Microsyst Technol

DOI 10.1007/s00542-017-3423-8

REVIEW PAPER

Design and analytical analysis of a large‑range tri‑symmetrical

2R1T compliant mechanism
Guangbo Hao1,2,3 · Fukang Dai1,2 · Xiuyun He3 · Yufei Liu1,2

Received: 31 March 2017 / Accepted: 19 April 2017

© Springer-Verlag Berlin Heidelberg 2017

Abstract This paper details design, modelling and analy- assembly (Howell 2001; Howel et al. 2013). A compliant
sis of a class of compliant mechanisms with tri-symmetri- mechanism can be regarded as a traditional mechanism
cal planar structures, which are able to achieve three large- coupled with spring with linear or nonlinear stiffness, from
range out-of-plane motions. Each compliant mechanism, the point of view of a pseudo-rigid body model (Howell
using identical wire beams with square cross sections, 2001). It can have lumped compliance or distributed com-
consists of three double 2-beam modules in a tri-symmet- pliance, and its stiffness can be of positive, negative or zero
rical way. Each double 2-beam module has an embedded stiffness (aka constant force) (Howell 2001; Howel et al.
serial configuration with a common intersection point of 2013). Due to these merits and performance characteristics,
beams. Based on assumptions of small deformations, ana- compliant mechanisms have gained successful applica-
lytical compliance models of a generic mechanism are then tions, which can be generally broken down into seven cat-
obtained to consider all design parameters. Finally, influ- egories (Hao et al. 2016) including (1) precision motion/
ences of main parameters on the compliances (and there- positioning stages, (2) metrology instruments, (3) medical/
fore degrees of freedom or degrees of constraint) are fully health devices, (4) MEMS/NEMS, (5) energy harvesting/
analysed followed by design insights. vibration absorption, (6) material and structure design, and
(7) design for no assembly. This paper focuses on a class
of 2R1T (R: revolute, T: translational) compliant mecha-
1 Introduction nisms, which are required to provide three large-range tip-
tilt-piston (out-of-plane) motions. This type of compliant
Compliant mechanisms transfer motion/load/energy by mechanism can be typically used as micro-mirror arrays
deformation of their materials, which are jointless design or a force/moment sensor (Hopkins and Panas 2014; Zhao
(Howell 2001; Howel et al. 2013). They can offer mer- et al. 2016).
its such as no backlash, no friction, no wear, and free of When designing a compliant mechanism, it is required
to have a large stiffness/small compliance in the DOC
(degrees of constraint) direction, and have a small stiff-
* Guangbo Hao ness/large compliance in the DOF (degrees of freedom)
G.Hao@ucc.ie direction, towards the target that the stiffness in the DOC
* Yufei Liu direction is commonly 100 times larger than that in the
yufei.liu@cqu.edu.cn DOF direction (Hao and Kong 2013). In addition, it is
1 desirable to design a symmetrical but compact structure in
Key Laboratory of Optoelectronic Technology and Systems
(Chongqing University), Ministry of Education, Chongqing, order to minimize the parasitic motion (unwanted motion
China in the DOC direction) in a low-mobility compliant mech-
2
College of Optoelectronic Engineering, Chongqing anism (Hao and Li 2015; Hao 2017a, b). Symmetry can
University, Chongqing 400044, China also improve the mechanism’s performance characteristics
3
School of Engineering‑Electrical and Electronic Engineering, in other aspects, such as reducing the moment of inertia
University College Cork, Cork, Ireland for better dynamics, alleviating the thermal and buckling

