CHAPTER 5

AN INTRODUCTION TO THE
SYNTHESIS OF FOUR BAR MECHANISMS
5.1 Introduction
This chapter is concerned with the synthesis or design of four bar mechanisms.
That is, given a desired movement design the four bar mechanism that provides that
movement. We start this kinematic synthesis with a neecl to provide some motion, not
a mechanism. Our goal is to find the mechanism that meets our needs. This is very
different from the analysis tasks discussed in the previous three chapters where we
were given the mechanism, and asked to find positions, velocities and accelerations of
parts of the mechanism.
There are several stages in the general synthesis or design process. We are starting
with a need to provide a motion. However, considerable effort may have been made
by the designer prior to this stage to define the overall system in which a four bar
mechanism may be only one part. Given a need to provide a particular motion, it is
important to consider the means that mav be used to provide that motion. Clearly, a
four bar mechanism is not the only means at our disposal for providing a specified
motion. Recalling the back hoe shovel discussed in Chapter 1., the four bar function
generator served to rotate the shovel. Other means could have been used for that task
including elechic or hydraulic motors that produce a rotational output directly. Using
such clevices would eliminate the need for a four bar mechanism. The designer had to
choose between these and manv other options, but decided after consideration of
multiple factors that a four mechanism driven by a hydraulic cylinder was the
preferable means for providing the motion of the shovel. Some of the factors which
entered into this decision making process may have been cost, maintenance, reliabilitv
and the available power sources. Choosing between alternatives is an important parf
of the engineering design process. In kinematic synthesis this particular decision
making process is sometimes referred to as type synthesis (ref). While this has been a
somewhat qualitative process, some relatively recent efforts have been made to make
this process more quantitative. For our purposes, we will accept that the clesirecl
means of providing a motion is through a four bar mechanism. Thus, we will
concentrate on finding the four bar mechanism that suits our needs. This process is
sometimes referred to as climensional synthesis (re0.
5.2 Classification of Synthesis Tasks
In Chapter 1 u.e saw that mechanisms may be designed for different purposes. The
mechanism that drove the back hoe shovel was called a function generator. In this
case the mechanism is designecl to provide a coordinated motion between the input
and output links. This motion is coordinated by a mathematical function of one
variable. The mechanism shown in Figure 5.1 is an example of another use of a
function generator. This mechanism is used to control the air-fuel mixfure in a
furnace. In this case it was important to maintain a linear relationship between the air
and fuel content such that when the fuel supply decreased toward zero the air supply
also decreased toward zero. Thus, this mechanism expresses a linear functional
relationship with a zero
"y
intercept" between the rotation of the shafts attached to the
input and output links. However, it was also necessarv to provide a family of straight
lines through the origin of the fuel air relationship to account for different operating
conditions. To provide this linear relationship, over a range of conditions, this
particular mechanism was made adjustable: pneumatic cylinders in the coupler and
output links could be used to change the link lengths to maintain the desired
functional relationship. Another example of a function generator is a door closer
(Figure 5.2). Here the rotation of the door is to be controlled, and the door is a link
which is attached to the frame. In each of these examples the mechanism is designed
to meet is some specified relationship between the rotation of the input and output
links.
The front loader discussed in Chapter 1 is an example of a motion generating, or
rigid boclv guidance mechanism. This mechanism was designed to produce a
specified rigid bodv motion of the coupler link. In some automobiles the windshield
wiper is attached to the coupler link of a four bar rigid body guidance mechanism
(Figure 5.3). Such a mechanism allows the wiper to clear a larger area of the
windshield than would be possible if the blacle was rigidly attached to a single
rotating arm. In each of these motion synthesis mechanisms the angular rotation of
the coupler as well as the displacement of a point on the coupler was specified bv the
designer. Specif,ving these three variables fullv locates a rigid body in a plane, a,ncl
hence the name rigid body guidance.
If only the displacement of a point on the coupler is specified in a desigry then the
design task is known as path svnthesis. Some common paths for which mechanisms
have been designed are straight line and circular segments, but the clesign process is
not limited to these paths. Some well known straight line path generating
mechanisms are shown in Figure 5.4. It is important to realize that manv other types
of mechanical svstems can produce shaight line motion such as the.utu-follo*"i t.
rack and pinion shown in Figure 5.5. One feafure that distinguishes four bar
mechanisms is that the driving stroke (the specified path of a point on the coupler)
and refurn stroke are nof in general, the same. This is a useful feafure in systems
such as the film advance mechanism shown if Figure 5.6. The point P on the coupler
develops straight line motion for the driving stroke fram ,\ to ,\ butit then clears the
film for the return skoke,
4
to
4
to
4.
5.3 Rotation of a Vector - The Rigid Body Rotation Matrix
The analytical methods of kinematic svnthesis, to be developed in the next few
sections, will require that we have some means for expressing the vector that results
from the rotation of a given vector about a specified axis. For our purposes, the
vector to be rotated represents a link in a mechanism, although the methods we will
use may be applied to any vector. Since we are sfudving planar mechanisms, we will
sfucly the rotation of a vector in a plane. In Chapter 9 these ideas will we generalized
to three dimensional motion.
