sinθ12 Þ
 
þ C xpX11 −C xpX12 ðp1 cosθ12 þ f 1 sinθ12 Þ ð15:2Þ

The external forces on each building block have been transformed into the forces at the moving end of the kinematic chain, by
using the load relationships Eq. (12.1) and Eq. (12.2), so the internal forces in the chain are eliminated.

Table 1
Dimensional parameters and Young's modulus of the material in the mechanism.

L/mm T/mm W/mm E/Pa

50 0.5 6 0.73 × 1011


22 H. Zhao et al. / Mechanism and Machine Theory 50 (2012) 15–28

4.3. The model of the linear-motion mechanism

The compliant linear-motion mechanism can be treated as compliant parallel four-bar mechanism, hence the loads, applied on
the moving platform and on the moving ends of the two kinematic chains, must satisfy the relationship as

8
< f1 þ f2 ¼ f
p1 þ p2 ¼ p ð16Þ
:
m1 þ m2 þ ðp1 −p2 Þs ¼ m:

The geometric compatibility condition between the moving platform and the moving ends of the two kinematic chains can be
obtained.


dx ¼ ðdx1 þ dx2 Þ=2
ð17Þ
dy ¼ ðdy1 þ dy2 Þ=2

For a typical linear-motion mechanism, θ12 and θ22 are approximately identical, so according to Eqs. (15.1) and (15.2), the
following expressions can be rearranged as

   
ðdy11 þ dy12 Þ þ ðdy21 þ dy22 Þ≈C yθ θ211 þ θ212 þ θ221 þ θ222 þ H yθX11 þ HyθX12 þ H yθX21 þ H yθX22
   
C yf X11 −C yf X12 þ C yf X21 −C yf X22 C ypX11 þ C ypX12 þ C ypX21 þ C ypX22 ð18:1Þ
þ ðf cosθ12 −p sinθ12 Þ þ ðp cosθ12 þ f sinθ12 Þ
2 2

 
3 3 3 3
ðdx11 −dx12 Þ þ ðdx21 −dx22 Þ ¼ C xθ θ11 −θ12 þ θ21 −θ22 þ ðHxθX11 −H xθX12 þ HxθX21 −H xθX22 Þ
   
C xf X11 þ C xf X12 þ C xf X21 þ C xf X22 C xpX11 −C xpX12 þ C xpX21 −C xpX22 ð18:2Þ
þ ðf cosθ12 −p sinθ12 Þ þ ðp cosθ12 þ f sinθ12 Þ:
2 2

Meanwhile, the following relationship exists for this compliant linear-motion mechanism.

θ11 ≈−θ12 ≈θ21 ≈−θ22 ð19Þ

So Eqs. (18.1) and (18.2) can be further rearranged as

2
ðdy11 þ dy12 Þ þ ðdy21 þ dy22 Þ≈4C yθ θ12 þ H yθ þ C yf ðf cosθ12 −p sinθ12 Þ þ C yp ðp cosθ12 þ f sinθ12 Þ ð20:1Þ

3
ðdx11 −dx12 Þ þ ðdx21 −dx22 Þ ¼ −4C xθ θ12 þ Hxθ þ C xf ðf cosθ12 −p sinθ12 Þ þ C xp ðp cosθ12 þ f sinθ12 Þ: ð20:2Þ

The expression for the higher order terms (Hxθ and Hyθ), and the compliances (Cxf, Cxp, Cyf, and Cyp) are listed in Appendix A,
owing to the complex form.

Table 2
Geometric parameters of five typical compliant linear-motion mechanisms using in FEA.

Mechanism No. α a μ

Mech 1 π/3 0.6 0.800


Mech 2 π/4 0.5 0.778
Mech 3 π/6 0.4 0.738
Mech 5 π/4 0.1 0.2
H. Zhao et al. / Mechanism and Machine Theory 50 (2012) 15–28 23

Fig. 9. The finite element model under deformation.

