You are on page 1of 13

Mechanismand MachineTheory,1978, VoL 13, pp. 125-t3L Pergamon Press.

Printed in Great Britain

Location of Burmester Synthesis Solutions with Fully Rotatable


Cranks

K. J. Waldront

Received 17 December 1976

Abstract
A method is developed for locating regions of the Burmester circle-point curve which
give the driving cranks of crank-rocker or drag-link linkages when combined with a
given driven crank. The technique used is a graphical iteration procedure, When used as
suggested convergence is rapid. The technique is compatible with the author's
procedures for avoiding solutions requiring branch changes and ensuring correct order
of progression through the design positions.

1. Introduction
As is well known, not all planar 4-bar linkages possess fully rotatable cranks. It is usually
desirable to have such a crank so that the linkage can be driven by a continuously rotating
motor. Therefore, in any synthesis procedure, it is important to be able to separate solutions
which do possess fully rotating cranks. The classical Burmester synthesis procedure lacks any
convenient means of accomplishing this. Other than the trial-and-error approach, the only
available techniques are the rather laborious methods of Beyer[1] and Filemon[2, 3]. Both use
plots of linkage parameters against curvilinear distance along the circle-point or center-point
curve. Beyer plots link length sums whilst Filemon uses joint angles.
It is the purpose of the present work to develop a more efficient a priori method of locating
solutions with fully rotating cranks. The method developed is suitable for either graphical or
numerical use. It is also fully compatible with the author's solutions to the order and branch
problems of Burmester synthesis[4,5]. In combination with these techniques the method
described below eliminates almost all trial-and-error from the Burmester synthesis procedure.
The problem treated here has been referred to as the "Grashof Problem"[4] because
Grashof's rules are used to distinguish linkages with fully rotating cranks. It will be shown that
it is possible to distinguish linkages of any desired Grashof type, but most particularly the
crank-rocker type which is usually the most useful from the practical point of view.

2. Rules Based on the Order and Branch Solutions


Before proceeding to a method of precisely identifying regions of the circle-point curve
which give solutions of nominated Grashof type when combined with a chosen driven crank we
will develop some useful qualitative rules. These make use of information from the order[5]
and branch[4] solution techniques.

(a) Significance of image poles


It was shown in Ref. [5] that the image poles divide the circle-point curve into segments
which give constant order of progression through the design positions. That is, driving crank
circle-points chosen from the same segment all give driving cranks which move the coupler
through the design positions in the same order when driven with uniform direction. Clearly, the
selection of driving crank circle-point is restricted to those segments giving the desired order.
Let us suppose that one of the image poles bounding a segment which gives the desired
order is chosen as the driving crank circle-point. Now this point has the same position in two
adjacent design positions. If these design positions are finitely separated this implies that the

tAssociate Professor, Department of Mechanical Engineering, University of Houston, Houston, TX 77004, U.S.A.

125
126

crank has reversed direction between the two positions. A continuously rotating crank passes
through each of its positions only once in each cycle. Therefore a choice of driving crank
circle-point at the boundary of the curve segment giving the desired order cannot give a driving
crank capable of complete rotation. Circle-points chosen near the segment boundaries will also
not give cranks capable of complete rotation. Tfierefore: (a) The driving crank circlepoint
should be chosen on a segment of the curve giving the desired order but well away from the
image poles bounding that segment.

(b) Significance of Filemon lines


In Ref. [3 and 4] it is shown that, once a driven crank has been selected, a circle-point
outside a region bounded by two lines must be selected for the driving crank. If this is not done
the solution linkage will not be capable of traversing all design positions.
If a circle-point on one of the lines is chosen as the driving crank circle-point then, in one of
the design positions, the driven crank and coupler are aligned. This results in either of the
configurations shown in Fig. 1. In either case the driving crank is at an extreme position and
must reverse direction. Thus the driving crank cannot perform a complete rotation. Circle-
points on portions of the curve near the Filemon lines will also give driving cranks incapable of
complete rotation. Therefore: (b) The driving crank circle-point should be chosen on a segment
of the curve not excluded by Filemon's construction and well away from the bounding lines of
the excluded region.
Note that, since the circle-point curve is cubic, each Filemon line will have either two or no
intersections with the cubic in addition to the driven crank circle-point. Thus, Filemon's
construction allows, at most, three segments of curve for choice of driving crank circle-point.