As shown in Figure 5.7 a vector r in the xy plane has components
r,=rCoSQ (5.1)
and
r,
=
rsitlS (5.2)
where r is the magnifude of the vector r. Note that the tail of this vector passes
through the origin of the xv coordinate system. Rotating this vector about the z axis
by an angle @ produces a new vector r'. The components of r' are given bv
r'= r'cos(0 + 0)i + r'sin(g + 0)i (5.3)
where r' is the magnifude of the vector r'. The magnitudes of the vectors r and r'are,
however, equal since t' was produced by a rotation of r. Using the fact that
lr I
=
lr'l
and the trigonometric identities for the sine and cosine of sums of angles, Equation
(5.3) may be rewritten as
1'=
dcosg
cos0
-
sing sinO)i +
r(sing cose + sino cosg)
'
(5'4)
Recalling the expressions for r, and r, from Equations (5.1) ancl (5.2) we can write the
above expression for the vector r' in terms of the vector r as
r'= (4 cose
-
I
sinO)i +
(4sine+!cose)j
(s.s)
Recognizing that the components of r'are linearly related to the components of r, this
expression can also be written in matrix form as
or
[r']= [n][r]
$.n
The matrix
[R]
is known as the planar rigid body rotation matrix. It is an operator
which transforms the vector r into the vector r'. It is important to recognize that this
expression was derived for the rotaUon of a vector passing through the origin of a
coordinate system, and that the axis of rotation also passed through the origin of the
coordinate system.
The matrix
[R]
has some interesting properties. The inverse of this matrix is equal
to its transpose
[n]'= [n]-'
(s.8)
While this can be proved rigorously, it can also be seen by simply multiplying
[R]
by
its transpose and observing that the result is the identity matrix. It can also be shown
that the determinant of
[R]
is always
+1. This is known as a proper transformation
since it represents an acceptable rigid body motion (ref). In a homework problem it
will be demonstrated that a determinant of -1 corresponds to kinematically
impossible, or improper motion of a rigid body: furning it inside out
EXAMPLE 5.1: Find the vector that results from a rotation of the vector r = 10i by'135,
180, and 270 degrees about the z axis. Before proceeding to the solution given below,
which makes use of the rotation matrix, you should first solve this problem
geometricallv.
['',.l_ fcos0
-sinel[rl
L'',1
-
|
sine cose
l[r,J
For a 135 degree rotation
For a 180 clegree rotation
r' r
=
[T;l,';;i,:li;;i][';]
=
ll
I)
,-r,',
-
fcos(180)
L r
I
sin(180)
h'r
=
l:ftTi
":l!#i][';]
=
[
?,]
-sin(raoll[tol
=
[-1ol
cos(rsoilloj
L
o
I
For a 270 degree rotation
5.4 Kinematic Inversion
The concept of kinematic inversion is important for the understanding of graphical
techniques of kinematic synthesis as well as many other aspects of kinematics. Given
a mechanism, its inversion is created by allowing the ground link to be mobile, and
letting any other link be attached to the frame. This is illustrated in Figures 5.8a and
5.9a, where the mechanism in Figure 5.8a is inverted in Figure 5.9bby letting the
coupler be the ground link. A sequence of positions for the initial and inverted
mechanism are shown in Figures 5.8a-c and 5.9a-c. These positions were chosen such
that the angle between links
Qr{
and
QQ
were the same in both mechanisms. It is
important to realize that in these two mechanisms aII relahve motions are the same.
Thus, @,
= 0, ,
@,
= $r,
@,
= 0,
and Oo
= $0.
The equivalence of relative motions is not
limited to angular positions. Distances between specified points are also the same in
both mechanism. For example, AO,
=
A'Q, BO,
=
B'Q and PQ
=
PO, etc. The
equivalence of these and other distances in both mechanisms can be verified by
making measurements on the figures.
5.5 General Characteristics of Mechanism Synthesis
A review of the literafure would show that a large number of methods have been
developed for the synthesis of four bar mechanisms. These include both graphical
and analytical methods. Graphical methods are of interest since they illustrate some
of the basic concepts of mechanism synthesis. Graphical methods for svnthesizing
motion, path anci function generator mechanisms will be illustrated beiow. Whil;
graphical methods have been generally replaced by analytical methods,
programmable computer aided drafting svstems have made the graphical approach a
viable alternative to analvtical methods. Using CAD programmed graphical methods
it is sometimes possible to see how a mechanism may be modified to improve the
motion that it produces (ref). In addition to the graphical solutiory one, generally
applicable analytical approach will also be developed for motion, path and function
synthesis. This basic approach is based on a vectorization of methods often
implemented using complex numbers to represent planar vectors (refs).
We will design mechanisms to provide a specified movement. However, the
contirut.ous desired movement is not specified in the design. Rather, selected positions
are specified, and the mechanism is designed to conform to these positions. In
between these positions, the motion produced by the mechanism may not fit the
specified motion.
5.5 Motion Synthesis
We will begin the sfudy of specific methods of four bar mechanism synthesis by
considering motion synthesis or rigid bodv guidance. In this case we want to proiuce
a specifiecl rigid body motion of an object which is attached to the coupler of the four
bar mechanism.
Three positions of a desired rigid body motion are shown in Figure 5.10. Clearlv,
the desired motion is continuous, and includes positions between those shown, but
the synthesis will be based on just these three positions known as accuracv points or
precision points. While this may appear to be an arbitrary choice, we will see that the
number of accuracy points has a profound influence on the solution procedure used to
size the links in our design. We will develop both graphical and analytical methods
for synthesizing four bar mechanisms.
5.6a Motion Synthesis - Graphical Method
In all of the graphical solutions, subscripts 1,2 and 3 will be used to denote the
location of a point in the firsf second and third accuracy point. If a subscript is not
assigned to a point then it refers to the first accuracy position, that is, A has the same
meaning as
4.
We will start the graphical construction by assigning locations of the moving pivots
A and B on the coupler link as shown in Figure 5.11. This choice of locations for A
and B was arbitrarv. The ground pivots,
Q
and
Q
are the centers of the circles haced
by points A and B respectively. To find these centers we construct the perpendicular
bisectors of the lines ArA, and ArAr. The ground pivot
Q
is located at the intersection
of these two bisectors: a point equidistant from
4, 4
and
4.
Similarly, the ground
pivot
Q
is located at the intersection of the perpendicular bisectors of B,B, and BrB..
We have now svnthesized a four bar mechanism, qABq, that will transport its
coupler through the three specified positions. This graphical synthesis procedure is
now summarized in the following steps:
1. Pick three positions in the rigid body motion that r,r'ill be used as accuracy
positions.
