Mech. Mach. Theory Vol. 31, No. 7, pp.

891-911, 1996
~ Pergamon 0094-114X(95)00115-8
Copyright© 1996ElsevierScienceLtd
Printed in Great Britain. All rights reserved
0094-114x/96 $15.00+ 0.00

COMPUTERIZED DESIGN A N D GENERATION OF

CYCLOIDAL GEARINGS

F A Y D O R L. L I T V I N and P I N - H A O F E N G
Department of Mechanical Engineering, University of Illinois at Chicago, Chicago, IL 60607, U.S.A.

(Received 9 November 1994; in revised form 7 September 1995)

Abstract--The generation and design of cycloidal (trochoidal) gearings that have found application
in Wankel engine, Root's blowers and pumps are considered. The paper covers the following topics:
(i) generation and geometry of planar cycloidal gearings and rotors of screw Root's blowers, and
(ii) improved design that allows to avoid profile and surface singularities. Computer programs in support
to proposed analytical solutions have been developed. Numerical examples that illustrate the developed
theory are represented. Copyright © 1996 Elsevier Science Ltd

NOMENCLATURE
C~Center of the pin circle
Oi--Center of the centrodes
/--The instantaneous center of rotation
p,--Radius of the pin circle
r~--Radius of the centrodes
a--Distance between the center of the pin circle and the center of the centrode 1
0, 0~--Parameter of the generated curve
~b~--Angle of gear rotation
~--Angle of rotation in the screw motion (Root's blower, Section 8)
p--Screw parameter (Root's blower, Section 8)
~¢2--Curvature of the generated curve
r~~) Position vector of a point in coordinate system S~
m21~ear ratio

1. I N T R O D U C T I O N
Cycloidal gears have initially f o u n d application as watch and clock gears. At present, such gears
are applied in Wankel engine, p u m p s and R o o t ' s blowers (see Figs 1-4).
We consider a planar cycloidal gearing whose one link is provided with pin-teeth [1-3]. The
profiles o f the teeth o f the mating link are curves that are conjugated to the pin-teeth (Figs 1 and 2).
C o n j u g a t i o n o f rotor 1 with c h a m b e r 2 (Fig. 3) is another example of a cycloidal gearing with
pin-teeth applied in Wankel engine [4-7]. In this case the rotor (link 1) performs a planetary motion
in two components: (i) the transfer motion, rotation a b o u t O2 and (ii) the relative motion, rotation
a b o u t Ol (Fig. 3). The c h a m b e r (link 2) is at rest. R o o t ' s blower [1-3] is also an example o f the
discussed type o f cycloidal gearing (Fig. 4). However, we have to consider in this case that the
rotor's t o o t h (the lobe) has an a d d e n d u m (it is the pin), and a d e d e n d u m whose profile is a curve
conjugated with the circular arc o f the a d d e n d u m o f the mating rotor. Figure 4(a) and 4(b) shows
R o o t ' s blowers with two and three lobes, respectively.
In the case o f gearings shown in Figs 1-3 the n u m b e r Nj o f pins o f the driving gear and the
n u m b e r N~ o f "teeth" o f the driven gear must satisfy the relation [1]

N~-N2= 1
Therefore, the gear ratio is
N~ N~
m2! = - - ~ - - -
N: Nl -- 1
In the case o f a R o o t ' s blower we have that N 2 = N1, and m2j = 1.

891
892 Faydor L. Litvin and Pin-Hao Feng

Cenfrode 2 -- ~ ~ Link f
Centrode 2 ~ Link ! k..2
Fig. 1. Planar cycloidalgearing with pin teeth and internal Fig. 2. Planar cycloidalgearing with pin teeth and internal
tangency, tangency.

2 (Chamber)

Centrode 2

" ' ~ 0 2 ' "

,,, ".. ,'

• I

1 Centrode f
Fig. 3. Wankel's chamber and rotor.

Several scientists have performed research in the area of cycloidal gearings. Beard e t al. [5, 8, 9]
have investigated the cycloidal gearing applied in Wankel engine. Litvin [1-3] has investigated the
meshing of Root's blowers and watch gears. Colbourne [10] has considered the geometry of
trochoid envelopes. Sakun [11] has investigated the meshing of screw rotors applied in compressors.
A general approach for the computerized design and generation of planar cycloidal gearings is
proposed in this paper. The contents of this paper cover the following topics:
(1) Determination of equations of cycloidal profiles of gear teeth.
(2) Investigation of the "behavior" of the contact normal (its orientation) during the process
of meshing.
(3) Determination of the curvature of cycloidal profiles, and relations between the design
parameters that correspond to positive and negative curvatures of the profiles, of a stationary point
(when the curvature reaches its extreme value), and conditions that allow to avoid singularities of
the generated profiles.
 Generation and design of cycloidal gearings 893

Centrodes

x ~ ~ ~ Dedendum

2r
(a) I-- =

- ' ~ A ddddendum

(1
2r
(b) =
Fig. 4. Schematicof Root's blowers.