It is worth while to note that the internal forces are entirely eliminated by the load relationship (Eq. 16) and the geometric
condition (Eq. 17). So the developed model allows parametric insight into the properties of the compliant linear-motion mech-
anism. Moreover, the advantages of the building blocks method are displayed. Thus this method can be extensively exploited
in the design of complex flexural pivot and compliant mechanism.

a) Mech1

b) Mech2

Fig. 10. Primary motion versus parasitic motion.


24 H. Zhao et al. / Mechanism and Machine Theory 50 (2012) 15–28

Fig. 11. Comparison of parasitic motion for Mech 3–Mech 5.

Furthermore, substituting Eqs. (20.1) and (20.2) into Eq. (17) will result in the parasitic motion and primary motion for the
compliant linear-motion mechanism.
" #
ðdy1 þ dy2 Þ 1 ðθ þ θ12 Þ2 þ ðθ21 þ θ22 Þ2
dy ¼ ¼ f½ðdy11 þ dy12 Þ þ ðdy21 þ dy22 Þ cosθ12 −½ðdx11 −dx12 Þ þ ðdx21 −dx22 Þ sinθ12 −ðμ−aÞ 11
2 2 2
! ð21:1Þ
−½1−2ðμ−aÞ
θ212 þ θ222
2 g
ðdx1 þ dx2 Þ 1
dx ¼ ¼ f½ðdy11 þ dy12 Þ þ ðdy21 þ dy22 Þ sinθ12 þ ½ðdx11 −dx12 Þ þ ðdx21 −dx22 Þ cosθ12 þ ðμ−aÞ½ sinðθ11 þ θ12 Þ
2 2 ð21:2Þ
þ sinðθ21 þ θ22 Þ þ ½1−2ðμ−aÞð sinθ12 þ sinθ22 Þg

5. Analysis of the compliant linear-motion mechanism

5.1. The compensation condition for parasitic motion

It is worthy of note that the parasitic motion of the compliant linear-motion mechanism will decrease significantly when the
following condition is satisfied. Because using the dominant terms of these flexural pivot, the inherent parasitic motion of the par-
allel four-bar mechanism is compensated for, according to Eqs. (20.1)–(21.2).

1−2ðμ−aÞ ¼ 4C yθ ð22Þ

Rearranging Eq. (22) in terms of Eq. (11.2), the compensation condition for parasitic motion is obtained. Therefore the para-
sitic error will be diminished greatly when the geometric parameters μ, a, and α satisfy the following equation.
  h i
2 2 2 2
4−30 cos α μ þ 15 cos α ð2a þ 1Þ−36a μ þ 36a ¼ 0 ð23Þ

Fig. 12. Compliant double parallel four-bar mechanism.


H. Zhao et al. / Mechanism and Machine Theory 50 (2012) 15–28 25

Meanwhile, the geometric parameters are subject to the following range


 pffiffiffi  pffiffiffi
0bμ−ab1=2 and 3− 5 =6bλb 3 þ 5 =6: ð24Þ

Then the parameter space is illustrated in Fig. 7. Here the half angle of intersection α is limited in the range of [π/6, π/3], in
light of the practical application.
Fig. 7 can be utilized as a design tool to choose the suitable geometric parameters, in order that the dominant term of the par-
asitic motion is compensated for. In order to design a linear-motion mechanism with large stroke, the primary motion is also
taken into account in what follows.