2. Assign locations to the moving pivots A and B as in Figure 5.10.
3. Construct the perpendicular bisector of ArA, and At,\. The intersection of these
bisectors determines the location of the fixed pivot
Q.
4. Construct the perpendicular bisector of BrB, and BrBr. The intersection of these
bisectors determines the location of the fixed pivot
Q.
5. Compare the mechanism characteristics with design specifications such as the
overall rigid body motion and the transmission angle. Redesign the mechanism as
needed bv modifiiing the free choices.
However, there is no guarantee that this mechanism will suit all of the designers
specifications. The mechanism only reproduces the specified accuracy positions. The
motion of the coupler link, between the specified positions, may not be acceptable. It
is also possible that the ground pivots may fall in unacceptable locations on the frame,
the input and output links may sweep over an unacceptable region of the frame, the
transmission angle mav not be acceptable at some mechanism positions or that the
overall mechanism mobility may not be compatible with the designers need. For
example, the designer may need a crank and rocker mechanism, but the svnthesis
procedure may have vielded a drag link mechanism. While there mav be many
potential problems with the mechanjsm we have svnthesized, there are an infinite
number of mechanism that can be found to produce the specified motion when three
positions are given.
We can begin to identifu other mechanisms by recalling that the locations of the
moving pivots, A and B, rvere arbitrary. Placing one or both of these moving pivots at
a different location on the coupler will produce a different mechanism. It is important
to recognize that the moving pivots may be placed anywhere in space as long as they
are atbached to the part whose motion we are tr"ving to produce. Another mechanism
synthesized to produce the motion shown in Figure 5.11 is shown in Figure 5.12. Note
that in Figure 5.12 extensions were added to the moving link to provide locations for
the moving pivots, but the motion of the moving link was not changed. If this second
mechanism is still not acceptable we can relocate the moving pivots and svnthesize yet
another mechanism. In searching for different mechanisms, the designer is making
use of
"free
choices", that is the freedom to position the moving pivots to produce an
acceptable mechanism. The fact that we have these free choices, or free variables, at
our disposal in the synthesis process, implies that there are an infinite number of
mechanisms that can be designed to produce a prescribed motion, at least when the
design is based on three accuracv points.
5.6b Motion Synthesis - Analytical Methods
Motion synthesis is concerned with the design of a mechanism to produce a
specifiecl rigicl body motion of the coupler link. As such, the clisplacement of a point
on the coupler, and the rotation of the coupler are specified. The analytical approach
to motion synthesis is based on a vector loop equation around the mechanism in
groups of two of its accuracy points as shown in Figure 5.13. Writing the vector loop
equation from
Q
to
Q
over mechanism in position one and back to
Q
over the
mechanism in the k-th position
vielcls:
p+q- s-r+rr +sr
-Qr -P,
= 0
From Figure 5.13 we can see that
P+q-9r -Pr -
-dr
and
-f-S+Sr+f*=d*
(s.e)
where d* is the displacement of a point on the coupler.
Comparing Equations (5.9), (5.10) and (5.11) we see that the loop equation may be
satisfied by satisfying Equations (5.10) and (5.11) individually. In terms of the
synthesis, this means that we can find the vectors p and q independent of the vectors
r and s. This mav not be too surprising given our experience with the graphical
approach where we determined the input and ouQut links based only on their
respective moving pivot locations. That is, the position of the moving pivot A,
determined
Q,but
had no effect on the position of q.
(s.10)
(s.11)
We will develop the solution of the synthesis equations for the vectors p and q
from Equation (5.10). The solution for r and s from Equation (5.11) will be exactly the
same as that for p and q. The vectors p* and q* in Equation (5.10) are related to p and
q by a simple rotation
lp*l= lP-l[p]
[q*]= iq-liql
(5.12)
and
where
[4]
and
iQ*]
u.. the rotation matrices defined in Equation (5.6). Substituting
Equations (5.12) and (5.13) into Equation (5.10) leads to
lP-
-
tl[p]+
[e*
-
r]lql
=
lo-l
(5.13)
(s.14)
where
[t]
is tne identity mahix. This is an equation in the unknown link vecto.s
[p]
and
[q]
and the rotation matrices
[4]
anct
[q*].
fnis is the fundamental equation of
kinematic synthesis. We will see that the equations for path and function synthesis
are of the same form.
Before beginning to develop the solution procedure, it must be recognized that
Equation (5.14) relates the
frst
and k-th accuracy points of the mechanism. If we are
synthesizing a mechanism by specifying three positions then we have two equations
of the form of (5.14), one which relates the
frst
and second positions, and one which
relates the
first
and third positions. We will concentrate on the solution for the three
position case. Our goal, now, is to develop an approach for solving this equation.
Using Equation (5.6) we can explicitly write the rotation matrices
ffr]
ana
ffr]
lRl
=
fcosO,
-sinO,-l
L
zr
I
sin@, cosO,
]
lcoso^
-sino^l
LP,l
=
[rino,
coso;J
where the angles @, and @, are the rotations of the input link, p, from the first
acflrracy point to the second, and from the first accuracy point to the third. These
angles are unknown. Similarly, the matrices
[Qr]
u"a
[Q,]
ur"
where the angles
S,
and
0,
are the coupler rotations when moving from the first to the
second mechanism positions and from the first to the third positions. These angles are
known: they are specified by the designer since this is motion synthesis.