13
 Microsyst Technol

sensitivity (Panas and Hopkins 2015) for better robust- as an individual group was employed to replace each
ness. Hao (2017b) has proposed a class of tri-symmetrical 2-beam module that employed in Hao (2017b), in order
2R1T compliant mechanisms using identical wire beams, to improve the motion range (also reduce the actuation
whose motion range, however, is limited, and may result in stiffness). The double 2-beam module has a common
nonlinear stiffness characteristics over large deformation. intersection point so that there are three non-overlapping
It is therefore the motivation of this paper to present and intersection points from three double 2-beam modules for
analyse a new tri-symmetrical design towards an improved constraining the undesired in-plane rotation. Each group
motion range with a compact configuration. has an embedded serial arrangement for compactness
This paper is organised as follows. Section 2 presents consideration. Note that a secondary stage has been intro-
the new generic large-range tri-symmetrical design using duced in the double 2-beam module, which is between
three double 2-beam modules with embedded serial lay- the base and the motion stage. The secondary stage may
outs, which considers conditions to avoid interferences of help mitigate nonlinear effects over large deformation
the mechanism. The analytical parametric modelling is fol- such as generating an approximate constant primary stiff-
lowed in Sec. 3. Section 4 provides an assessment of main ness even over large range of motion (Xie et al. 2016).
parameters on the compliances of the mechanism. Conclu- Due to the use of tri-symmetry, the generic design
sions are drawn in Sec. 5. is only defined by six independent geometrical param-
eters (t, l, α, bI, bE, rb) that are indicated in Fig. 1.
Herein, rb denotes the overall dimension (footprint)
2 Design of a large‑range tri‑symmetrical of the mechanism. We also have relations: bI < bE, and
mechanism β = arccot[cot(α)bE/bI] + π. In order to make the design
more compact, we need to impose limitation on two
The large-range tri-symmetrical 2R1T mechanism with angles: α ∊ (π, 2π), β ∊ (π, 2π). If α = 3π/2 (β = 3π/2),
a planar configuration is presented in Fig. 1, based on a specific case of the proposed design is shown in Fig. 2.
the constrained-based design method (Awtar and Slo- The following conditions should be firstly met to
cum 2007; Hale 1999; Li et al. 2016). It consists of effectively constrain the unwanted in-plane rotation of
12 identical wire beams in the same plane, which are the mechanism:
arranged in an appropriate way to constrain the in-plane 
DOF. The present design is an improved version of the ri = |rb + 0.5bI tan α| >> 0
design in Hao (2017b), where a double 2-beam module
, (1)
bI >> 0 if α is very close to 3π/2

bE
Group 1 Note: all beams are
bI Secondary
identical with square
stage cross sections: length l
and thickness t

Base Base X1-E1 Y1-E2 X1-E2

Oi
Y1-E1
ds Y1-I1
ri Y rb Base
db Base
Y1-I2
X1-I1 X1-I2
dm
O
Base X Base
2 /3 Y
Motion stage
Group 2 Group 3 X
O
Four local coordinate
Base Base systems of group 1-double
2-beam module

Fig. 1  A large-range tri-symmetrical 2R1T compliant mechanism with a generic structure

13
Microsyst Technol

Fig. 2  A special case (α = β = 3π/2). a Top View b 3D view

where the first equation refers to that the intersection lower-case symbols. Note that all parameters used in
position relevant to the motion stage centre should be Fig. 1 are lower-case ones, meaning that l = 1, bI = BI/L;
much larger than zero. bE = BE/L; t = T/L, rb = Rb/L, and ri = Ri/L.
Then, we need to satisfy the following conditions to Let the global coordinate system (O-XYZ) be set
avoid interferences of the mechanism: up at the centre of motion stage, and let external load-
 ing vector (fs − x, fs − y, fs − z, ms − x, ms − y, ms − z) and
bI ≥ 2l cos α if α ∈ (3π/2, 2π ) ⇔ α ≤ 2π - arccos(bI /2/l)
the corresponding displacement vector (xs, ys, zs, θs − x,

√



 dm = 2[ 3rb /2 − bI /4 + l sin(α + π/6)] ≥ 0 θs − y, θs − z) both defined at the same point. The stiffness

√ ,

 db = 2[ 3rb /2 − bE /4 + l sin(β + π/6)] ≥ 0 matrix of the first double 2-beam module (group 1) is
√ first derived with respect to the global coordinate system