5.2. The characteristics of parasitic motion and primary motion

From the observation of Eqs. (20.1), (20.2), and (21.2), the primary motion can be expressed as
0 0 0 0
dx ¼ C xθ θ12 þ Hx þ C xf f þ C xp p: ð25Þ

As far as the contribution of the dominant term is concerned, the primary motion can be approximately calculated as
0
dx≈C xθ θ12 : ð26Þ

If the rotational stroke of a single pivot is prescribed, the linear-motion range for the mechanism is determined by the coeffi-
cient of the dominant term. The physical meaning of the coefficient is the length of rigid link, which is illustrated in Fig. 2. So this
coefficient can be obtained by calculating the distance between the two initial intersection points on a kinematic chain.
0
C xθ ¼ 1−2ðμ−aÞ ð27Þ

On the other hand, the geometric parameters are subject to the compensation condition, in order to reduce the parasitic mo-
tion. So the two factors are taken into account to satisfy the requirement in primary motion and parasitic motion simultaneously.
When the half angle of intersection α is π/6, π/4, and π/3, the curves of Eq. (23) are illustrated in Fig. 8. In addition, the straight
line (Eq. 27) is also depicted in the figure when C′xθ is zero. Thus the design of a compliant linear-motion mechanism with high
performance is translated into the following optimization problem:
If a point (determined by the two parameters μ and a) is designed on the curve and the distance from this point to the line is
farthest, the linear-motion mechanism will achieve the largest motion range as well as small parasitic error.
In terms of Fig. 8, the range of motion will increase with the increasing of the geometric parameter α. So for a designer, the
angle of the intersection needs to be chosen as large as possible.
Similarly, the parasitic motion after compensation can be expressed as the following form
0 0 0
dy ¼ H y þ C yf f þ C yp p: ð28Þ

According to Eqs. (25) and (28), the physical meanings are captured for both primary motion and parasitic motion: the dom-
inant terms and the higher order terms are the kinematic components arising from the parallel four-bar mechanism; these com-
pliances, C′xf, C′xp, C′yf, and C′yp, are related to both the load effect and kinematic behavior, thus they reveal the compliant
properties of this linear-motion mechanism.
As the load effect is evaluated, the evidence is given that the external moment indirectly contributes to the primary motion
and parasitic motion through the rotational displacement of flexural pivots. Nevertheless, the external forces f and p not only de-
termine the kinematic behavior of these flexural pivots, but also influence the linear-motion mechanism directly. Furthermore,
the conclusion in our previous work [19] can be taken account of, because only the flexural building blocks contribute to the
load effect. So the parasitic motion dy is mainly influenced by the vertical force p, especially when the dominant term vanishes.
However, the vertical force p affects the primary motion slightly, according to the magnitude demonstrated in Appendix A. It
needs to be emphasized again that the evaluation of load effect must satisfy the compensation condition.

5.3. The comparison of parasitic motion

In order to display the advantages in parasitic motion, two kinds of compliant parallel four-bar mechanism, which are con-
structed by two leaves (Fig. 1(a)) and four cartwheel flexural building blocks (Fig. 1(b)), are introduced for performance compar-
ison. Here they are named as configuration II and configuration III respectively, and the proposed linear-motion mechanism are
named as configuration I.
First of all, the three configurations derive from the parallel four-bar mechanism, so all the primary motions can be approxi-
mated as Eq. (26), even though the coefficients C′xθ are different. Thus these primary motions are the order of θ12.
26 H. Zhao et al. / Mechanism and Machine Theory 50 (2012) 15–28

The parasitic motion for configuration I is compensated for, so the higher order term H′y is of the order of θ 4 12 in light of
Eq. (28) and Appendix A. Consequently, the relationship between the primary motion and parasitic motion is
 
4
dy ¼ O dx : ð29Þ

According to Ref. [5], the following relationship exists for configuration II

3 2  
2
dy ¼ − dx ¼ O dx : ð30Þ
5

As far as configuration III is concerned, the parasitic motion can be derived as illustrated in Fig. 2. The flexural building block
exploited is the cartwheel flexural pivot, and the center shift in the y direction will be obtained readily in terms of Ref. [1].

2
θ
dyp ≈− : ð31Þ
30 cosα

After the parameter transformation, the parasitic motion for configuration III can be approximately obtained by taking advan-
tage of the principle as illustrated in Fig. 2.