DRIVING
CRANK DRIVEN

(a)

(b)
Figure 1. Configurations arising from selection of a circle-point on a Filemon line.

(c) Significance o[ the circle diagram


In Ref. [4] it is shown that solution linkages capable of traversing all the design positions
can only be obtained if the driven crank circle-point is chosen from certain segments of the
curve. These segments are identified by means of a diagram consisting of the line joining two
image poles with a common subscript and the circle whose diameter is bounded by those poles.
Repeated use of this diagram allows the permissible segments for driven crank circle-point to
be precisely mapped.
If a point lying at the boundary of one of these segments is chosen as the driven crank
circle-point, the angle between the Filemon lines becomes 180°, the two lines coincide and the
whole plane is excluded except for points on the coincident lines. Thus, as is explained above, a
fully rotating crank is impossible. If a point near the segment boundary is chosen, the angle
between the Filemon lines will be close to 180° and the usable region will be very small. Thus it
will not be possible to obtain a fully rotating driving crank since all circle-points in the usable
127

region will be too close to the Filemon lines. Therefore: (c) The driven crank circle-point should
be selected on a permissible region of the curve well away from the boundaries of that segment.
In practive, the rule (c) above would be applied first since the procedure for a~voiding branch
change [4] requires that the driven crank be selected first. Rules (a) and (b) would be supplied
together when selecting the driving crank.
One more rule, mentioned in Ref. [4] is useful. If a crank-rocker solution is sought it is
useful to select a driving crank circle-point from a region excluded for driven crank circle-
points by the circle diagram method. This ensures that the driving crank will rotate through
more than 180° relative to the coupler in traversing the design positions. Such a choice excludes
any possibility of a drag-link solution since the driving crank cannot rotate through more than
180° relative to the coupler in a drag-link.
The designer should also bear in mind that a crank-rocker solution is only possible if the
smallest net rotation of the coupler in traversing the design positions(in any order) is less than
180°. In fact it is difficult to get a crank-rocker solution for net coupler rotations much above
90° . Drag-link solutions are possible for net coupler rotations greater than 180° .

3. The Grashof Problem


Grashof's rules are:
(i) If i is the length of the longest link, s is the length of the shortest link and p and q the
lengths of the other two sides, the linkage can only have a fully rotatable crank if

I+s<-p+q, (1)

(ii) The shortest link must either be a crank or the base. If it is a crank the linkage is a
crank-rocker with the shortest link the one fully rotatable crank. If the shortest link is the base
the linkage is a drag-link with both cranks fully rotatable.
The presence, or absence, of a fully rotatable crank is a property of the linkage as a whole.
In this respect the Grashof problem resembles the branch problem. It is a question of
compatibility of the two chosen cranks. Thus the primary question is: Given a driven crank;
which driving crank circle-points produce, in combination with that crank, linkages which
satisfy Grashof's inequality (1)? The driven crank is chosen first for compatability with the
branch solution. If we can locate, on the circle-point curve, those driving crank circle-points for
which the inequality (1) becomes an equality, these points will divide the curve into segments.
Each segment will contain only circle-points which give linkages which either do or do not
satisfy the inequality. In principal this can be accomplished by expressing

l+s=p+q (2)

as an equation in the coordinates of the circle and center-points and solving it simultaneously
with the circularity conditions used in deriving the circle and center point equations. Unfor-
tunately the apparent simplicity of the above equation is deceptive. When expressed in terms of
the circle and center point coordinates it gives an eighth order equation, although the highest
power of any individual variable is the fourth. Using the quadratic circularity conditions to
eliminate the center point coordinates and obtain a locus which can be intersected with the
circle-point curve obviously gives very high order curve. Further, it is necessary to generate not
one, but three such loci to take account of the different possible locations of the longest and
shortest links. It is then necessary to remove from the resulting segments those on which the
coupler is the shortest link. Evidently this procedure is not practical. Therefore we take a
different approach which will be developed in the next section.