(2)

ds = 2( 3rb /2 − bE /4) ≥ 0

as below:
where the first equation guarantees that the two internal
beams in each group have no interference; the second equa- KG1 = [(TI1 )T · K · TI1 + (TI2 )T · K · TI2 ]−1
tion vouchers that the internal beams between two groups (3)
+ [(TE1 )T · K · TE1 + (TE2 )T · K · TE2 ]−1 .
have no interference; the third equation can make sure that
the external beams between two groups have no interfer- In Eq. (3), K is the stiffness matrix of each wire beam
ence; the final equation finally guarantees the three second- with respect to its own local coordinate system in group
ary stages not to interfere each other. 1 (Hao and Kong 2013) (there are four local coordinate
systems for group 1 as shown in Fig. 1), as shown below:
 
d 0 0 0 0 0
3 Analytical modelling 0
 12 0 0 0 −6 

0 0 12 0 6 0 
K= , (4a)
Analytical modelling is carried out in this section to ena- 0
 0 0 δ 0 0 
ble performance quick analysis in the next section. As 0 0 6 0 4 0 
recommended in (Hao and Kong 2013), the normalisation 0 −6 0 0 0 4
method is used to effectively process all parameters for
derivations, which is described as follows. Length param- where d = 12/t2, δ = 1.69 G/E (G is shear modulus of
eters are divided by beam’s actual length L, force param- material) (Hao 2017b).
eters are divided by EI/L2 (E is material’s Young’s modulus, In Eq. (3),TIi or TEi (i = 1 or 2) are the transformation
and I = T4/12 is the area moment of inertia of square cross matrix of an internal/external wire beam from its local
sections), and moment parameters by EI/L. All parameters coordinate system in group 1 (Fig. 1) to the global coor-
after normalisation are represented by their corresponding dinate system, which is elaborated as below:

13
 Microsyst Technol

where Ti is the matrix for transforming group 1’s stiff-

 
cos(−θIi ) − sin(−θIi )
0 0 0 0
 sin(−θIi )
 cos(−θIi )0 0 0 0 ness matrix (Eq. (3) to group 2’s or group 3’s one with
 0 0 1 0 0 0 respect to the global coordinate system, which is detailed
TIi = 

 0 0 0 cos(−θIi ) − sin(−θIi ) 0 as:
 0 0 0 sin(−θIi ) cos(−θIi ) 0  
0 0 0 0 0 1 cos(−φi ) − sin(−φi ) 0 0 0 0
   sin(−φi ) cos(−φi ) 0 0 0 0
1 0 0 0 dzIi −dyIi 
 0 0 1 0 0 0

0 1 0 −dzIi 0 dxIi  Ti = 
 
  0 0 0 cos(−φi ) − sin(−φi ) 0 
0 0 1 dyIi −dxIi 0   
× , (4b)  0 0 0 sin(−φi ) cos(−φi ) 0 
0 0 0 1 0 0 
  0 0 0 0 0 1
0 0 0 0 1 0 
0 0 0 0 0 1 × (i = 2 or 3; φ2 = 2π/3, φ3 = 4π/3).
Finally, the 2R1T mechanism’s compliance matrix (a
  diagonal one) is obtained:
cos(−θEi ) − sin(−θEi ) 0 0 0 0
 sin(−θEi ) cos(−θEi ) 0 0 0 0  

 0 0 1 0 0 0
 c11 0 0 0 0 0
TEi =   0 c22 0 0 0 0 
 0 0 0 cos(−θEi ) − sin(−θEi ) 0  
   0 0 [c33 ] 0 0 0 
 0 0 0 sin(−θEi ) cos(−θEi ) 0 C = (K)−1 =
 0
, (6)
0 0 0 0 0 1  0 0 [c44 ] 0 0 
   0 0 0 0 [c55 ] 0 
1 0 0 0 dzEi −dyEi 0 0 0 0 0 c66
 0 1 0 −dzEi 0 dxEi 
 