2 
θ12 ð1−μ Þ 2 μ ð1−μ Þ 2
dy≈− − θ12 ¼ − þ θ12 ð32Þ
15ξ cosα 2 15 cos α
2 2

The parameter μ is in the range (0, 1), so the inherent parasitic motion of the rigid link can not be compensated for by the flex-
ural building blocks. Moreover, the relationship between primary motion and parasitic motion is
 
2
dy ¼ O dx : ð33Þ

In all, the parasitic motion of the proposed compliant linear-motion mechanism is diminished significantly by the compensa-
tion principle, and its advantages in precision engineering is pronounced in terms of the performance analysis for the three linear-
motion mechanisms. Furthermore, the quantitative comparison will be carried out by simulation data in the following section.

6. FEA verification

In order to verify the validity of the model, FEA is implemented using commercial package ANSYS, which is suitable for large-
deflection and nonlinear analysis. In order to implement the simplified procedure, a model is drawn considering all the members
as lines, and BEAM3 element is selected for all members. This element has three degrees of freedom at each node: translations in
the nodal x and y directions and rotation about the nodal z-axis. The platforms of mechanism and the stages of pivots are simpli-
fied as some rigid beams that connect the leaves. In order to make these rigid beams as stiff as possible, value of Young's modulus
considered for these rigid beams is thousand times higher than the value considered for the leaves, and the former is five times
thicker than the latter. So the relative deflection for the platforms and stages can be negligible.
For all of the leaves in the compliant linear-motion mechanisms, the shape parameters and material properties are the same,
which are listed in Table 1. Five typical mechanisms with different geometric parameters are simulated, as listed in Table 2. For
Mech 1, the finite element model under deformation is illustrated in Fig. 9
Firstly, the primary motion versus parasitic motion is simulated for Mech 1 and Mech 2, as shown in Fig. 10. Here the horizon-
tal force for driving Mech 1 and Mech 2 is up to 3.8356 and 6.5753 respectively, which is not indicated in the figure. It is a remark-
able fact that the parasitic motion dy is affected greatly when the vertical force (payload) is applied. Nevertheless, the primary
motion is slightly influenced by the payload. These characteristics mentioned in the previous section are therefore verified. Mean-
while, the model data is in good agreement with the data obtained by FEA, even if the payload is large enough.
Furthermore, two compliant linear-motion mechanisms are introduced to compare the parasitic error with Mech 3. Here Mech
4 and Mech 5 are the compliant parallel four-bar mechanism constructed by two leaf-springs (Fig. 1(a)) and four cartwheel flex-
ural building blocks (Fig. 1(b)) respectively. The geometric parameters used for Mech 5 are listed in Table 2. Consequently, as
demonstrated in Fig. 11, the parasitic motion for Mech 3 is greatly reduced by the compensation of the generalized cross-
spring pivots. And the precision advantages of the proposed linear-motion mechanism are remarkable on the basis of quantitative
comparison.

7. The compliant double parallel four-bar mechanism

As shown in Fig. 12, the compliant double parallel four-bar mechanism can be combined by the proposed linear-motion mech-
anism, in order to further diminish the parasitic motion. Meanwhile, the range of motion is increased by reducing the stress in the
generalized cross-spring pivot, and the material can be fully utilized.
H. Zhao et al. / Mechanism and Machine Theory 50 (2012) 15–28 27

Of course, the higher-precision compound compliant linear-motion mechanism can be designed by making use of this mech-
anism as a flexural module. Therefore, the future of this linear-motion mechanism in precision engineering is promising.