(4) A modified method for generation of chambers of Wankel engines by application of a tool
with an increased diameter has been proposed.
(5) The design and meshing of screw rotors of Root's blower is considered as well. An improved
design for avoidance of surface singularities has been developed.
The developed theory is illustrated with numerical examples and complemented with developed
computer programs.
Special attention has been paid to application of the most effective methods of theory of
gearing [1-3] to the investigation of discussed cycloidal gearings.

2. G E N E R A T I O N OF T R O C H O I D A L (CYCLOIDAL) CURVES
2.1. Appliedcoordinatesystems
We consider a mechanism formed by two movable links and a fixed link, the frame of the
mechanism. Coordinate systems S~, $2 and Sf are rigidly connected to links 1, 2 and f, respectively.
The movable links perform rotation about parallel axes with a constant ratio of angular velocities.
Henceforth, we will consider two cases when the rotation is performed in the same, or opposite
 894 Faydor L. Litvin and Pin-Hao Feng

Yl

(~) (b) Ye k C x2
2

xl
1

Fig. 5. Centrodes.

Fig. 6. For generation of trocoidal curve.

directions, respectively. The centrodes of the movable links are two circles of radii r I and r2 that
are in internal or external tangency as shown in Fig. 5.

2.2. Equation of trochoidal curve

We consider initially the case when the trochoidal curve is generated in coordinate system $2 by
point C that is rigidly connected to coordinate system S~ (Fig. 6). The rotation of links 1 and 2 is
performed in the same direction, the centrodes of links 1 and 2 are circles of radii rt and r2, that are
in internal tangency. Point I is the instantaneous center of rotation. Point C is represented in St as

r~' ~ = [ 0 a 0] T (1)

The trochoidal curve (designated as 0"2) is represented in $2 by the following matrix equation

r2 (¢) = M2~ (q~)r~~ (2)

Here: ~b = $~ is the generalized parameter of motion.

.• b. 2. .
rl
m21 (3)
(~1 r2

where m2t is the gear ratio. In this particular case we have that m2~ = co(2)/co°) is constant.
Matrix

cos(~,,- ¢,2) -- sin(~b I - ~b2) 0 - ( r I - r2) sin ~b2 "]

!
cos(,, - '2) 0 - ( r , -- r2) cos ~2 /
M2,(~b)= l sin(C!- ¢2) (4)
0 1 0 /
0 0 1 J

describes the coordinate transformation from St to $2.

Equations from (2) to (4) represent the generated curve in parametric form, and ¢ is the variable
parameter.

2.3. Curvature of generated curve 0-2

We may determine the curvature of curve 0 2 (it is generated by point C) using the following two
approaches:
 G e n e r a t i o n and design of cycloidal gearings 895

Approach I. The curvature of a parametric curve represented by vector equation r2(~) is

determined by the equation [3]
K2_r~ "n
r~ (5)
where
2r 2
r ~ = 8~----~ (6)

k x r¢// &r2 )
n = 1-~71 ~r~--- ~ , k=-k2 (7)

n is the unit normal to the generated curve; k2 is the unit vector of axis z2 that is perpendicular
to plane (x2, Y2), where curve a2 is located.
The final expression for the curve curvature is
r¢~.(k x r¢)
~c2- Ir~ 13 (8)