 0 0 1 dyEi −dxEi 0 
× , (4c) where compliance entries c11, c22, and c66 correspond to
0 0 0 1 0 0   three DOC directions, which are the translation along the
0 0 0 0 1 0 
X-direction, the translation along the Y-axis, and the rota-
0 0 0 0 0 1
tion about the Z-axis, respectively. Compliance entries c33,
where θIi (θEi) denotes the rotation of the internal c44, and c55 correspond to three DOF directions, which are
(external) local coordinate system, O ­ 1 − IiX1 − IiY1 − Ii the translation along the Z-direction, the rotation about
­(O1 − EiX1 − EiY1 − Ei), with respect to the global coordi- the X-axis, and the rotation about the Y-axis, respectively.
nate system about the Z-axis (Fig. 1); dxIi, dyIi and dzIi Here, c11 = c22 and c44 = c55.
(dxEi, dyEi and dzEi) denote the translations of the inter- In Eq. (6), compliance entries (cii, i = 1,2,…,6) are
nal (external) local coordinate system with respect to the determined by 5 geometrical parameters (t, α, bI, bE,
global coordinate system along three-axes (Fig. 1). All rb) and 1 material parameter (δ). Thus, it is impossible
those position parameters of local coordinate systems are to derive symbolic forms for them (it is trial to obtain
calculated based on Fig. 1 as: the complex expressions), which are not presented in
dzIi = 0 (i = 1 or 2); θI1 = α, θI2 = 3π − α, this paper. Note that the numerical values for the com-
dxI1 = lcos(α) − bI/2, dyI1 = rb + lsin(α); pliance entries under specific geometrical conditions
dxI2 = −lcos(α) + bI/2, dyI2 = rb + lsin(α); are to be discussed in the following section. However, if
dzEi = 0 (i = 1 or 2); θE1 = β + π, θE2 = 2π − β, α = β = 3π/2 (Fig. 2), compliance entries in Eq. (6) are
dxE1 = lcos(β) − bE/2, dyE1 = rb + lsin(β); simplified to the following symbolic form:
dxE2 = −lcos(β) + bE/2, dyE2 = rb + lsin(β).
2
Using the results of Eqs. (3) and (4a–4c) and on the c11 = c22 = , (7a)
3(12 + d)
basis of the tri-symmetry property, we can therefore
derive the stiffness matrix of the whole compliant mecha-
1
nism with respect to the global coordinate system: c33 = , (7b)
36
T T
K = KG1 + (T2 ) · KG1 · T2 + (T3 ) · KG1 · T3 , (5)

13
Microsyst Technol

3bE2 + 3bI2 + 2δ
c44 = c55 = ,
3(9bE2 bI2 + 18bE2 rb2 + 18bI2 rb2 + 3bE2 δ + 3bI2 δ − 18bE2 rb − 18bI2 rb + 12rb2 δ + 6bE2 + 6bI2 − 12rb δ + δ 2 + 4δ)
(7c)
2(bE2 d + bI2 d + 8)
c66 = .
3(bE2 bI2 d 2 + 24bE2 rb2 d + 24bI2 rb2 d − 24bE2 rb d − 24bI2 rb d + 10bE2 d + 10bI2 d + 192rb2 − 192rb + 64) (7d)