4. Iterative Location of Segment Boundaries


Since it appears impractical to directly locate segments of the circle-point curve which give,
in combination with a given crank, linkages which have fully rotating cranks, we will use an
iterative technique. Let us suppose that

l+s-p-q = 8L (3)

MMT VoL 13, No. 2--C.


128

where 8L is small relative to ! + s. Let us further suppose that the links are arranged as shown
in Fig. 2. X~X* is the chosen driven crank. YI Y* is the driving crank. The longest link is the
coupler and the shortest is the driving crank. Let us suppose that the driving crank circle-point
is moved a small distance 8s along the curve to point Z~, Z* being the corresponding
center-point. Let us further suppose that the linkage X~X*Z*Z1 satisfies eqn (2). Then

(1 + c$s cos/3) + (s + 8s cos a - 8s* cos a*) - p - (q + 8s* cos/3*) = 0

giving

8L + 8s cos/3 + 8s cos a - 8s* cos a* - 8s* cos/3* = 0

which can be rewritten

8L
8s (4)
(Ss*lSs)(cos a* + cos/3*) - cos a - cos/3"

It can be seen that we need the direction of the center-point curve tangent and the ratio
8s*lSs in order to use eqn (4) to locate the crank Z~Z* for which eqn (2) is satisfied. These may
be found without plotting the center-point curve by referring to Fig. 3. The basis of the result is
the fact that a crank of a 4-bar subtends angle o~d2 at the pole Pu where i = 2, 3, 4. Thus Y~ Y*
and Z~Z* subtend equal angles at the poles P~2 and P~3. Using the sine rule we get

8s Pi2 Yt 8s PI3 YI
&L sin~12 6v sinai3
@* PI2Y* 8s* PI3Y*
8/~ sinm, 8v sin~3"

Elimination of 8~ and 8v gives

8s* Pt2Y* sin ~12 P , s Y * sin f~s (5)


8s PI2YI sin *!12 P13YI sin t/13"

Now, the angle subtended at a center-point by the line Pt2PB is ~ 1 2 [4]. We use the sign
convention that angles are positive in the clockwise direction. Thus, in Fig. 3 d~23/2 is negative

CI RCLE- POINT ~ , , ~ ) Xt
: ~ TANGENT ~.-~ /

/y,~ q vX

Figure 2, Solution linkages derived from two different driving-crank circle-point


selections separated by a small distance 85.
129
CIRCLE- POINT
TANGENT

"r/=z

CENTER - POINT
TANGENT

Bu

8u

v
P,,
P,,
Rgure 3. Angular relationships for two circle-points separated by a small distance 8S
and the correspondingcenter-points.

and we have

¢•23
r/t3 = 'rh2 2 " (6)

Combining eqns (5) and (6) to eliminate ~hs we get, after some manipulation

1 [ ~ PI2YIX PtsY* sin ~t3]


cotv/12-sin(~b2s/2) cos 2 PiEY*XPl3Y=sinse,2J" (7)

Equation (7) can be used to find angle vh2 and hence the direction of the center-point
tangent from the direction of the circle-point curve. "q~2can then be substituted into eqn (5) to
give 8s*lSs. Thus we now have the data needed to use eqn (4) to locate a driving crank
circle-point, Z, which, in combination with the chosen driven crank X~X* gives a linkage which
satisfies eqn (2). However, use of these equations in their present form in combination with eqn
(4) requires no less than 14 careful measurements on each iteration. This can be reduced to 8 by
making use of the angle relationships shown in Fig. 4. We have

a--~12-?--~13 'q2/is
a*=-~+'r/]2 -~+'r/13
and, using the sine rule

P n Y, - P n Y* Y, Y*
sin ~1212 sin ~12/2 sin 0n12
PIs YI -Pl3 Y* Y1Y*
sin 4~1s/2 sin ~,3/2 sin O,s12"
130

CIRCLE-POINT
TANGENT

~s
2

CENTER-POINT
TANGENT

m
2

Plz
Figure 4. Angular relationships for poles P12 and P13, a crank Y1Y* and the circle and
center-point curve tangents at Y1 and Y*.