We can now summarize the equations and the known and unknown quantities in
these equations. For three position synthesis we have two planm vector equations of
the form of (5.1a). One for the displacement of the mechanism from accuracy point
one to accuracy point two
lP,
-
tl[p]+
[e,
-
t]lql
=
lo,l
(s.17a)
and one for the displacement of the mechanism from accuracy point one to accuracy
point three
[P,
-
r][p] +
fe,
-
tl[q]
=
lo,l
(5.18b)
These two vector equations represent four scalar equations. The known quantities are
re,r=[:TJ;
":iJ;]
re,r=
[::;f;
":i,l,]
l0
the displacement vectors d, and d, of a specified point on the coupler and the
rotations
$,
and
$,
of the coupler. The unknown quantities are the vectors p and q
and the rotations O, and O, of the input link. These are six scalar unknowns which
must satisfy the four scalar equations resulting from (5.17a) and (5.17b). This
seemingly awkward sifuation, more unknowns than equations, has a simple
resolution: we arbitrarily assign values to two scalar unknowns. Again, this sifuation
should not be surprising based on our experience with the graphical solution to
motion synthesis. We saw that the designer was free to position each moving pivot on
the coupler, which resulted in an infinite number of possible mechanisms. In the
analytical solution we also see that there is an infinite number of solutions: this is the
implication of having more unknowns than equations. However, as we will now see,
the nafure of the analytical solution will depend on which variables we assign.
If we assign anv two scalar components of the vectors p and
e,
for example pr, and
4r,,
then the variables p*
Q,r,
@, and O, will be unknown. Equations (5.17a) and
(5.17b) can then be solved for these variables. However, these equations are nonlinear
in the angles @, and @r, and a nonlinear, numerical, solution algorithm such as the
Newton-Raphson method would have to be used. Alternatively, we could assign
values to O, and @r, in which case the four scalar components of the vectors
fp]
and
fl]
wouta be unknown. In this case, Equations (5.77a) and (5.17b) would be linear in
the unknown quantities and are given in matrix form by
lQ, -
rrlItpr.l=
[td,]l
tq, -
rllltqll
LIo,tl
Ilq
Lre
-rl
-
rI
(5.18a)
or in
"shorthand"
form
lslln]= lpl
(5.18b)
where each of the partitions in the coefficient matrix represents a 2 x2 arcay ancl the
partitions in
[n]
and
fn]
are2x1 arrays. Using Equations (5.15) and (5.16) the
coefficient mahix in Equation (5.18a) can be written in a more explicit form as
l1
fcos@r-1 -sin@,
cos$r-l
-sin$,
I
l*r:l
sinO, cosO,-1 sin$, cos$r-tl
L"r
lcos@r-l -sinO,
cos$.-1
-sin$,
I
I
sinO, cos@,
-
I sin$, cos$,
-
1-l
(5.18c)
The solution to Equations (5.18a) can then be found by inverting the coefficient
matrix
or by any other means for solving simultaneous linear equaUons.
The details of the synthesis are given in the example that follows.
EXAMPLE 5.2: Using the analytical synthesis method developed above, design a
mechanism to produce the motion shown in Figure 5.10. For this example we will
pick our free choices from the graphical solution. The mechanism we will design will
then be quite similar to that shown in Figure 5.11.
Since this is a motion synthesis, the displacement of some point on the coupler,
and the coupler rotation must be specified. Using point p on the coupler (Figures 5.10
and 5.11), the displacement vectors d, and d, are
dr=-1.85I+4.60J
d,
= -9'751+3.941
and the coupler rotations are
4, =
70"
0t =
100'
For the dyad composed of the input link
QA
and the coupler segment AP, the fuee
choices are the input link rotations @, and @r. These are chosen to be @,
=
56.37" and
@r
=
154.7". The coefficient matrix, Equation (5.18c) can now be determined, and is
given by
lffi]=[fi-;]
l3:-lll
[[,]l
(s1e,
12
l-o.uo -o.833 -o.658 +.q+ol
tst
=
I
0.833 4.446 o.s4o ,.urr
I
L r
I
-1.904
4.427
-1.174 -0.985
I
lo.or,
-r.eo4
0.e85
-t.t
o)
I
tr+r 0.208
-0.785 -o.ttzf
lsl_,
=
|
-0.208
t.t4t 0.372
-0.785
I
L r
l -1
413 0.433 0.606 0.38r
I
l-0.+lo -t.4ti -0.38r
o 606
j
l-o.ztz
-0.685
-o.658 -o.e+ol
fst
=
I
0.685 4.272 o.e4o +.osa
I
L r
|
-1.534
{.8,16
-1.174 -0.985
|
I
o.t+o
-1.s34
o.eB5
-t.n+ l
and the inverse of
[S]
is
Using Equation (5.19), p and q ane:
p
=
5.041- 1.085J
q
=
0.200I +0.400J
Which are similar to the graphical solution shown in Figure 5.10. For the dyad
composed of the output lir:dr' qB and coupler segment 8P the displacement vectors d,
and d, and the coupler rotations
$,
and
S,
are unchanged. Picking the free choices as
ltz =
43.24' and
Vt =122.25"
and using these angles in Equation (5.18c) leads to
and its inverse
l3
o 062l
1.0s3 I
o 118
I
0667l
I t.ttt
rrl=
|
:,^::,,
[-o
rn,
-0.444
1.337
0.793
-1.401
-1.053
4.062
0.667
-0.118
From Equation (5.19) the vectors describing the right hand dyad of the mechanism are
found
r=5.99I+1.78J
s
=
0.199I
-1.2031
By now
you
may be wondering why we have concentrated on three position
synthesis? The answer lies in the nafure of the solution procedure for the synthesis
equations. Specifving more than three mechanism positions leads to an inherently
nonlinear set of equations. That is, we cannot assign values to some of the unknown
variables and produce a set of equations which are linear in the remaining unknowns,
if more than three positions are specified. This point is illustrated in one of the
homework problems.
5.7 Path Synthesis
In path synthesis, only the curve traced bv a point on the coupler if of interest,
there are no constraints on the angular position of the coupler as we had in motion
synthesis. We will illustrate the synthesis procedure using three accuracy points on
the desired continuous path. The graphical procedure depends on the number of
accuracy points. The
fornutlntiott
of the analytical approach is independent of the
number of accuracy points, however, the solrttioll of the synthesis equations will
depend on the number of accuracv points, and how free choices are chosen as in
motion synthesis.