4 Compliance analysis mentioned above. In Fig. 3, ri rises with the increase of α,

and reaches an infinite at α = β = 3π/2. After α = 3π/2, ri
In this section, the influences of several parameters on first decreases with the increase of α, and quickly reduces
compliances are studied. For convenience, in the following to zero, and then goes up again. It is shown that with the
analysis we can assign beam’s length and thickness as well decrease of rb, the α value for ri = 0 is delayed. The results
as material properties to have fixed values: L = 40 mm, in Fig. 3 can further provide insight into analysing Fig. 4.
T = 1 mm (d = 19,200); E = 69 GPa, G = 26 GPa (i.e., Figure 4a explains the relations between the com-
δ = 1.69 × 26/69). Under these conditions, from the quali- pliance ratio (c44/c66) and two parameters. The c44/c66
tative point of view, c11, c22, c33 do not change too much value gradually enlarges with the increase of α, and its
with other parameters, leaving us to study on c44 and maximum can be up to 3000. After the maximum, the
c66. In particular, from Eq. (7a–7d), we can observe that c44/c66 value goes down, and reduces to the minimum at
if α = β = 1.5π and d is fixed, c11, c22, c33 are all con- a value of α corresponding to ri = 0 (the minimal c44/c66
stants. Moreover, let bI = 0.8, bE = 1 for the follow- is approximate 2). After the minimum, c44/c66 increases
ing parameter analysis so that we only study on how the again. Since the c44/c66 value is significantly affected by
main parameters, rb and α, affect c44, c66 and ri. Note that the value of α, we should avoid selecting a value of α
β = arccot[cot(α)bE/bI] + π, which is determined by α close to that for ri = 0. As observed, rb also influences
only when bI = 0.8, bE = 1. c44/c66. In addition to guaranteeing a large enough value
During parameter analysis, we should always stick of c44/c66, there is a necessity to have a large enough c44
to the conditions as defined in Eq. (2) to avoid interfer- so that the large stiffness in the DOC direction and the
ence. Therefore, when all other parameters have fixed large compliance in the DOF direction can be gained
values, the value range of α is non-continuous, and simultaneously.
α ≤ 2π − arccos(bI/2) = 5.124. If α = 2π - arctan[rb/ Figure 4b interprets the influences of parameters,
(bI/2)], one can yield ri = 0, resulting in that the in-plane rb and α, on c44, which contributes to a better under-
rotation DOC degrades to a DOF, and c44 and c66 are both standing of Fig. 4a. When α = 2π − arctan[rb/(bI/2)]
the maximums. (i.e., ri = 0), c44 achieves its maximum. When rb = 1.6,
Figures 3 and 4 illustrate analysis results, with intermit- α = 2π − arctan[rb/(bI/2)] ≈ 4.96; when rb = 1.4,
tent curves due to the non-continuous value range of α as α = 2π − arctan[rb/(bI/2)] ≈ 4.99; when rb = 1.2,
α = 2π − arctan[rb/(bI/2)] ≈ 5.03. Apparently, with the
decrease of rb, the value of α corresponding to ri = 0 is
rb changing
10
postponed.
rb=1.6 The effect of two parameters on c66 is plotted in
9
rb=1.4 Fig. 4c. It is shown that c66 rises as α increases, and
8
rb=1.4 reaches its maximum at a value α for ri = 0, and then
7 rb=1.2
decreases.
From the above analysis in Figs. 4 and 5, it is con-
6 rb=1.2
cluded that (a) when α = β = 1.5π, ri has its maximum,
5 but c44/c66, c44 and c66 are not at their maximums; (b)
ri

4 when ri = 0, c44 and c66 both have their maximums, but

3
c44/c66 is at its minimum. Considering both easy manu-
facturability and desired compliance characteristics, it is
2
convincing that the specific design as shown in Fig. 2 is a
1 desired candidate.
0 In order to verify the analytical modelling results,
3 3.5 4 4.5 5 5.5
small-deformation FEA (finite element analysis) is car-
ried out for the specific case with α = β = 1.5π. We
Fig. 3  Influences of parameters on intersection position ri also let material be AL6061-T6 with E = 69 GPa,

13
 Microsyst Technol

Fig. 4   Influences of param-

eters on compliances
(a)

(b)

(c)

G = 26 GPa, and let actual geometrical parameters be was adopted for the FEA with fine meshing and others
assigned values: L = 30 mm, BI = 19 mm; BE = 25 mm; in default. The compliance values obtained by FEA are
T = 1 mm, Rb = 69.64 mm. Solidworks 2016 simulation shown in Fig. 5, which are in good agreement with those

13
Microsyst Technol

Fig. 5  FEA simulations and

comparisons with analytical
results. a FEA for translation
along Y-axis b FEA for transla-
tion along Z-axisc FEA for
rotation about Y-axis d FEA for
rotation about Z-axis

obtained by analytical modelling. It is observed that the alternative unit/segment to compose a continuous robot for
FEA compliance in each direction is slightly larger than high-flexibility use (Qiu et al. 2016).
the analytical one, which complies with our expecta- It is hoped that the findings in this paper will provide
tion. This is because FEA regards all members as elastic inspiration for the design and analysis of other types of
bodies while analytical modelling assumes base, motion symmetrical complaint mechanisms. Following up this
stage, and secondary stages to be ideal rigid bodies. work, nonlinear modelling and analysis with experimental
testing are to be investigated.