Also
012= 4h2+"Igz and 013= ~bl3+~13.

Substitution of these relationships in eqn (7) gives, after considerable manipulation

tan a* = pl2(l - cos ~13) - p13(1 - cos ~b12)


P12 sin ~13 - p 1 3 sin ~bl2 (8)

where pli = cos ot - c o s (a + 01i - t~li); i = 1, 2.


Substituting into (5) we get

8s* sin ( ~ ) sin (2a + 022- ~bt2)


- (9)
8s . ~b,2 . / 2 a * - & l z )
sm T sin [~

These equations are convenient because 012 and 013 are constant for all circle-points (0o is
the angle through which the coupler rotates in moving from position i to position j). Thus these
equations require repeated measurement of only qh2, dh3 and a.
Of course, the above technique is only approximate for finite 8L. It is, in fact, a graphical
equivalent of a Newton-Raphson algorithm and can be used in a similar iterative fashion. As
will be seen in the example below, convergence is usually rapid and it is seldom necessary to
use more than two iterations to get within drawing accuracy.
Equation (4) above is only good if the longest and shortest links are in the positions shown
in Fig. 2. There are, in fact, 12 possible distinct arrangements. In order to conveniently describe
them we adopt the convention of writing the link lengths in the following order: driving crank,
coupler, driven crank, base. Thus, the arrangement shown in Fig. 2 is written slpq. There are
three possible distinct arrangements with the driving crank shortest slpq, splq and spqi. Notice
that the order of p and q is immaterial. The other 9 possible arrangements are obtained by
permuting these three so that the shortest link becomes, successively, the coupler, driven crank
and base. In each case eqn 4 assumes a different form except that arrangements obtained from
one another simply by interchanging 1 and s have the same equation. The different expressions
obtained for (BLISs) are summarized in Table 1.
131

Table 1.

Link arrangement

spql: lpqs
(g>(COSa*-coSp-cosatcosf3
- $
qslp; dSP
( > a* - (cos cos p*) + cos a - cos fi

qspl; PI@ - g (COSa*tcoS/3*)tcosa-cos~


( >

qpsl: pqls - $ * (cosa*tcos/3*)tcosntcos/3


( >

The way in which the above technique can be used to map regions of the circle-point curve
which give linkages which satisfy Grashof’s inequality is demonstrated by the example given
below. By first applying the rules for avoiding branch change and obtaining the desired order of
progression through the design positions, the length of curve to be examined is drastically
reduced. Obviously there is no point in applying the present method to curve segments which
can be excluded on other grounds. This assists in two ways. Firstly it reduced the number of
curve segments to be examined to two, or zero, and hence reduces the number of points
satisfying eqn 2 to be found. Secondly, the relatively short segments which result are unlikely
to have more than two points which satisfy eqn 2.
Once the segments of the curve on which the Grashof inequality is satisfied are identified, it
is necessary only to locate the shortest link of the linkage given by any circle point on that
segment in order to determine the Grashof type everywhere on the segment. It might be thought
that the type could change as one moves along the segment, for example a crank-rocker could
change to a drag-link. However, this is not, in fact, possible. Consider the transitional case
between two linkage types which satisfy the inequality. The linkage type is determined by the
location of the shortest link. Thus, in the transitional case, two links have equal length and both
are shorter than the other two sides. Thus s = p and

1+ssq+s.

Therefore the Grashof inequality is not satisfied. Hence the Grashof type cannot change on any
segment on which the inequality is everywhere satisfied.

