True Straight-line Linkages Having

a Rectilinear Translating Bar

E.A. Dijksman

Precision Engineering
Faculty of Mechanical Engineering
Eindhoven University of Technology
The Netherlands

Summary - Eight- and ten-bar straight-line mechanisms with a rectilinear translating bar, have been
derived from two basic types, namely from Watt's - and from Stephenson's type of Hart's true
straight-line mechanism(s). Though the generalization of Watt's type, representing Hart's inversor,
was easily done, the other one, namely Stephenson's type, representing Hart's 2"" straight-line
mechanism, required a new design circumventing Burmester' s design of focal linkages. In the end
spectacular results emerged based on a new but easy going design of these true straight-line
linkages. Because of the simplification obtained here, the applicability of these "lifting devices" now
comes within easy reach of the designer.

1. Introduction

Planat linkage mechanisms producing an exact straight-line are very rare. The more so
when the elementary pairs contained in the mechanism, are restricted to revolute pairs
(i.e.tuming-joints) only. Four-bar linkages for instance, do not produce a straight-line at
all. But, six-bar linkages may produce them, provided the dimensions are specifically
chosen. It appears that principally only two solutions exist. Both have been found by
H.Hart, ref.[l], and are known as respectively, the first and the second straight-line
mechanism of Hart. The first one is a six-bar of Watt's type, whereas the second one may
be recognized as a six-bar of type Stephenson, ref.[2].
The first one represents the inversor mechanism (figure 1), while the 2nd one represents a
special case of Kempe's focal mechanism (figures 2 and 9), ref.[3,4].
Hart's inv~rsor has further been generalized, which lead to the so-called quadruplane
inversor of Sylvester and Kempe (figure 3). The result though, may stiH be recognized as
a six-bar linkage mechanism of Watt's form.(ref.[S])
The generalization of the focal type, however, does lead to an eight-bar but then contain-
ing a rectilinear translating bar. (figure 4, see also figure 10 of ref.[4])
4ll
A. J. Lenartic and B. B. Ravani (eds.), Advances in Robot Kinematics and Computationed Geometry, 411-420.
© 1994 Kluwer Academic Publishers.
412

Eight-bar linkages with such a rectilinear translating bar, may also be obtained from a
design based on Harts' inversor. (figure 5, see also figure 10 of ref.[6].)

jJ

1" straight-line mechanism of Hart

mechanically interconnected with the
Peaucellier-Lipkin inversor of 1864.
Figure 1

Two curve-cognates each representing the

2.. straight-line mechanism of Hart(1877)
Figure 2

Quadruplane inversor of Sylvester and

Kempe (a generalization of Hart's inversor)
Figure 3

8-bar linkage mechanism with bar A "B"

moving perpendicular to the frame
A random 4-bar and a re-
Figure 4
flected similar one, built
on tap Figure 7
 413

Ao

Hart's 2"" straight-line mechanism

incorporated Figure 8

J c
8-bar linkage mechanism with a rectilinear
moving bar PR (the random contra-parallelogram,
ASĂS, represents a sub-chain of the 8-bar)
Figure 5

Hart's 2"" straight-line mechanism

obtained by multiplication Figure 9

8-bar coupler cognate

having a bar A "B"
moving perpendicular 8-bar linkage mechanism containing a
to the frame Figure 10 bar PR' moving in an invariable but
OB~nA.mA ",.,. 0.8"4..t,.A"' oblique direction with the frame Figure 6
.$'/",rrvw-"" ..,.;;"" o/
~~mN'br# ,

2. Eight-bar Inversor with rectilinear translating bar

In practice, only one equilateral dyad-linkage is needed to adjoin Hart's inversor in order
to obtain an 8-bar with a translating link moving perpendicular to the frame, (figure 5).
The mechanism contains a contra-parallelogram and two identica! dyad chains, remaining
parallel during the motion.
The same procedure may be applied at the quadruplane inversor of Sylvester and Kempe.
414

Then, the equilateral dyad PR'D' adjoined to the 6-bar, remains parallel to the identical
dyad OQ0 'Q' giving a rectilinear translating bar PR' moving straight with respect to the
fixed link OOo'. (figure 6) The angle enclosed between the fixed link and the direction of
the straight-lines, traced by points of PR' , then differs from 90° and equals
(.i. S'DA - 90°).
We conclude that this kind of generalization practically results into a change of direction
of the straight motion of the translating bar. (See also figure 27 of ref.[?].)

