IOP Conference Series: Materials Science and Engineering

PAPER • OPEN ACCESS

Kinematical analysis of a generalized Cardanic joint

To cite this article: F C Ciornei et al 2019 IOP Conf. Ser.: Mater. Sci. Eng. 477 012037

View the article online for updates and enhancements.

This content was downloaded from IP address 178.171.66.239 on 19/02/2019 at 01:28

International Conference on Applied Sciences IOP Publishing
IOP Conf. Series: Materials Science and Engineering 477 (2019) 012037 doi:10.1088/1757-899X/477/1/012037

Kinematical analysis of a generalized Cardanic joint

F C Ciornei, S Alaci, I C Romanu, I Mihai and G A Tibu

1
”Stefan cel Mare” University, Mechanics and Technologies Department, Suceava,
Romania, Universitatii str., no. 13, 720229 Suceava, Romania

E-mail: florina@fim.usv.ro

Abstract. The necessity of transmitting the rotation motion between two shafts is commonly
met in mechanical engineering. The most frequent solutions are represented mechanical
systems known as couplings. The coupling solutions presented in literature depend on two
main features: the relative position between the shafts and the variation of the transmission
ratio. One of the most popular couplings is the Cardanic joint meant to transmit motion
between intersecting axes. From technological point of view, to obtain the concurrence of all
axes of the joints of the coupling it is a very difficult task. To overcome this aspect, it is
assumed that the axes of the shafts are not intersecting and from the four joints of the Cardanic
coupling, only the input joint remains a rotation one; the other three joints transform into
cylindrical joints and thus allow for relative linear displacement of the elements besides
rotation. In this manner, the new mechanism is a RCCC one. The paper presents the positional
analysis of the mechanism applying the method of homogenous operators proposed by
Hartenberg and Denavit. The analytical expressions of all displacements from the joints of the
mechanism are obtained. The aspects referring to the obscurity occurring when inverse
trigonometric functions are involved in describing the displacements and the way to avoid
these are discussed.

1. Introduction
The main task of a mechanism is to transmit motion and power from the driving element to the final
one under well stipulated conditions [1]. The problem of transmission of rotation motion between two
shafts is often met in technology [2]. When a rigorously constant transmission ration is required, the
gear mechanisms are used [3], [4]. When the constant transmission ratio is not a stringent necessity,
the use of linkage mechanisms [5-7] which are more reliable and economical than gears is a handy
solution.
The kinematical study of a linkage mechanism supposes establishing the position of the mechanism
for a given position of the driving elements [8]. This means that for stipulated position parameters of
the driving elements, all displacements from the rest of kinematical pairs of the mechanisms are
required. Every method from technical literature is based on the condition of closed kinematical chain
from the structure of the mechanism [9], requirement imposed by the definition of a mechanism itself.
The condition of closed kinematic chain is founded on the manner of coordinate transformation of a
point using a set of coordinate systems attached to the elements of the chain, chosen in such way that
the first frame coincides to the last one. The condition of closed kinematic chain is expressed in a
manner that depends on the structure of the mechanism and the solutions depend on the mathematical
apparatus involved.

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
International Conference on Applied Sciences IOP Publishing
IOP Conf. Series: Materials Science and Engineering 477 (2019) 012037 doi:10.1088/1757-899X/477/1/012037

2. Methods applied in spatial kinematical analysis

For plane mechanisms [10], the condition of closed kinematic chain can be expressed in vector format
or using complex numbers. For spherical mechanisms [11], the closed loop condition is formulated via
3x3 type rotation matrices. The closing equation has the form:
R12 R23 .....Rn ,1  I 3 (1)
where Rk ,k 1 is the rotation matrix that brings the frame ( k ) over the frame ( k  1 ) and I 3 is the unit
matrix of third order. The general from of a rotation matrix which describes the rotation of  angle
around an axis of e versor is, according to Angeles [12]:
R  ee T  ( I 3  ee T ) cos   E sin  (2)

where E is the anti-symmetric matrix attached to the e versor. The condition to be fulfilled by two
rotation matrices in order to be identical is to have the angles and the versors of the axes, respectively,
identical. From here it results the conclusion that from the nine scalar equations provided by equation
(1) only three are independent. Another extremely useful instrument in expressing the spatial rotations
is the quaternions algebra, [13]. For the case of spatial mechanisms the closing condition has a form
identical to the relation (1) [14].
T12T23 .....Tn ,1  I 4 (3)