5. An Example
Figure 5(a) shows four positions of a line segment AB fixed in a moving lamina. The
auxiliary diagram is used to find the angles &j and the rotational order of the design positions at
infinity[S]. The poles and image poles have also been located. Figure 5(b) shows the circle-point
curve and the segments into which it is divided in the course of the branch solution[4]. The
thick portions of the curve are those from which the driven crank circle-point can be chosen.
X1 is chosen as the driven crank circle-point and XZ, X3, X4 are generated by making triangles
A&X,, AzBzXZ, A~BsX~ and A.&X4 congruent (construction not shown). The driven crank
center-point X* is located at the center of the circle on which XI, X2, X3 and X4 lie. Figure S(c)
shows the construction of the Filemon lines [4]. Either the method of Ref. [4] or that of Ref. [6]
can be used. The latter is probably more convenient and the construction shown here is as for
that method. The auxiliary diagram is used to find Yij from the angle Yij/2 subtended by the line
joining the image poles Pik and Pjk at the circle-point X1. The segments giving order 1234, or its
reverse[S], are shown on this figure as the thick segments of the curve.
The various constructions used in Figs 5(b and c) are needed to avoid solutions with branch
change or solutions which move through the design positions in the wrong order. Thus far none
132
oPt3

opt,

op;, P;,
I e A, oPi4 oPl,~ Bz

2 ~ B,

Figure 5(a). Four design positions A/B,, auxiliary diagram, poles and image poles for
example,

TI4

T~/

Xj
Q34

Q=4
Ui2

T2~
Figure 5(b). Circle-point curve and segments mapped for driven crank selection.
Selected crank is X1X*.

P;,

-2-

Figure 5(c). Filemon's construction for example.


133
of the techniques described in the present paper have been employed. Notice, however, that the
portion of the curve from which the driving crank circle-point can be selected is already quite
restricted, Only the segments CID1 and ElF1 can be used. It is this reduction in the length of
curve to be considered which makes the present technique viable.
In order to find the curve segments which give cranks which, in combination with X * X ,
give linkages satisfying the inequality (1) we will start with points CI, DI, El and FI as initial
guesses. This is, of course, not obigatory. However, it ensures convergence to both end points
of each segment which satisfies the inequality since, for reasons explained in Section 2 above,
points C1, DI, El and FI certainly lie outside these segments. Initial choices within the segments
might lead to repeated convergence to the same point.
Using point CI as initial guess we first draw the circle-point tangent at Ch as shown in Fig.
5(d) and assign a positive direction on it. Next the corresponding center-point C* is located. A
particularly convenient method of doing this is by constructing <CIPnEC equal to 01d2 and
< C I P ] 3 c ' equal to 013/2. The intersection of Pi2c and P]3c' is the center-point C*. Note that the
signs of the angles are important. If 012, as given by Fig. 4(a), is clockwise <CIP12c must be
clockwise about P12 and vice-versa. Similarly <C~Pi3c' must have the same direction about PI3
as 013.

, ~ CIRCLE- POINT
C' / TANGENT

CENTER-POINT ~ C ° -

---...,.

P,2

Rgure 5(d). First iteration: location of center-point C* and tangent to center-point curve
corresponding to circle-point C1. Computer position of circle-point
approximately satisfying Grashof equation is J1.

The lengths of the four links X ' X 1 , XICI, C|C*, C * X $ can now be measured. They are

X * X n = 2.80 = p

Xi CI = 2.40 = q
C I C * = 1.68 = s

C'X* = 5.41 = !.
Therefore
8 L = l + s - p - q = 1.89.

Notice also that, in the notation of Table 1, the link arrangement is sqpl. Since p and q are
interchangeable we use the table entry for spql to give BLISs. Hence

8L
8s ( S s * / S s ) ( c o s a * - cos/~*)- cos a + cos f f (12)
134

The angle a from C'C1 produced to the positive direction of the circle-point tangent is now
measured. Once again, it is important to assign it the correct sign. a is positive if, when
measured from C*C~ produced to the positive tangent direction, it is clockwise. The angles
t~12/2 and $13/2 are also measured. $12/2 is measured from C'C1 to P12C*produced and $~3/2 is
from C'C1 to PI3C* produced. Both are positive if clockwise. Ot2 and 013 are assumed to have
already been measured from the auxiliary diagram of Fig. 5(a).
Equation (8) can now be used to calculate a*

012 = 34.0° 013 = 70.6 °


$t2 = 149.0° $13 = - 1 0 1 . 0 °
a = 170.7°

giving

p12 = -1.981 pl3 = 1.940

and hence

a* = 52.7 °.