3. Six-, eight- and ten-bar linkages derived from the focal type

Hart's 2m1 straight-line mechanism, which is a six-bar of type Stephenson, is to be general-

ized in the way Kempe did with his focallinkage.(ref.[8] and [4])
However, it is much easier to follow a different road: Doing that, one obtains Hart's 2m1
straight-line mechanism as a besides. In order to design the mechanism as successively
drawn in the figures 7 and 8,
l. one starts with a completely random four-bar AoABBo ,
2. then, one adjoins the chain Ao' A'B'B 0 ' in such a way that the 4-bar Ao' A'B'B 0 ' is
reflected similar to the initial four-bar AoABBo , whereas
=
Ao' B and B0 ' A. =
3. One further adjoins the linkage-dyad B'C'B in such a way that B'C'BA' forms a
linkage parallelogram. ·
4. One finally adjoins the bar C'C0 ' , such that BoBC'C0 ' too forms a linkage parallelo-
gram, of which B0C0 ' represents the frame-link.

The result is a 10-bar linkage mechanism of which the bar A'B' moves perpendicular to
the frame.

Considering the stage in which the three bars C'B , C'B' and C'C0 ' are not yet installed
and assuming that the 4-bar Ao' A'B'B 0 ', built on top of the random 4-bar AoABBo ,
remains reflected similar with the latter, it is still easy to prove that A'B' remains parallel
to the frame AoBo:
<l: (B'A',A.Bo) = <l: AAoBo- <l: AnAB'+ <l: A'B'Bo' =

=<l: AAoBo- <l: AnAB + <l: AoBoB + <l: BaBA = CJ, (1)

practically using the fact that the four angles of the random four-bar are ali tobe indicated
=
at the tuming-joint A B0'. (The sum of them being 360°.) Thus, if we draw a horizontal
passing through A, the bar B0 'B' enclofes the same angle J.l with the horizontal as well as
 415

with tbe bar A'B'.

Further note that because of the reflected similarity the bar AB encloses tbe same angle
with the fixed link as with the bar A'B'. (Actually, this remark represents the sbortest
proof for the parallel motion of A'B')
Tbe reflected similarity of the four-bars may be enforced in different ways. One of them
is sbown in figure 8 througb the adjoining of the three bars baving C' as tbeir common
joint. Tben, tbe parallelism of tbe bar A'B' is enforced. But otber methods are allowed
too. (Think, for instance of the later to be explained possibility to replace the bar C'B by
a bar interconnecting M and C 0 'C', simultaneously avoiding the appearance of linkage
parallelograms.)
Notwithstanding the sbown parallel motion, it still necessary to prove that a singular point
of A'B' moves perpendicular to the frame. Cboosing the point B', it suffices to demon-
strate that the dyad-cbain AoAB' represents an eccentric crank-and-slider mecbanism.
Whence, it remains to prove that tbe eccentricity "e" of tbat mecbanism sbows to be a
constant.
Clearly, tbe distance e between the fixed center Ao and the perpendicular let down from
B' on to tbe fixed link may be determined by projection of tbe dyad AoAB' on to tbe
frame:
e = a.cos cp - c' .cos J.l = a.cos cp - (bc/d)cos J.l (2)

in wbicb cp = <l: AAoBo and J.l = <l: BoB A (3)

(thus, cp and J.l are opposite angles in the initial four-bar)
Further note, that the subsequent dimensions of the reflected similar four-bar are respect-
ively: a' = ab/d ; b' = b2/d ; c' = bc/d and d' =b , (4)
tbe initial dimensions being a, b, c and d respectively.
Tbe Rute of Cosine twice applied in tbe random four-bar (abcd) on top of wbicb the
reflected similar four-bar (a'b'c'd') bas been built as sbown in figure 7, gives

(5)

Wbence after substitution apparently

(6)

Tbus, since the eccentricity remains a constant, point B' bas to trace a straigbt-line
running normal to tbe frame, tbat is presuming the reflected similar four-bar on top of tbe
initial one, remains reflected similar. So, ali in ali we proved that the motion of the bar
B' A' is in fact a rectilinear translation in the mentioned direction.