5.7aPath Synthesis - Graphical Method
A desired path, and three positions on that path which will be used in a
mechanism synthesis are shown in Figure 5.14a. It is important to realize that the
positions Pr, P, and.( represent three locations of a point, P, fixed on the coupler.
The point P moves through the three positions
\,
P, and
{
as the mechanism is being
driven. Realizing that P is the location of a point fixed to the coupler will help our
understanding of the graphical synthesis of this mechanism. The graphical approach
will make use of the concept of kinematic inversion. It should be remembered that an
l4
inverted mechanism maintains the same relative motion as the original mechanism.
To synthesize a path generating mechanism we first pick, arbitrarily, the length
and orientation of the ground link, and the length of the input link. One selection of
these links is shown in Figure 5.1.4a along with the path we are trving to svnthesize.
The window in this figure also shows a generic four bar mechanism to be used as a
reference for identifving the labels of pivots and visualizing the relative motion of
links. Refurning to the synthesis, we arbitrarilv pick the length of one side of the
coupler from A to P. This allows us to determine three positions of the moving pivot
A,
4, \
and A, are each locatecl at a distance AP from
\,
P, and
{
respectivelv
(Figure 5.14b). There are two possible intersections that determine the locations of
these points. We picked the intersection so that they followed in sequence from ,4, to
4
to,4, as the input link rotates counterclockwise. If we coulcl locate the moving
pivot, B, then the mechanism woulcl be complete. To locate B we will invert the
mechanism at the first accuracv point, making AP the fixed link, and
QQ
"
moving
link. In this inverted mechanism B, which is fixed, is the center of the circle traced by
Q.
W. will now locate the positions of links
QA
and
QQ
in the inverted mechanism
at the second and third accuracy points. The points
Q
ancl O, at the second and third
accuracy points, will be denoted by (O,)r, (q)r, (Q), and (q)r. At the first accuracv
point,
Q
ancl
Q
will be named without the additional subscript.
In the true mechanism the angle y, between the input and coupler links decreased
when going from the first to the second accuracy point (Figure 5.14c). In the inverted
mechanism the same relative motion must be maintained. The position of
Q
in the
second position of the inverted mechanism, (Q)r, is located as shown in Figure 5.14d.
The point (Q
),
must lie on a circle with center at (Q), and with a radius of
QO.
as
shown in Figure 5.14d, and labeled circle #1. To locate (Q), ot circle #1 we use the
fact that the distance between P and
Q,
ut measured in the true mechanism, must be
the same in the inverted mechanism. Recognizing that at the second accuracy point P
is at Pr, we now draw the circle with center at
\
and radius Prq. Where this
intersects circle #-1, determines the location of (Q)r. But there are two such
intersections. Can we choose either of them? In general only one of these
intersections will yield (Q), in a position which is consistent with the inversion of the
mechanism. The intersection marked in Figure 5.14d is consistent with the inversion
as can be seen bv comparing the angle between links
QA
and
QQ
in the true
mechanism at the second accuracy point (Figure 5.14c) with that in the inverted
mechanism. We could also identif"v the correct intersection by comparing the
l5
diagonal
44
i" the true mechanism with the corresponding diagonal in the inverted
mechanism. These distances must be the same in both mechanisms. The positions of
Q
and
Q
at the third accuracv point are found in exactly the same manner as were
(Q
),
and (q)r. These are shown in Figure 5.14e. We now have three positions of
Q
in the inverted mechanism. These points q, (q), and (Q
),
lie on a circle with its
center at B. The intersection of the perpendicular bisectors to qQ), and (q)r(q),
locates point B (Figure 5.141). The mechanism
qAPBq is our path generator shown in
the first precision point in Figure 5.149. This synthesis procedure is summarized in
the steps listed below.
1. Pick three accuracy points on the continuous path.
2. Assign the length and orientation of the ground link (Figure 5.14a)
3. Assign the length of the input link (Figure 5.14a)
4. Assign a length to the segment of the coupler from A to P and then graphically
determine the location of the points
4, 4
and .\ (Figure 5.14b).
5. Draw the input lir:Jr- qA, and the coupler segment AP at the three accuracy
points (Figure 5.14c). Determine the angle y at each accuracy point.
6. Invert the mechanism, making the coupler the fixed link. Graphicallv locate the
position of
Q
in the second accuracy point (Q
),
(Figure 5.14d). Now, locate the
position of (Or), at the intersection of two circles, one with radius
QQ
and center at
(Q
),
and the other with radius Prq and center at
{
(Figure 5.14d). Remember, that
while there are two intersections of these circles, only one represents the inverted
mechanism.
6. Locate (Q
),
and (4
),
i. the third accuracy point in the same way that these
points were determined in the second accuracy point (Figure 5.14e)
7. Construct the perpendicular bisectors of q(q), and (Q
)r(q)r.
The intersection
of these bisectors determines the location of the point B.
8. We now have a path synthesis mechanism in the first accuracy point (Figure
sJ, g). The performance of this mechanism should be compared with the design
specifications, and redesigned as needed.
t6
As was observed with motion synthesis, it is possible that the first mechanism that
we synthesize will not meet all of our design criteria. These may include size of the
mechanism, position of the ground pivots, and transmission angle, among others. If
this mechanism is found to be unacceptable then any or all of the free choices may be
changed, and another mechanism synthesized. The free choices for this synthesis
include the length and orientation of the ground link, the length of the input link and
the length of the link AP.