8. Conclusion

In this paper, the center shift of the generalized cross-spring pivot is utilized to compensate for the inherent parasitic motion
of the parallel four-bar mechanism. The model for primary motion and parasitic motion is developed based on the building block
method, and the internal loads are eliminated by taking account of the load relationship between the flexural modules and the
two kinematic chains. Then the compensation condition for parasitic motion is derived, and the design space for geometric
parameters a, μ, and α is illustrated by a graph. In order to achieve the large range of motion and diminished parasitic error simul-
taneously, the approach to choose the suitable parameters is proposed. The evidence is given that a large angle of intersection will
satisfy the requirement. In addition, in terms of the evaluation of load effect, the payload influences the parasitic motion signif-
icantly, but affects the primary motion slightly.
Furthermore, the validity and effectiveness of the developed model are verified by FEA. The simulation data displays the ad-
vantages of the proposed compliant linear-motion mechanism, in light of the comparison with other two configurations of the
complaint parallel four-bar mechanism.
Meanwhile, the compliant double parallel four-bar mechanism is designed by combining the proposed linear-motion mecha-
nism as a flexural module, for the sake of diminishing the parasitic motion further. The more complex compliant linear-motion
mechanism will be constructed based on the building block approach. Consequently, the prospect of this rectilinear mechanism
in precision engineering is promising.

Acknowledgments

The authors gratefully acknowledge the financial support of National Natural Science Foundation of China (Grant Nos.
51105014, 50975007) and Ph.D. Programs Foundation of Ministry of Education of China (Grant No. 20091102110023).

Appendix A

In Eqs. (20.1) and (20.2), the expression for the higher order terms (Hxθ and Hyθ) and the compliances (Cxf, Cxp, Cyf, and Cyp) are
rearranged. These expressions are valid for the generalized cross-spring pivot whose geometric parameter λ is in the range [0,
0.5]. When the position of the intersection point is in the range (0.5, 1], the result can be easily obtained in terms of Eqs. (3)
and (4), but they are not listed here due to the complexity in the form.

!
2
2400 cosα ð2λ−1Þθ12 ξ 2
H xθ ¼ þ A θ
2 12
A3 ξ sin2 α d
ðA:1Þ
5
θ12 h i 
2 2 2
− −1200A1 sin α−120 cos α ð12λ−1Þð2λ−1Þ 9λ þ 15λ−1
6300A3 ξ cos α
3

" #
1 −6 tan2 α 2 4  4 3 2

4
H yθ ¼ A1 þ 2592λ −3024λ þ 1338λ −241λ þ 2 θ12
2ξ cosα 5 1500
! !

f
1 3 tan2 α 4θ212 ξ2 h i
2 2 2 2
þ 1− A1 θ12 þ A2 θ12 −1200A1 sin α−120 cos α ð12λ−1Þð2λ−1Þ
2ξ cosα 10 A3 cos2 α d ðA:2Þ

" ! #
þ
4800ð2λ−1Þθ212 cot2 α 12λ−1 ξ2
A3 10 d
2 A tan2 α
þ A2 θ12 þ 1
1400
2
ð2λ−1Þθ12 g
!
2
100 ξ 2
C xf ¼ þ A2 θ12 ðA:3Þ
ξ3 A3 sin2 α d

" !#
1 −θ312  2  10ð12λ−1Þθ
12 ξ2 2
C xp ¼ 9λ þ 15λ−1 þ þ A2 θ12 ðA:4Þ
A3 ξ3 126 cos α
2
sin2 α d

!
10 3 tan2 α 2
C yf ¼ 1− A 1 θ12
A ξ3 cos2 α 10
( 3 ! " ! # ) ðA:5Þ
2
ξ 2 2 12λ−1 ξ2 2 tan2 α 2
−ð12λ−1Þθ12 þ A2 θ12 þ 10 cot α þ A2 θ12 þ ð2λ−1ÞA1 θ12 θ12
d 10 d 1400
28 H. Zhao et al. / Mechanism and Machine Theory 50 (2012) 15–28

" # !
2 2
100 3 tan α 2 ξ 2
C yp ¼ 1− A1 θ12 þ A2 θ12 ðA:6Þ
ξ2 A3 cos2 α 10 d

where

−2  2 
A1 ¼ 9λ −9λ þ 1
15
1  2 
A2 ¼ 9λ −9λ þ 11
6300
2 2
A3 ¼ ð12λ−1Þ θ þ 100:

The geometric parameter can be transformed as the following relationship

λ ¼ 1−a=μ:

References

[1] S.T. Smith, Flexures: Elements of Elastic Mechanisms, Gordon and Breach Science Publishers, New York, NY, 2000.
[2] L.L. Howell, Compliant Mechanisms, Wiley, New York, 2001.
[3] J.M. Paros, L. Weisbord, How to design flexure hinges, Machine Design 37 (1965) 151–156.
[4] N. Lobontiu, Compliant Mechanisms: Design of Flexure Hinges, CRC Press, Boca Raton, FL, 2002.
[5] S. Awtar, A.H. Slocum, E. Sevincer, Characteristics of beam-based flexure modules, ASME Journal of Mechanical Design 129 (6) (2007) 625–639.
[6] S. Henein, P. Spanoudakis, S. Droz, L.I. Myklebust, E. Onillon, Flexure pivot for aerospace mechanisms, 10th European Space Mechanisms and Tribology
Symposium, San Sebastian, Spain, 2003.
[7] D. Chakarov, Study of the passive compliance of parallel manipulators, Mechanism and Machine Theory 34 (3) (1999) 373–389.
[8] B.P. Trease, Y.-M. Moon, S. Kota, Design of large-displacement compliant joints, ASME Journal of Mechanical Design 127 (4) (2005) 788–798.
[9] P. Spanoudakis, P. Schwab, P. Johnson, Design and production of the METOP satellite IASI corner cube mechanisms, 10th European Space Mechanisms and
Tribology Symposium, San Sebastian, Spain, 2003.
[10] S. Henein, I. Kjelberg, S. Zelenika, Flexible bearings for high-precision mechanisms in accelerator facilities, Proceedings of NANOBEAM 2002- 26th Advanced
ICFA Beam Dynamics Workshop on Nanometre-Size Colliding Beams, Lausanne, Switzerland, 2002.
[11] B.R. Cannon, T.D. Lillian, S.P. Magleby, L.L. Howell, M.R. Linford, A compliant end-effector for microscribing, Precision Engineering 29 (1) (2005) 86–94.
[12] N.B. Hubbard, J.W. Wittwer, J.A. Kennedy, D.L. Wilcox, L.L. Howell, A novel fully compliant planar linear-motion mechanism, Proceedings ASME IDETC/CIE
2004, Salt Lake City, Utah, USA, 2004.
[13] G.A. Roman, G.J. Wiens, MEMS optical force sensor enhancement via compliant mechanism, Proceedings ASME IDETC/CIE 2007, Las Vegas, USA, 2007.
[14] S.H. Chang, S.S. Li, A high resolution long travel friction drive micropositioner with programmable step size, The Review of Scientific Instruments 70 (6)
(1999) 2776–2782.
[15] Y.C. Wang, S.H. Chang, Design and performance of a piezoelectric actuated precise rotary positioner, The Review of Scientific Instruments 77 (10) (2006)
105101-1 –105101-5.
[16] R. Duarte, M.R. Howells, Z. Hussain, et al., A linear motion machine for soft x-ray interferometry, Proceedings of the SPIE 1997 Conference, San Diego, CA,
1997.
[17] K.B. Choi, J.J. Lee, M.Y. Kim, Cartwheel flexure-based compliant stage for large displacement driven by a stack-type piezoelectric element, International
Conference on Control, Automation and Systems 2007, COEX, Seoul Korea, 2007.
[18] H.Z. Zhao, S.S. Bi, Stiffness and stress characteristics of the generalized cross-spring pivot, Mechanism and Machine Theory 45 (3) (2010) 378–391.
[19] H.Z. Zhao, S.S. Bi, Accuracy characteristics of the generalized cross-spring pivot, Mechanism and Machine Theory 45 (10) (2010) 1434–1448.

Vous aimerez peut-être aussi