Substitution in eqn (9) now gives

85'
- 1.290.
8s

The center-point tangent can now be drawn at angle a* to C'C1 as shown in Fig. 5(d). Angle/3
is measured from XICt produced to the positive direction of the circle-point tangent. Similarly
/3* is measured from X'C* to the positive direction of the center-point tangent. In this case
they are

B = 81-3° /3* = -52.8 °.

Substitution of these values and the value of 8L obtained above in eqn (12) gives

8S = 1.66.

Rgure 5(e). Second iteration: location of center-point J* and tangent to center-point


curve corresponding to circle-point Jl. Computed position of circle-point
satisfying Grashof equation to drawing accuracy is/(1.
135

Measuring a distance 1.66 units along the curve in the positive direction of the circle-point
tangent we get the point J~. Here the entire process is repeated, Fig. 5(e). The actual numbers
are

X ' X 1 = 2.80 = p
X d t = 2.74 = q
8L = I + s - p - q =0.55
J d * = 1.71 = s
J ' X * = 4.38 = !
~12 = 139.6 ° ~13 = -131.4 °

a = 150.5 °

giving, from eqn (8),

pl2 = --1.579 p13 = -1.862

and

a* = 15.4°.

Using eqn (9)

85*
- - = -1.034.
8s

The center-point tangent is now drawn in as shown in Fig. 5(e). Measuring from the figure

/3 = 77.0° /3* = -55.8 °.

The link arrangement is still sqpl so we use eqn (12) again to get

8s = 0.81.

Measuring this distance along the curve in the positive direction of the circle-point tangent we
get point Kt. Construction of the corresponding center-point gives K*. Then

X*Xt = 2.80 = p
XIK, = 2.94 = q
8L = l + s - p - q = -0.01.
K~K* = 1.84 = s
K ' X * = 3.89 = I.

Thus, despite the fact that 8L at C~ could certainly not be described as small in relation to
! + s, 8L has been reduced to 0, within drawing accuracy, in only two iterations. This process is
repeated with DI, El and F~ as initial guesses to obtain points Lt, M~ and N~, Fig. 5(f). In no
case is more than two iterations needed. In fact the starting 8L's are, in all three cases, very
much smaller than at C~.
We conclude that circle-points on the segments KIL~ and M~N~ give solution linkages
which satisfy the Grashof inequality[l]. It is, of course, possible in principle for there to be an
additional pair of zeroS of 8L on either segment. However, in view of their short length this is
quite unlikely. It can easily be verified that it does not happen in the present example.
At all four points K~, L~, Mr, N~ the driving link is the shortest link. Thus, we conclude that
the segments KILt and MtN~ give only crank-rocker linkages.
Since K~L~ is considerably shorter than M~N~ the driven-crank circle point Y~ is selected on
136

K, /
L,

Figure 5(f). Segments of curve giving solution linkages which satisfy the Grashof
inequality are K1L1 and MIN~. Y1 is the chosen driving crank circle-point.
X2

B~
A3 X~
X3
B3

Y3

A4

~ Q

Y,
A~
Figure 5(g). The solution linkage in the four design positions.

the latter segment. This is done because, since KIL~ is relatively short, the magnitude of 8L can
be expected to be close to zero for all linkages obtained from that segment. This leads to
linkages with poor force transmission characteristics. Figure 5(g) shows the four positions of the
solution linkage.