5 Conclusions Acknowledgement The work in this paper is supported by Visiting

Scholar Foundation of Key Laboratory of Optoelectronic Technol-
ogy and Systems (Chongqing University), Ministry of Education of
A large-range tri-symmetrical 2R1T compliant mecha- China, which is much appreciated.
nism has been proposed, analytically modelled, and para-
metrically analysed. Non-interference conditions have been Compliance with ethical standards
taken into account during parameter analysis. The derived
analytical results can be used to quickly assess the com- Conflict of interest The authors declare that they have no conflict of
interest.
pliance characteristics of any 2R1T compliant mechanism
based on specific design/manufacturing requirements.
The proposed design can not only be used as a stand- References
alone manipulator actuated by three non-contact magnetic
actuators, but also form the foundation to design other Awtar S, Slocum AH (2007) Constraint-based design of paral-
novel compliant mechanisms. For example, by adding an lel kinematic XY flexure mechanisms. ASME J Mech Des
extra translational constraint along the Z-direction (using a 129(8):816–830
Hale LC (1999) Principles and techniques for designing precision
wire beam), the 2R1T mechanism reduces to a 2R point- machines, Doctoral dissertation, Massachusetts Institute of
ing mechanism (Xie et al. 2016), which can be useful for Technology
space applications. One can combine two same 2R1T Hao G (2017a) Determinate design and analytical analysis of a class
mechanisms in parallel to obtain a linear guiding mecha- of symmetrical flexure guiding mechanisms for linear actuators.
ASME J Mech Des 139(1):012301
nism for designing voice coil actuators Hao (2017a; Kim Hao G (2017b) Determinate synthesis of symmetrical, monolithic tip-
et al. 2013). We can also use the 2R1T mechanism as an tilt-piston flexure stages. ASME J Mech Des 139(4):042303

13
 Microsyst Technol

Hao G, Kong X (2013) A normalization-based approach to the mobil- Li H, Hao G, Kavanagh R (2016) Position-space-based compli-
ity analysis of spatial compliant multi-beam modules. Mech ant mechanism reconfiguration approach and its application in
Mach Theory 59:1–19 reduction of parasitic motion. ASME J Mech Des 138(9):092301
Hao G, Li H (2015) Nonlinear analytical modeling and characteristic Panas RM, Hopkins JB (2015) Eliminating underconstraint in dou-
analysis of a class of compound multibeam parallelogram mech- ble parallelogram flexure mechanisms. ASME J Mech Des
anisms. ASME J Mech Robot 7(4):041016 137(9):092301
Hao G, Yu J, Li H (2016) A brief review on nonlinear modeling meth- Qiu C, Qi P, Liu H, Althoefer K, Dai JS (2016) Six dimensional com-
ods and applications of compliant mechanisms. Front Mech Eng pliance analysis and validation of orthoplanar springs. ASME J
11(2):119–128 Mech Des 138:042301
Hopkins JB, Panas RM (2014) Flexure design for a high-speed large- Xie Y, Pan B, Pei X, Yu J (2016) Design of compliant universal joint
range tip-tilt-piston micro-mirror array, Proc ASPE 29th Annual with linear stiffness. ASME 2016 International Design Engineer-
Meeting, Boston, MA, November ing Technical Conferences and Computers and Information in
Howel LL et al (2013) Handbook of compliant mechanisms. Wiley, Engineering Conference
New York Zhao Y, Jiao L, Weng D, Zhang D, Zheng R (2016) Decoupling prin-
Howell LL (2001) Compliant mechanisms. Wiley, New York ciple analysis and development of a parallel three-dimensional
Kim C, Song M-G et al (2013) Design of an auto-focusing actuator force sensor. Sensors 16:1506
with a flexure-based compliant mechanism for mobile imaging
devices. Microsyst Technol 19:1633–1644

13