After transformations of equation (8), we are able to represent the curvature of the curve by the
following equation
- - ( m 2 1 - - 1))~ 2 - - m21 + ( 2 m 2 1 - - 1)2 c o s ~b
/(2 = F 1 (/'n21 - - 1)(22 - 22. cos ~b + 1) 3/2 (9)
where 2 = a/rl.
The positive (negative) curvature shows that the curvature center is located on the positive
(negative) direction of the normal or the unit normal that is determined with equation (7).
Approach 2. This approach is based on representation of the generated curve a2 in two-parameter
form with an additional relation between the parameters. Curve tr2 is represented in $2 by the
following equations
r2(~b, 0), f(~b, O) = 0 (10)
The advantage of this approach is the possibility to visualize the orientation of the normal to
the generated curve, the "behavior" of the normal in the process for generation, and develop an
alternative approach for derivation of the curve curvature.
The upcoming derivations are based on the following procedure.
Step 1: Normal CI to the curve passes through the instantaneous center of rotation I (Fig. 6).
The unit normal to the curve is represented in coordinate systems S~ and $2 as follows
n~=[sin0 -cos0 0]T (11)

n 2 = L21 (qS)nl (12)

where L:t that represents the direction cosines is the 3 x 3 submatrix of M21, and 0 is measured
from O~ C in the counterclockwise direction (Fig. 6).
Step 2: The procedure of computations of n 2 and the curve curvature requires the derivation of
equation
f(O, ~b) = 0 (13)
that relates parameters (0, ~b). This derivation is based on the consideration that the velocity ¢12)
in relative motion is directed along the tangent to curve ~2, and therefore we have
N . v (~2)= n. v (12) 0 = (14)

where N and n are the normal and the unit normal to a2, respectively. The scalar product in
equation (14) is invariant with respect to the applied coordinate systems, and therefore we can
represent equation (14) as
N i • v~ 12) = n i • v112) = 0 (i = 1, 2 , f ) (15)
896 Faydor L. Litvin and Pin-Hao Feng

The subscript " i " in equation (15) indicates the coordinate system Si where the vectors of the
scalar product are represented. In the following derivations we will choose i --- 1.
The relative velocity v~12) is represented as [1-3]
"" [a-rlnC°S~b 1
v~12)= ((.D~I)- £0~2)) x r I - ( O 1 0 2 ) x £0~2) = (.ol(m21 - 1) r I s 49 (16)

where rn21 = o9~2)/¢oo) is the ratio of angular velocities.

Equations (1 1), (15), and (16) yield that
f(O, 49) = ri sin(0 + ~b) - a sin 0 = 0 (17)
Step 3: Curve 0"2 and the unit normal to the curve are represented in coordinate system $2 by the
following equations

r2(~b) = m2,(49)r~ c), n2(0 , 49) = L21(~b)nl(O), f(O, ~b) = 0 (18)

Equations (18) represent in coordinate system $2 the trochoidal curve generated by point C
and the unit normal to the curve by two related parameters 0 and 49.
Similarly, we can represent the generated curve a2 when the rotation of links 1 and 2 is performed
in opposite directions.
Step 4: The investigation of equation (17) shows that the relation between parameters (0, 49)
depends on the ratio a/r I.
Differentiating equation (17), we obtain that
dO cos(0 + qg)
d49 - 2 cos 0 - cos(0 + q~) (19)
where 2 = a/rl.
Using equations (1 7) and (1 9), we obtain after transformations the following alternative equation

dO ). cos 49 - 1
d--~ - 2 2 - 22 cos 49 + 1 (20)

1.5708'

0 0 4 1 r.. - ~m~z.,

-!.5708
0 3.1416 6.2832

Fig. 7. Functions O(~)for various coefficients 2 = a/p:O--,~ = 1 , ~ - - , 2 = 2, C)---*2 = 3, p/r = 0.1333.

 Generation and design of cycloidal gearings 897

0"

0 -3.1416 ~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

-6,2832
0 3.1416 6.2832
(p

Fig. 8. Functions 0(~b) for 2 =0.8, p/r =0.1333.

At the position determined as

1
cos ~b = [sin 0 [ = ~- (2 > 1) (21)

we obtain that angle 0 will reach its maximal value, and the speed of the normal C-] with respect
to position vector O~ C becomes equal to zero.
Figures 7 and 8 show the relation between 0 and q~ for the cases when a/q >1 1, and a/q < 1,
respectively. Figure 9 shows the two branches of the curve that is generated by point C when a = q.
The common point of the branches is singular.

i ",, ,~ 0 ~ ," "'"ii

•" r'
,i

-- ...... ..o'°

Fig. 9, Two branches of trochoidal curve with singularity point.