5.7b Path Synthesis - Analytical Method
The analytical approach to path svnthesis follows that developed for motion
synthesis. For a desired path, displacement vectors d, anddr, between accuracy
points, are specified. These are defined in the same way as the coupler displacements
in motion synthesis. Vector loop Equations (5.10) and (5.11), developed for motion
synthesis, are then applicable to path synthesis. For three position path synthesis we
can then use Equation (5.17) to synthesize one dyad (the input link and one part of the
coupler) and an analogous set of equations to synthesize the other dyad (the output
link and the remaining half of the coupler). The equations for the input - coupler
dyad are
[e
-
t][p]+
le,
-
tl[q]
=
[0,]
[q
-
t]lpl *[e,
-
t][r]
=
[0,]
(5.20a)
(s.20b)
which are the same as (5.17a) and (5.17b). We again have two vector or four scalar
equations, but now there are eight scalar unknowns, pr,, pry,
e1,, 4r,
@r, @r,
0r, 0,
and assigning values to @,
,
@r,
6,
and
$,
results in a set of linear algebraic equations.
By assigning values to the input link rotations, @, and O, we are prescribing the
position of this link with respect to the position of the tracing point P on the coupler.
This is known as path synthesis with prescribed timing.
EXAMPLE 5.3 We will design a mechanism to synthesize the path shown in Figure
5.1.4a. We will determine our free choices from the graphical solution which should
result in a mechanism that is similar to that shown in Figure 5.14g. Since this is path
synthesis, only the displacement of a point on the coupler is specified in the design.
For the path shown in Figure 5.-1,4a, the displacement of point P is given by the vectors
l7
d, =
-0'
15I
-
0.38J
dl
= -1'391+0.241
For the dyad composed of the input lir*, qA, and the segment AP of the coupler, the
free choices are selected as @,
=
21.43' and @,
=
73.3T for the input link
Q,4
and
d, = -7.05'
and
S, =
5.81' for the coupler link.
The coefficient matrix
[S]
from Equation (5.18c) is
-0.365 -0.008
-0.069 -0.123
-0.958 -0.005
-0.713
0.101
Taking the inverse of
[S],
and using Equation (5.19), the vectors p and q found,
P =
0. 4891+0.8671
=
O,A
I=3.95I+1.88J=AP
These results are similar to those found graphically in Figure 5.1,4g.
For the dyad composed of the output lirk qB and the coupler segment BP, the
displacement vectors d, and d, are, of course, the same as given above. Likewise, the
chosen values for the coupler rotations
S,
and
$,
must be the same. The output link
rotations are chosen ars,
lrz =
-1A
and ry,
= -137.5".
Computing the matrix
[S],
and
using Equation (5.19) we find
r
= 0.4761- 0.2891
s
=
0.776I+3.591
5.8 Function Synthesis
Concepfually, function synthesis is quite different from either motion or path
[-0. ooq
tst=|
0365
L J
I
-0.713
Iorrt
o.tn 1
-0msI
-0
101 I
-o
oo5l
l8
synthesis. In function synthesis we are concerned with producing a coordinated
motion between the input and output links. We are not explicitly concerned with any
motion of the coupler. Despite these concepfual differences we will see that the form
of the equations used for function svnthesis closely follows that used for motion and
path synthesis.
The overall goal of function sl.nthesis is to produce a mechanism that generates the
function y
=
f{x)
as shown in Figure 5.15. When the input link
"points"
to a value of
the inclependent variable, r, the output link
"points"
to the corresponding value f(x).
Any mechanism that we design for this purpose will onlv produce the functional
relationship accurately at the precision points as shown in Figure 5.16. The difference
between the synthesized and desired function is known as the strucfural error.
Minimization of the strucfural error would appear to be a desirable feafure of any
function svnthesis technique.
We can think of the svnthesized function as a polynomial approximation of the
specified function. It can be shown that a Chebyshev polvnomial deviates the least
from a specified function. To obtain the best fit of a Chebyshev polvnomial to a set of
data we pick values of the independent variable, x, according to the formula
(s.12)
whereT is an index corresponding to the number of the precision point, Ax is the
range of r over which the function is to be synthesized and n is the total number of
precision points. Unlike motion or path synthesis, the location of precision points in a
function generator are not arbitrarily assigned by the designer. However, the
designer does specify the range of r for which the mechanism will be synthesized, as
well as the range of rotation of the input and output links that corresponds to the
range of variation of the independent and dependent variables. In fact, an important
relationship exists between these variables and the angular position of the input and
output links. The range of rotation of the input (output) link is scaled by the range of
the independent (clependent) variable, and this scaling is the same for intermediate
rotations of the input (output) link between precision points
xi=xo-+['-*'(1;)^]
t9
@,
-@,
=
ao
xi-x, Ax
(s.22)
Vi-Vr=AV
!1-\
AY
where the position of these links is referenced to the first precision point. That is,
@,
=
0 and ry,
=
0. Note that the absolute initial position of the input and output links
is not known or specified at this time, yet we are specifying angular deviations
@r, @j, y, and yrfrom this unknown position.
Finding precision points, and rotations of the input and output links is a precursor
to both the graphical and analytical synthesis of function generators. This initial set of
calculations will be illustrated in the following example.
EXAMPLE 5.4: Find the precision points and corresponding input - output link
rotation for the function
!
=
xt for the range of x from 0 to 2. The specified rotations
of the input and output links corresponding to this variable range are A@ = 60 deg,
and Ary = 90 cleg.
We find the location of the precision points from the Chebyshev spacing formula,
Equation (5.21)
t
{t -
"orf'4
-
t),,}
(sa)
xj=xo*tl'-"""\
2,
)''J
where xo
=
0, Lx
=
2 and n
=
3. Substifuting these values into (5a) we obtain
x,
=
0'134
rz
=
l'000
rr
=
1.866
(5b)
It is helpful to organize the data used in this calculation into a table
20
xi v;
0 0.0 0.0
1 0.134 0.018
2 1.000 1.000
a
J 1..866 3.482
4 2.000 4.000
We are now in a position to calculate the relative rotations of the input and output
links from Equation (5.22)
o,
=+
(x,-x,)
'
Ax'
o,
= 9(t.o-
0.134)
=
25.98
o.