where Tk ,k  1 is the displacement matrix that superposes the ( k ) frame over the ( k  1 ) frame and
having the general form:
 Rk .k 1 d k ,k 1  (4)
Tk ,k 1   
 0 1 

where Rk ,k  1 has the same significance as in equation (1), d k ,k  1 is the vector that moves the origin
Ok over Ok  1 origin and I 4 is the unit matrix of fourth order. It is easily noticed that the relation (3)
supposes the identity between the rotation matrices and displacement vectors of the matrices from both
members of equation. Thus, a system of six scalar equations actually results. The kinematic study of
spatial chains was substantially simplified by Hartenberg and Denavit [15] who proved that for a
kinematic chain with lower cylindrical pairs, the number of necessary parameters needed for
stipulation of the relative position of two elements may be reduced from six to four by the suitable
selection of the position and orientation of the axes of the frames attached to the chains’ elements. To
complete this operation, it is required that the axes of the cylindrical pairs should be chosen as Oz
axes of the frames and each of the Ox axis should be along the common normal of two neighboring
axes, as in Figure 1, [15]. With the axes chosen in this manner, the ( k ) coordinate system can be
superposed over the ( k  1 ) frame by a roto-translation of parameters  k and sk followed by a roto-
translation of parameters  k ,k  1 and a k ,k  1 .

 Rkz,k 1 d kz,k 1   Rkx,k 1 d kx,k 1 

Tk ,k 1    
 0  0  (5)
 1  1 
The two displacement vectors are expressed under the form:
d kz,k 1  0 0 sk T
(6)
d kx,k 1  ak ,k 1 0 0 T

2
International Conference on Applied Sciences IOP Publishing
IOP Conf. Series: Materials Science and Engineering 477 (2019) 012037 doi:10.1088/1757-899X/477/1/012037

zk 1 zk
a k ,k  1
 k ,k  1 z k 1

k

xk ,k  1 sk
xk ,k 1

Figure 1. The Hartenberg Denavit convention

In order to obtain the rotation matrices from relation (1), this relation is applied for e  k ,    k
and e  i ,    k ,k  1 respectively, and it results:

cos   sin  0 
 
Rkz ,k 1   sin  cos  0  (7)
 
 0 0 1

and:
1 0 0 
 
Rkx,k  1  0 cos  k ,k  1  sin  k ,k  1  . (8)
 
0 sin  k ,k 1 cos  k ,k 1 
 
The following notations are introduced for a simpler writing manner:
sin   S (9)
cos   C
The Tk ,k  1 operator has the matrix:

C k  S k 0 0 1 0 0 ak .k  1 
   
 S k C k 0 0 0 C k ,k  1  S k ,k  1 0 
Tk ,k  1      (10)
 0 0 1 sk  0 S k ,k  1 C k ,k  1 0 
   
 1 0 0 1  0 0 0 1 

Another especially useful tool in kinematical analysis of spatial mechanisms is the dual numbers
concept introduced by Ball and Clifford [16]. Similarly to complex numbers use, they propose the
employment of numbers having the form:
x̂  x  x0 (11)
acknowledged as dual numbers where  is the imaginary unit defined by the Phillips equality [17]:
2 0 (12)

3
International Conference on Applied Sciences IOP Publishing
IOP Conf. Series: Materials Science and Engineering 477 (2019) 012037 doi:10.1088/1757-899X/477/1/012037

Based on the Kotelnikov transfer principle [16] that states: all relations deduced within spherical
trigonometry remain valid if the real numbers are replaced by dual numbers, any spatial mechanism
may be studied using a spherical mechanism. Founded on this remark, Yang [18] analyses the spatial
mechanisms with five elements employing matrices of 3x3 type with dual numbers. It is also Yang
who studies spatial mechanisms applying dual quaternions.

3. The kinematical analysis of the proposed mechanism

Next, the positional analysis of generalized Cardanic mechanism is presented. The mechanism of the
Cardanic transmission presented in Figure 2 is a spherical one.