At this point, it is easily seen that <l: AoAB' = <l: C0 'C'B',wbence the five-bar AoAB'C'C0 '
alone represents a constrained linkage-pentagon. It is further possible to find a singular bar
416

interconnecting the sides M and Co'C' of that pentagon. To obtain that bar, we observe
the pantograpb BoBC'C0 ' , simultaneously taking M as the fixed link. Point B of tbat
pantograpb then traces a circle about A with respect to M. wbence equally the intersec-
tion (AoB x C0 'C') of AoB and C0 'C' traces a circle about the intersection of M and a
line running parallel to AB but meeting the point (AoB x C0 'C'). Tbe resulting six-bar
containing the pentagon and the adjoined bar appears to be Hart's 2nd straigbt-line linkage
mecbanism since among others B' traces a straigbt-line.
Multiplying the six-bar with the factor A 0 DIA0 B 1 in whicb D = AoB' x BoA' then
results into an enlarged Hart's 2m1 straigbt-line mecbanism, also sbown in figure 9. This
mecbanism tben consists of tbe pentagonal linkage AoEDCB 0 and the bar AB. Naturally,
tuming-joint D of tbe pentagon similarly traces a straigbt-line like B' did in the initial six-
bar.
Tbe six-bar obtained this way, is to be set up by the initial four-bar AoABB 0 and the
additional linkage-dyad CDE. (Note tbat any four-bar contains sucb a unique point D. Tbe
point plays a center-role in further generalization.)

It is quite possible to turn the 10-bar into an 8-bar linkage. To carry this out, one
multiplies the reflected four-bar Ao' A'B'B 0 ' geometrically from Hart's dyad-joint D.
(figure 4) Tbe multiplication-factor is then decided by tbe cboice of, for instance, B" at
AoD. Wbence the factor equals DB"IDB'.
Thus, D Ao"A"B"B 0" = (DB"IDB').D A0' A'B'Bo', (7)
Note that corresponding vertices of these two four-bars remain at the same ray joining the
ray-center D.
The initial four-bar (AoABB0) and the obtained one (Ao"A"B"B0 ") are furtber interlinked at
tbe common tuming-joints E" and C", being the intersections of corresponding sides.
Thus, E" = M x B"B0 " and C" =BoB x Ao"A". (The connectivity at E" for example is
due to the fact that the eccentric slider-crank AoAB' bas been turned into tbe similar one
AoE"B" througb the similarity center A0 of these sub-cbains.) In fact, the former common
joint A bas now moved to E" while former common joint B moved to C".
Clearly, the bar A"B" moves perpendicular to the frame. Now, tbe parallel motion of
A"B" doesn't bave to be sustained by linkage parallelograms, needed earlier for the
motion of A'B'. Tbus, an eigbt-bar is obtained with a rectilinear motion of A"B" moving
normal to the frame.
However, in case the rectilinear motion is badly transmitted - as may be the case
sometimes - one still bas the possibility to adjoin linkage parallelograms to sustain the
motion. To avoid the then appearing overconstrainedness, one may simultaneously omit
for example the superfluous bar Ao"B 0 ". The result is a 10-bar linkage to be recognized in
figure 4.
 417

4. The cognate 8-bar

Hart's 2"d straight-line mechanism occurs in pairs. That is to say, for each mechanism of
this type, a curve cognate exists producing the same part of a straight-line. (figure 2 and
also figure 9 of ref.[2].)
Both cognates may further be generalized in a way as demonstrated in the figures 4 and
10. However, it is quite possible to let them also have the same (rectilinear) moving bar
A"B". To attain this, one takes for both generalizations the same point B" at the ray AJ)
(and/or the same point A" at DB 0 .). The two generalizations then appear to be each others'
coupler cognate producing the identica} (rectilinear) motion for the common bar A"B".
Figure 11 demonstrates the two of them in one figure, whereas figures 4 and 10 show
them separately.
So, for each point B" at AJ), two 8-bar coupler cognates exist, producing the same
straight motion perpendicular to the fixed link they have in common .
..b (ceAke ~h!'i:A-/"'!lJl/onj