=
+(1.866-0.134) =
51.96
Aw
V,=
^LOr-Yr)
'^y
vr, =ry(r.ooo
-
o.or 8) =
22.10
4'
90_
Vr
=
+
(3.482
-0.018)
=77.95
The calculated rotations of the input and output links are shown in Figure 5.17.
5.8a Function Synthesis - Graphical Method
The graphical approach will now be introduced using the results of Example 5.4.
First, we arbitrarily fix the length and orientation of the ground and output links as
shown in Figure 5.18a. If we could locate point A, consistent with the desired
function, we would have a complete mechanism. ff we invert this mechanism,
making the input link the fixed link, then the unknown point A is at the center of a
circle traced by point B. We can
perform this inversion since we have computed the relative rotations of the input and
output links Or, 03,
V,
and ry' In the true mechanism the angle between the input
and ground links increased by 25.98 deg. when moving from the first to the second
precision point. This same relative motion must be maintained in the inverted
mechanism, therefore
QQ
mustrotate clockwise by 25.98 deg. This locates the
second position of O, which we are calling (Q
), Gigure
5.18b). To locate B, we
2t
observe that in the true mechanism
QB
rotates counter clockwise. We therefore rotate
(Or)rB counterclockwise, by 22.10 deg, relative to (Q)rB'. This establishes the
position of B, (Figure 5.18b). The positions of (Q), and B, are determined in the same
manner as (q
),
and B, as shown in Figure 5.18b. We now have established three
positions of B in the inverted mechanism, B, B, and Br. These points move on a
circular path with its center at the unknown pivot A. The intersection of the
perpendicular bisectors to BB, and BrB, establishes the location of A as shown in
Figure 5.18c. The mechanism qABq in Figure 5.18d is our desired function generator
at the first precision point. The steps in this graphical synthesis are summ arized
below.
1. Determine the values of the independent variable at the precision points
using the Chebyshev spacing formula in Equation (5.21). Determine the scale factors,
and the relative rotation of the input and output links from Equations (5.22)
2. Assign the length and orientation of the ground and output links (Figure 5.18a)
3. Invert the mechanism about the (unknown) input link. Graphically locate the
positions (Q), and Brby rotating the ground link by an angle
-O,
and the output link
by and angle
-y,
using the reference lines shown in Figure 5.18b.
4. Repeat the procedures in step 3 to determine the locations of (q), and Br.
5. Construct the perpendicular bisectors of BB, and BrBr. The intersection of these
bisectors determines the location of the pivot A.
6. Evaluate the performance of the mechanism with regard to the design
specifications, and redesign by altering the free choices as needed.
As with our previous designs, we made a series of arbitrary choices of link lengths
and orientations when we designed this mechanism. We fixed the length and
orientation of the ground and output links. If this mechanism is unacceptable, these
free choices can be altered and a new mechanism designed. An interesting feafure of
function generators is that we can scale all link lengths by the same factor and still
produce the same function. Thus, if the mechanism we design is to small or to large,
it may be simply scaled up or down. The fact that we can multiply all link lengths by
a constanf without altering the functional relationship between input and output link
rotations, will be apparent in the next section on analvtical design of function
generators.
22
5.8b Funcfion Generation - Analytical Method
The analytical procedure for dimensional synthesis of function generator
mechanisms is based on a vector loop equation around the mechanism in two
positions as we did in motion and path synthesis. Flowever in function synthesis we
are not explicitly interested in the motion of any points on the coupler link. Therefore,
we do not need to include a vector to a tracing point P on the coupler. For a
mechanism in its first and k-th positiory the vector loop equation is given bv
p+t-r+rr
-t* -P* =0
(5.23)
where all vector quantities are defined in Figure 5.19. Using rotation matrices to write
r., t* and p* in terms of
9
t and p, Equation (5.23) may be written as
lp--lj[p]*lr--l][t]=
rr-t=ln--l]lr] (s.24)
where the right hand side was written in two forms to display two important physical
feafures. First, by examining the second term on the right hand side, we can see that
Equation (5.24) is linear in the link lengths. That is, if we were to multiply this
equation by a constant we would change the link lengths, but we would not change
the relationship expressed by this equation. Secondly, from Figure 5.19 it can be seen
that r,
-
r represents the displacement of point B when the mechanism moves between
the first and k-th precision point. Recognizing that the right hand side of Equation
(5.24) represents the displacement of point B it can be written as
lp-
-
rl[p]*
lr-
-
llltl
=
io-l
(s.2s)
where it must be remembered that d* is the displacement between the first and k-th
precision points. Equation (5.25) looks similar to (5.14) which we used for motion and
path synthesis. However, it is important to realize that Equation (5.25) is for the
whole mechanism as where in motion and path synthesis we had equations of this
form for each dyad.
For three position synthesis, we have two vector equations of the form of Equation
(s.2s)
le
-
tllpl *
lr,
-
rlltl
=
fo,l
lP,
-
tl[p]*[r,
-
t][t]=
fo,l
(5.26a)
(5.26b)
23
The solution procedure for Equations (5.26) is similar to that used for three position
motion and path synthesis. We will use the fact thaf in a function generator, the link
lengths can be scaled by a constant and assign the length of the output link. We will
also assign the orientation of this link in the first precision poinl With these choices
the vectors d, and d, are known since we have previously calculated the angles
ry, and
V3.
With these assignments we now have four scalar equations to find the six
scalar unknowns p1,, p1y,t1,, trr,
S,
and
Sr.
Using
Q,
and
0,
as our free choices leaves
us with four scalar equations to find four scalar unknowns, the components of the
vectors p and t.