6. Conclusions
It has been demonstrated that, for a given choice of driven crank, it is possible to locate all
choices of driving crank which give, in combination with that crank, solution linkages which
satisfy the Grashof inequality. The technique involves an iterative construction. If it were
necessary to consider all of the circle-point curve in this iterative process the method would not
be viable due to the potentially rather large number of Grashof neutral points which would have
to be located. However, as has been shown in Refs. [4 and 5], only limited segments of the
curve give solution linkages which are viable on other grounds. The boundaries of these
segments are shown in Section 2 to have properties relevant to the present problem. They are
used as initial points for the iteration in the example of Section 5. It is this restriction of the
segments of the curve to be searched which renders the method described here viable.
The importance of making the techniques developed in this paper available in numerical
137

form cannot be overstated. The complexity of graphical Burmester synthesis has been respon-
sible for the popularity of numerical methods based on the same theory but implemented on
digital computers. Space limitations dictate that a numerical formulation of the present theory
must be left for a later paper. In analytical format the present method is recognizable as simply
a Newton-Raphson solution of a system of simultaneous nonlinear algebraic equations. This
system consists of the equal distance conditions of the circle and center-points plus Grashof's
equation.
The feasibility of incorporating the branch and order solutions of Refs. [4 and 5] into
numerical solutions to restrict the segments of curve which must be considered has recently
been demonstrated by Spitznagei[7] who used the branch solution in an interactive graphic
display program.
The development of a practicable solution to the Grashof problem, along with the, earlier
solutions of the branch and order problem, finally eliminates all trial and error from the
Burmester synthesis technique. Of course a range of choices both for the driven crank and
driving crank is still presented to the designer. However, he is assured that the solution will not
be defective and will have the required Grashof type. Thus he can select for other charac-
teristics such as compactness, or eveness of driving torque. This is particularly true if the more
laborious parts of the process are handled on the computer.

Acknowledgement--The work described in this paper was supported by the National Science Foundation, grant number
ENG 75-20889.

References
I. R. Beyer, The Kinematic Synthesis o/ Mechanisms. McGraw-Hill,New York (1963).
2. E. Filemon, In addition of the Burmester theory. Proc. 3rd World Cong./or Theory o/Machines and Mechanisms. D,
63-78 (1971).
3. E. Filemon, Useful ranges of centerpoint curves for design of crank-and-rocker linkages. Mechanism and Machine
Theory 7, 47-53 (1972).
4. K. J. Waldron, Elimination of the branch problem in graphical Burmester mechanism synthesis for four finitely
separated positions. J. Engng Ind. Trans. ASME. Series B, 98, 176-182 (1976).
5. K. J. Waldron, The order problem of Burmester linkage synthesis. J. Engng Ind. Trans. ASME. Series B, 97, 1405-1406
(1975).
6. K. J. Waldron, Graphical solution of the branch and order problems of linkage synthesis for multiply separated
positions.ASME Paper No. 76-DET-16 to be presented at Design Engng Tech. Conf., Montreal, Canada, and published
in J. Engng Ind. (1976).
7. K. Spitznagel, Near-Global Optimum o/Synthesized Four-bar Mechanisms by InteractiveUse of Weighted Kinematic
Sequential Filters. M.S. Thesis, University of Florida (1975).

RECHERCHE DES S O L U T I O N S DE LA M E T H O D E DE S Y N T H E S E DE B U R M E S T E R QUI ONT LES M A N I V E L L E S CAPABLES


D'UNE R E V O L U T I O N COMPLETE.

K.J. Waldron

R ~ s u m ~ - On a d ~ v e l o p p ~ une m ~ t h o d e pour trouver les r ~ g i o n s de la c o u r b e des p o i n t s circulaires


de B u r m e s t e r qui d o n n e n t des m a n i v e l l e s pouvan~ tourner compl~tement q u a n d on les a s s o c i e avec
une d e u x i ~ m e manivelle sp~cifi~e. La m ~ t h o d e est une p r o c e d u r e d'it~ration graphique. Quand
on l ' u t i l i s e de la m a n i ~ r e propos~e, elle c o n v e r g e rapidement. La m ~ t h o d e est c o m p a t i b l e avec
les m ~ t b o d e s de l ' a u t e u r pour l'~limination des s o l u t i o n s donnant naissance ~ des e m b r a n c h e -
ments et qui a s s u r e n t que le m o u v e m e n t par les p o s i t i o n s successives sp~cifi~es se fera dans
l'ordre d~sir~.

You might also like