3
4

Figure 2. Cardanic transmission

The structure of the mechanism contains four rotation pairs whose axes should be theoretically
concurrent in a point. This requirement is extremely restrictive and therefore, from technological point
of view, special tasks must be fulfilled. To avoid these complications, the kinematical pairs of the
mechanism (excluding the driving pair) are transformed into cylindrical pairs. Thus, the resulting
mechanism is a RCCC spatial one, with the degree of mobility 1, and which doesn’t require the
concurrence of the axes of two pairs. For the mechanism obtained in the above manner, applying a
series of particular values for the constructive parameters, the positional analysis made by way of the
Hartenberg-Denavit method is presented next. The general case of the analysis of the RCCC
mechanism was completed by Yang who applies as manner of study the dual quaternions and by
Fischer who studies the kinematics of the mechanisms using matrices with dual elements. The
generalized Cardanic mechanism presented in Figure 2 consists in four elements: the driving element
1 , the intermediary coupling element 2 that has perpendicular (without being a strict requirement) but
not intersecting arms, the final element 3 and the ground, 4 . The elements 1 and 3 are identical as
shape but they distinguish by the joint they have with the ground. Thus, while the element 1 and the
ground 4 complete a rotation pair, R , the driven element 3 and the ground 1 make a cylindrical pair,
C . The intermediate element 2 creates cylindrical pairs with the driven element and with the driving
element respectively, and for this reasons the mechanism’s name - RCCC , is, from structural point of
view, justified. For the kinematical analysis of the mechanisms, the Hartenberg-Denavit methodology
is applied. To this purpose, the Oz axes of the pairs are chosen with the orientation presented in
Figure 2. Once the Oz axes are stipulated, the Oxk ,k  1 axes are chosen as to superpose the current
axis Oz k aver the next Oz k  1 axis by a screw motion with the  k ,k  1 rotation in trigonometric sense,
as shown in Figure 2. The motions of the mechanism will be described by the displacements from the
three cylindrical pairs (three rotations  2 ,  3 and  4 together to the translations s 2 s 3 and s 4 )
expressed as functions of the position of the driving element  1 . The constructive characteristics of the
mechanism will be represented by:

4
International Conference on Applied Sciences IOP Publishing
IOP Conf. Series: Materials Science and Engineering 477 (2019) 012037 doi:10.1088/1757-899X/477/1/012037

 12   23   34   / 2 (13)
and
a12  a34  0 . (14)

x34 z3
z2

z4
(3)

x41

(2)

x23 (1)

z1

x12

Figure 3. Generalized Cardanic mechanism

Z (  1 , s1 ) X (  12 ,a12 )Z (  2 , s 2 ) X (  23 , a 23 )Z (  3 , s 3 )
(15)
 X (  34 , a 34 )Z (  4 , s 4 ) X (  41 , a 41 )  I 4

where Z(  , s ) and X (  ,a ) are the displacements with respect to z axis and x axis respectively,
applying the equations (7) and (8) and I 4 is the unit matrix of fourth order. It is obvious that by
replacing the equations (1) and (2), the concrete form of the matrix equation(3) will be considerable
simplified. Though, to solve this equation is not an easy task. McCarthy recommends that the equation
(15) should be written under the form:
Z ( 1 , s1 ) X ( 12,a12 ..... (.......Z (  4 , s4 ) X (  41 ,a41 ))1 (16)

without stipulating the point where the two terms from the initial equation must be separated. In the
present case, the McCarthy method was applied by writing the equation (16) in the following manner:
Z (  1 , s1 ) X (  12,a12 )Z (  2 , s2 ) X (  23 , a23 ) 
(17)
[ Z (  3 , s3 ) X (  34 , a34 )Z (  4 , s4 ) X (  41 , a41 )]  1

5
International Conference on Applied Sciences IOP Publishing
IOP Conf. Series: Materials Science and Engineering 477 (2019) 012037 doi:10.1088/1757-899X/477/1/012037

where O4 is the zero matrix of 4 x4 type. The matrix equation (15) assumes the annulations of the
rotation matrix and of the displacement vector from the matrix on the left member of equation (15).
The displacement vector doesn’t create problems as it means the simultaneous cancellation of the first
three elements from the last column of the left member matrix, but concerning the rotation matrix, it is
characterized by 9 elements (the 3x3 type matrix from upper left).
C1C 2  C 3C 4 S1C 2  C 3 S 4 C 41  S 3 S 41 S 2  C 3 S 4 S 41  S 3Ca41 a23  s1S 2  a41C 3C 4  s4 S 3 
 
 S  S C  C1  S 3 S 4 C 41  C 3 S 41  S 3 S 4 S 41  C 3C 41  s2  a41S 3C 4  s4 C 3 
 1 3 4 O
  4
 C S
1 2  S 4 S  S
1 2  C C
4 41  C  2  C  S
4 41 s C
1 2  a 41S   s 3 
 
 0 0 0 0 
(18)
From the nine equations obtained, only three are independent. A rigorous solving assumes
separation of the symmetrical parts from the two matrices:
R  RT (19)
Rs 
2
which permits finding the angle of rotation and the disjointing of the asymmetrical parts:
R  RT (20)
Ra 
2
which permits finding the versor of the axis of rotation.
For the current case, a different technique will be used consisting in the convenient selections of
the three equations from which the angles  2 ,  3 and 4 should be found. The elements from the
positions (3,1) and (3,2) are equaled to zero and the following system results:
C1S 2  S 4  0 (21)

S1S 2  C 4 C 41  0

that allows for the 4 to be found.