24
EXAMPLE 5.5: The function synthesis mechanism, designed above by graphical
methods will now de designed using analytical methods. To define the vectors d,
and d, we must first assign the position of the output link, at the first precision point,
with respect to the ground link. For the example the initial position of the output link
was chosen to be he same as that determined from the graphical solution:
With this assignment and setting the length of
QB
=
2.82 in.,the vectors representing
the output link in its three precision points, r', r" and r"' are
r'
=
2.82fcos(68. 28)I + sin(68.28)Jl
r"
=
2.82lcos(90. 38)I + sin(90. 38)Jl
r "'
=
2.82fcos(146.22)I + sin(l 46.22)Il
The displacements,
d, and d., of point B are now found from:
d,
=
d"-d'=
-1.06I
+ 0.20J
d,
=
d"'-d"=
-3.38I -
1.05J
Since the relative rotations of the input link, @, and @, are knowry the matrices
ffr]and ffr]
are easily found. Choosing the coupler rotations to be
6, =
10 and
6, = -66,
the coefficient matrix in Equation (5.26b) is
[-0. ror
r,r=liffi
Iorrt
-0.438
-0.101
-0.788
-0.384
-0.060
4.342
-0.593
4.914
0s421
-0.060
I
0e14 I
-o
5e3J
25
-0
7e41
-o 831
|
-0.s261
-r
3e
_J
I
r+t
33t
ro.r l-tll
t.43
L'J-l
2.87
0.390
I
L_0.390
2.87
which yields
the vectors
-0.83
I
0.794
-1.39
0.526
and its inverse is
P =
2.791+ 1.98J
for the input link, ancl
t
=
2.29t +0.671
for the coupler link.
26
'56
'58
82
80
48
46
Fro^ U,
s.
? ale"^|"
4,za+,211
Ftaulze
s.
t
*5o
[xreur
LruK
+L S
2-
<-ovPl&
+L 54 Ooi?ut
LI^JK-
#, G 4
1%b
? r,.r €'Ltt^r.,4-ttt
CycrrtuCrs
Lrnl"t
5o,
Sz Ma
V- a(e W rr*ovi,t5 tinks
h &
4lar
$,*o+t"rr
Je,,,'<Ao..
-frtl
lory+)*s
{
actyW
and
o"\uk
link-s
Aft adryd,)il tL*
?
wuv^alu
cylind*,
4D
T*,d,
ce a.
fa*^
ty
of s*ar6t',.*" Lirr.e^s
4
di#ud
sQ<
6Ht
c-owt'wtr
y
\^terc?4t,
af"
(d
:
O
.
$i"3 <.
Fteuae
s. z-
W i'rdslrur
etJ
w
iper
B,lad e riqiUtv
Connq,1cd-ti cbc,gtvAB
ol
ln7,nt
LinlcorA
s.3
o,
Ftcau?e
tr\Gu Oe
\ vt
qoln
vnec.L,o:nis r,rr
\i,ne ovat o,t teast
?*t
of
,# ,,+,
s.4
pgr,vl
?
qeuetalesa
t+.oCt"t
lils
Tofu"
<{raigui \i,q
VtaOtr'an
fo l\on,er
#
=*ca'tqht
fivte
vtnduu'^.
piniu,n
Q"u)
(o
aT
CitM
trl 6uGe s,.
s
FI6UBE
s.
Q2
ta
tr l6u
rL€
s.?
ol
(a)
A
F(qu?e
s.6
trGu?€
s.1
?osit'o^
3
?osit o.,
Z
?osit'',n
\
F\quRE
S to
IBrser.I",
ol
Bz.Et
113isecto , ol
A.A3
/
IB1se,"lo.
ofB,B,
J- Bisec,to , o{
A
,A..
8,=D
P'P
=fi
trtquRs S.r\
JBistr.ob,
"f
A'.Nr
!8isec###BOT_TEXT###quot; "tts&
IBise"io.
ol
A, A.
A,= A
F
tOu RE
E< 12
2t'
=
3?J0ZlJ
Tesired
)
I
a*h'
Circl< r,t/itd
tr touRE
s, l4a-
\e
\o-
l-'
\c
\ r.^
\;
1
\ut
\e ###BOT_TEXT###lt;
vY-
it" \%
% \
\|\% \
\.d \
\P\
TI
o
C
p
cn
{
+
fl
\
h''
V
t\
7
r
#:.
/
ry
F
IGU RE
5 \tc-
11Z'r*
oYu
oP.
Ar.
ruil\
Orc uuit-{t
fadt.rr =
Q
COnt<
otA
r
r&dius =
O,c),
ce.nler
NctO,)
z
(og),
(o,).
o
o"
(Ar.
ruitl
fadilrs= PrO3
ce,'vrler
& P,
FIGURE
s d-_
o?,
A,
Arc 1^,i{'\
r adius=o,
cemkraL
A, \
_v
't;ru:*'";
lihc
u
fuicircta
oe
arc ru
iH"
radius =
Otob
c rznler ah(o r)
I
a
/
fC t^,'i{
fadrus=
P3O3
caal<r
&?t
trtGUt-2s S
l4e-
o?u
Co,)r
o3
ov'.
Q),
F tG uee
s.
taf
OPz
tr lGul-2E
=.
t|g
V alues of €mt
Vatuer
of
r
tr\GUR€
s.ls
,/*D.srted
.ft
nchon
f
cr)
FTGUZ€
s.tb
55.gB'
J0.00'
r2250'
78.350
Ft a
o&e s.t?
1l
o
d
(l
!
cp
P
'F
rl
d
qqt
's
3do9rc
n(o
t,gO.'trrr'rtr"'$::
"
a-
,og"'
\
=a,l
,0q
zz
/
v
,I[
BI
'(d
,00
0[
bz
.,O
B'
rrittttttlt'trrattttO
S
"
os)
u
C(Cru PF
S
l(.
TI
F
g
10
lrr
tn
:-
D
.0-
t
rs aznztJ