Two 8-bar coupler cognates having

a common bar A "B" moving per-
random choice for E"' (or C'")
pendicular to the common
Generalized 8-bar with link A '' 'B'''
frame AoBo Figure 11
moving normal to BoFu"' Figure 12

5. Special Cases

All 6-, 8- and 10-bar straight-line mechanisms of the focal type, have the advantage of a
random choice for the dimensions of the initial four-bar AoABBo contained in them.
Naturally, particular examples are found by taking specific four-bars to start the design
with. lf, for instance, the initial four-bar resembles a deltoid- linkage, possessing two pairs
of adjacent equal sides, the well-known 10-bar linkage of A.B.Kempe is obtained.
(ref.[7]). Note that in this case AB resembles the angle-bisector of the angle AoAB' .
In case the initial four-bar AoABB 0 represents a contra-parallelogram, the adjoined four-bar
418

Ao' A'B'B 0 ' coincides with the initial one. So, the design collapses. Section 2 then shows
how to proceed in that case.

6. Generalization through stretch-rotation

In figure 4 point E" joined the input-crank AoA. It is still possible, however, to generalize
the design by a completely random choice of this point. See, for instance figure 12 in
which we took the point E"' as a random point attached to the input-crank AoA. The
corresponding design of the generalized 8-bar straight-line mechanism having a rectilinear
moving bar would then read something like:
a. Start with the random choice of the 4-bar AoABB 0,
b. Make the four-bar O BA'B' A reflected similar with the initial 4-bar AoABB 0 ,
c. Determine the intersection point D = AoB'x BoA',
d. choose a random point E"' in the moving plane of the input-crank AoA.
e. Stretch-rotate AoAB' about Ao into AoE"'B'" determining point B"',
f. Stretch-rotate D BA'B' A about D into the 4-bar Ao"' A"'B"'B 0" ' ,
g. Form the similar and rigid triangles BoBC"' and AoAE"',
h. Finally, form the similar and rigid triangles B0 '"B"'E"' and Ao"'A"'C"',
i. Note, that 6.. B0"'E'"B"' = 6.. AE"'Ao = 6.. Ao"'C"'A'" = 6.. BC"'B0,

The choice of E' '' bas the advantage of choosing the motion-direction of the translating
bar A"'B"'. This bar moves in a fixed direction under an angle of (7t/2 + 6.. E"'AoA)
with AoBo· The length of the straight-line is govemed by the utmost positions of the
eccentric slider-crank AoE"'B"'. Whence, the length L of the straight-line, traced by the
point B"', is determined by the values of (A 0 E 711 + E 711B111 ) , IA0 E 711 - E 711B 771 1
and by the eccentricity e'
In order to sustain the rectilinear motion it is allowed to adjoin the three bars A'' 'F''',
E"'F"' and F0"'F'" such that E"'B"'A"'F"' as well as AoE"'F"'F0" ' form linkage-
parallelograms. After omitting either one of the bars AB or Ao'"B 0" ' , one obtains a ten-
bar linkage-mechanism still containing a bar always moving perpendicular to a fixed line
(F0 "'B0) in the frame, as proved hereafter:
Since for the 5-bar F0"'F"' A"'C"'B 0 , the 6.. F0"'F"' A"' = 6.. B0C"' A"', the rectilinear
motion of A'" will be perpendicular to F0"'B 0, the frame of the 5-bar. (See Wunderlich's
proof with isotropic coordinates in ref. [9])
Further, as the 5-bar F0"'F"' A"'C"'B 0 represents part of the foca! mecbanism, it is
possible to tind a bar interlinking the cranks F0"'F"' and B0C"'. This bar too may
replace the then superfluous bars AB and A0' "B0 ' " , again creating another 1O-bar linkage
with a rectilinear moving bar. The mechanism so obtained then represents Hart's 2ru1
straight-line mechanism adjoined with two linkage parallelograms, AoE"'F"'F0" ' and
E"'B"'A"'F"', in order to produce the parallel motion for the rectilinear moving bar.
 419