4  atan[C41C1 / S1 ] (22)
The elements from the positions (3,2) and (3,3) are equaled to zero:
S 1 S 2  C 4 C 41  0 (23)
 C 2  C 4 S 41  0

and it is found:
2  atan[tan4 /( S41C1 )] (24)
The elements from the positions (1,1) and (2,1) are zero:
C 1C 2  C 3C 4  0 (25)
S 1  S 3C 4  0

and it results:
3  atan* (tan1 / C 2 ) (26)
The displacements from the cylindrical pairs are obtained straightforward:

6
International Conference on Applied Sciences IOP Publishing
IOP Conf. Series: Materials Science and Engineering 477 (2019) 012037 doi:10.1088/1757-899X/477/1/012037

s3  s1C 2  a41S 4
a C C  ( a23  s1S 2 ) (27)
s4  41 3 4
S 3
s2  s4 C 3  a41S 3C 4

Next it will be shown that the employment of the function atan( x ) in the expressions of the
rotations (22), (24) and (26) may generate errors in the description of the kinematics of the
mechanism. Instead of function atan(y/x) , the use of the function angle( x, y ) , existing in any
calculus utilitarian, is proposed; this function returns the value of the angle from the domain [ 0,2 )
made by the vector radius of the point of coordinates ( x, y ) with the positive half-axis Ox . The
expressions of the rotation angles from the cylindrical pairs, as dependencies only of the angle of the
driving element  1 , are presented next:
 2  angle( S 41S1 ,C 41 )

 3  angle( S 41C1 , S 21  C 2 41C 21 (28)

 4  angle( S1 ,C 41C1 )

4. Discussions
The variation of the linear displacement from the pair between the driven element and the ground is
presented in Figure 4, with continuous line, found by means of relations (28), with broken line, found
by using the relations (22), (24) and (26). One can observe that, the two plots coincide only for the
first quarter of rotation of the driving element and differ fro the rest. At a first sight it should seem that
it is only a vertical translation, on portions, of the graph with continuous line.

Figure 4. The translation of the driven element calculated in two ways

If the graphs from Figure 4 are parts of the same continuous function, translated vertically on
intervals, their derivatives must coincide. In Figure 5 there are presented the derivatives of the two
functions from Figure 4. It can be observed that the plots are identical only for the portion on which
the graphs from Figure 4 coincide. This fact confirms the statement that the use of the function
ata n( x ) generates errors in the kinematical analysis of the mechanism. To clarify this aspect, one
must remark that the expressions of linear displacements from the cylindrical pairs depend on the
rotation angles  2 ,  3 and 4 as arguments of the trigonometric functions cos( x ) and sin( x ) .

7
International Conference on Applied Sciences IOP Publishing
IOP Conf. Series: Materials Science and Engineering 477 (2019) 012037 doi:10.1088/1757-899X/477/1/012037

Figure 5. The derivatives of the functions from Figure 4

The following simple functions are considered next in order to explain the effect of applying the
atan(x) function to the above mentioned ones:

f (  )  sin(angle(cos(  ),sin(  ))
f '(  )  sin(atan(sin(  ) /cos(  ))) (30)

Theoretically, these expressions should represent the same function or at most one of them could be
vertically translated on portions but in the end their derivatives should be equal.

Figure 6. The effect of the functions atan( sin(  )) and angle(sin(  ),cos(  ))

Figure 7. The effect of the functions atan( sin(  )) and angle(sin(  ),cos(  ))
upon the derivative of the same function cos( x )
A similar situation arises for the cos( x ) function. It can be concluded that, applying the
trigonometric functions sine and cosine to the functions a tan( y / x ) and angle( x , y ) different results
are obtained. Finally, using the relations (28), the rotation velocities and displacement velocities from
cylindrical pairs were found and there are shown in Figure 8 and Figure 9 respectively. Both the
rotation velocity and the translation velocity from the pair between the final element and the ground

8
International Conference on Applied Sciences IOP Publishing
IOP Conf. Series: Materials Science and Engineering 477 (2019) 012037 doi:10.1088/1757-899X/477/1/012037

are the result of the superposition of a periodic component over a constant component, the frequency
of the periodic component being twice the frequency of the rotation of the driving element. The law of
angular velocity variation is identical to the one of a Cardanic mechanism [2].

Figure 8. The rotation velocities from cylindrical pairs

Figure 9. The linear velocities from cylindrical pairs

Concerning the motions from the intermediate pairs, from Figure 8 it possibly will say that the
rotations from these pairs are the same only delayed with a quarter of period and of opposite signs.
However the plots of the translation velocities disagree with this hypothesis. For a true such
hypothesis, the plot s3  s3 ( s2 ) should be a symmetrical shape with respect to coordinate axes. The
graph from Figure 10 confirms that the motions from the two pairs present different characteristics.

Figure 10. The trajectory of the center of the coupling element

9
International Conference on Applied Sciences IOP Publishing
IOP Conf. Series: Materials Science and Engineering 477 (2019) 012037 doi:10.1088/1757-899X/477/1/012037

5. Conclusions
The paper presents the kinematical analysis of the spatial mechanism obtained from a Cardanic joint,
which is a spherical mechanism, by transforming the rotation pairs into cylindrical pairs. The new
mechanism presents the advantage that transmits the rotation motion identically to the spherical
Cardanic mechanism. But, compared to it, the condition of intersection of the axes of all pairs, which
is difficult to ensure in applications, must not be strictly obeyed.
After a short review of the methods applied in spatial analysis, the method of homogenous
operators proposed by Hartenberg and Denavit is chosen as technique for the analysis of the
mechanism. The analytical expressions for all the displacements from the pairs of the mechanism as
functions of the rotation angle of the driving element are resulting from the analysis.
In the final part of the work, the errors produced by the use of multiform function a tan( x ) are
highlighted and necessity of replacing this function with the uniform function angle( x , y ) is discussed.
The variations of the rotation and translation velocities from all the pairs of the mechanism are
presented and the remark that in the final pair the motion is helicoidally periodic with the period half
of the period of the driving element is made.

Acknowledgement. This work was supported by a grant of the Romanian Ministry of Research and
Innovation, CCCDI – UEFISCDI, project number PN-III-P1-1.2-PCCDI-2017-0404 / 31PCCDI ⁄
2018, within PNCDI III.

References
[1] Erdman A G, Sandor G N and Kota S 2001 Mechanism Design: Analysis and Synthesis, 4th ed,
Pearson
[2] Pennock G R, ShigleyJ E and Uicker J J 2014 Theory Of Machine And Mechanisms, 4th ed,
Oxford University Press
[3] David B. Dooner 2012 Kinematic geometry of gearing, Willey
[4] Litvin F L and Alfonso Fuentes A 2004 Gear Geometry and Applied Theory, 2nd ed,
Cambridge University Press
[5] Seherr-Thoss H C, Schmelz F and Aucktor E 2010 Universal Joints and Driveshafts: Analysis,
Design, Applications, Springer
[6] Dudita F 1974 Cuplaje Mobile Homocientice, Ed.Tehnica Bucuresti
[7] Dudita, F 1966 Cuplaje Cardanice, Ed.Tehnica Bucuresti
[8] Lung-Wen Tsai 2000 Mechanism Design: Enumeration of Kinematic Structures According to
Function, 1st ed, CRC Press
[9] Norton R L 2008 Design of Machinery: An Introduction to the Synthesis and Analysis of
Mechanisms and Machines, 4th ed, McGraw-Hill Higher Education
[10] Huckert J 1958 Analytical Kinematics of Plane Motion Mechanisms, The Macmillan Company
[11] Chiang C H 2000 Kinematics of Spherical Mechanisms, 2nd ed, Krieger Pub Co
[12] Angeles J 1989 Rational Kinematics, Springer Tracts in Natural Philosophy, Springer
[13] Brandt L 1947 Vector and tensor analysis, 1st ed, J. Wiley
[14] McCarthy J M 1990 Introduction to Theoretical Kinematics, 1st ed, The MIT Press
[15] Hartenberg R S and Denavit J 1964 Kinematic Synthesis of Linkages, 1st ed, McGraw-Hill Inc
[16] Dimentberg F M 1965 The Screw Calculus and Its Applications in Mechanics, Foreign
Technology Division Translation, FTDHT
[17] Phillips J 2007 Freedom in machinery, Cambridge University Press
[18] Yang A T and Freudenstein F 1964 Application of dual-number quaternion algebra to the
analysis of spatial mechanisms ASME Journal of Applied Mechanics 86 pp 300–308

10