Thus, from the 8-bar, many alternative 10-bar linkages are to be derived producing the
same rectilinear motion with sometimes better transmission.
The designer may choose which one he prefers.

s'?"+l---- -UJT--- -]'

IJ/, '

1 1
1 1
~1 1
1
1
1!
d
1 1
1
1
1

7. Zero eccentricity

In the particular case for which B'" traces a straight-line passing through Ao. the
eccentricity e' of the slider-crank AoE'"B'" equals zero, whence e =O and so
a2+ d 2 = b2 + c 2, since
e' = (A 0 E71! 1A 0 A).e = A 0 E 111 (a2+d2-b 2-c 2)/(2ad) (8)

(See figure 13 for vertical translation and figure 14 for a horizontal one.)
Then, the initial four-bar has equal transmission angles (~ 1 = ~2 ) in the two aligned
positions of the crank AoA with the frame AoBo·
And, since e = O, we find that

E 111B 1111A0 E 111 =cos <p 1 cos~= (bc)/(ad) (9)

=
If, additionally, E 111B 111 = A 0 E71! , we find that ~ ± <p and bc = ad (10)
Then, the initial four-bar has to be a kite, whereas the bar B' "E"' generates the elliptic
motion.
420

8. Historical Note on Linearizers (i.e. linkages with a rectilinear moving

link)
Apart from the mentioned literature of Hart and Kempe, a booklet of the russian L.D.
Ruzinov (ref.[lO]) bas been devoted to this subject. He mentioned more inventors, such as
Gagarin and others, but ali linkages of this kind seem to have been restricted to those
containing either deltoids (i.e. kites) or contra-parallelograms.
Ruzinov, for example, wrote, quote: "It is not possible to produce a linearizer based on a
four-bar chain", unquote, (page 61 of ref.[lO]). Though he searched for a more general
approach, he didn't succeed and really thought that such was not possible and only very
particular linkages were able to meet the rectilinear motion. That in reality, a random
four-bar can be used as a base to start the design of a linearizer with, he never uncovered.
Remarkable is also that even Kempe, who founded the generalization of his compound
linkage, later named the focal linkage by L.Burmester, did not saw the connection
between his three kite- or deltoid-linearizers and his own interconnected, so-called
conjugate four-piece linkages. Special cases of the latter were either restricted to Hart's 2"d
straight-line mechanism or to other more general linkages, but not to linearizers, having a
complete link moving rectilinear.
We conclude that the general approach as developed in the underlying manuscript, really
represents a generalized method simultaneously interconnecting ali linearizers known to
mankind.

References
[1] Hart,H.: A parallel motion, Proc.London Math.Soc.6(1875),p.l37-9
[2] Dijksman,E.A.: Six-Bar Cognates of a Stephenson Mechanism, Joumal of Mecha-
nisms,Vol.6(1970),Nr.1 ,p.31-57,(page 40)
[3] Hart,H.: On some cases ofparallel motion, Proc.London Math.Soc.8(1877),p.286-289.
[4] Dijksman,E.A.: Kempe's (Focal) Linkage Generalized, particularly in connection with Hart's
second straight-line mechanism,Mechanism and Machine Theory,Vol.l0(1974),Nr.6,p.445-
460,(figure 10)
[5] Kempe,A.B.: On a general method of producing exact rectilinear motion by linkworks,
Proc. Royal Soc.London,23(1875),p.565-77.
[6] Dijksman,E.A.: Overconstrained Linkages to be derived from perspectivity and rejlection,
Proc.71h World Congress on TMM,Sevilla, Spain,Vol.l(l987),p.69-73(figure 10)
[7] Kempe,A.B.: How to draw a straight-line-III,Nature,16(1811)p.l21
[8] Kempe,A.B.: On conjugate Four-piece Linkages,Proceedings of the London Mathematical
Society,9(1878)p.l33-147
[9] Wunderlich,W.: On Burmester's focal mechanism and Hart's straight-line motion, Joumal of
Mechanisms, Vol.3 (1968), Nr.2, p.79-86.
[10] Ruzinov,L.D.: Design of Mechanisms by Geometric Transformations, (translated from the
Russian into English), Iliffe Books Ltd., London,(l968).