898 Faydor L. Litvin and Pin-Hao Feng

Step 5: The determination o f the curvature o f the generated curve represented in two-parameter
f o r m is based on the equation [1-3]
~2 v~2) = - fi~2) (22)

where v~2) is the velocity o f the contact point that moves over a2, and fi~2)is the velocity o f the tip
o f the unit n o r m a l (in addition to the translational m o t i o n o f n ~z)) that moves over a2. In the
discussed case v~2) - ¢12) that is represented in S~ by equation (16). Vector ,,r-~2)is represented in S~
by the equation [1-3]

"rl~(2)= ,lrllll(l)+ to] 12) x n =

[dO+ (1 - m21 )
] [cos 0 sin 0 01T (23)

After transformations, we obtain the following equations for determination o f K2.

rl sin(0 + ¢ ) - a sin 0 = 0

[(1 + m21)a cos 0 - rim21 cos(0 + ¢ ) ] cos 0

1£2
(m21 -- 1)(a - rl cos ~b)[rl cos(0 + ¢ ) -- a cos 0]
[(1 + m21)a cos 0 -- rim21 cos(0 + q~)] sin 0
(24)
(m21 - - 1)rl sin ~b[rl cos(0 + ~ ) -- a cos 0]

Using the relations

sin ~b
sin 0 = (25)
(2 2 - 22 cos ~b + 1) I/2

2 - cos ¢
cos 0 = (2 2 _ 22 cos q5 + 1) I/2 (26)

we obtain after transformations that equations (24) yield equation (9) for x2.

3. D E T E R M I N A T I O N OF C U R V E ~2 C O N J U G A T E D TO P I N - T O O T H
The applied coordinate systems (Fig. 10) are the same as shown in Fig. 6. The pin-tooth is a
circle o f radius p. The goal is to determine the curve that is conjugated to the pin-tooth.
Cycloidal gears shown in Figs 1 and 2 are applied in planetary trains when a high gear ratio
o f the train is required. This requirement can be observed if the planetary train is designed as a

Yt l x2

Fig. 10. For generation of a curve conjugated to a pin-tooth.

 Generation and design of cycloidal gearings 899

set of at least two pairs of cycloidal gears shown in Figs 1 and 2 and because the difference of
number of teeth of the gears is equal to one. Theoretically, all pins of one gear in such a train can
be in tangency with the teeth of the mating gear, but under the load only half of the teeth share
the load. The pin teeth enter into mesh in sequence. The determination of the trochoidal curve as
conjugate to the pin-tooth circle must be determined in the range of ~b = 2~Z/Nl, where N~ is the
number of pins, ~ is the angle of rotation of the gear provided with pins.
The profile of the chamber of Wankel engine (Fig. 3) is a trochoidal curve that must be
determined in the range ~b = 2rrm, where m is a whole value that depends on the ratio m2~.
The determination of the trochoidal curve is based on the following procedure:
Step 1: We represent in parametric form the equation of the pin circle (designated by oj ) by vector
equation rl (0).
Step 2: We derive the equation of meshing that provides the relation between parameters 0 and ~b.
Step 3: Using the coordinate transformation from S~ to $2, we represent in $2 the family of pin
circles. Considering simultaneously the family of pin circles and the equation of meshing, we may
determine the curve that is conjugate to the pin circle.
The following is the detailed explanation of the mentioned above procedure.

3.1. Parametric representation of pin circle ~

Using the drawings of Fig. 10, we represent the pin circle and its normal by the equations

rl(0)=[psin0 a-pcos0 0 1]v (27)

n l(0)=[sin0 --cos0 0]T (28)

Angle 0, if positive, must be measured as shown in Fig. 10.

3.2. Representation of family of S,1 in $2

Using the coordinate transformation from S~ to $2, we obtain

r2(0, ~b) = M2~(~b)rl (0) (29)

where M21(~b) is represented by equation (4).

3.3. Equation of meshing

There are three alternative approaches for determination of equation of meshing.
(1) Approach 1 (applied in differential geometry) that is based on the equation

0F 2 0r 2
x ~-~ = 0 (30)

(2) Approach 2 (applied in theory of gearing) that is represented as

n I • V~ 12) = N l " v~12) = 0 (131)

(3) Approach 3 (based on Lewis' theorem). We use in this approach the theorem that the normal
to ~ at the point of tangency of E~ and Y~2 passes through the instantaneous center of rotation I.
Thus, we have
X,(49)-Xl(O) r,(c~)-y,(O)
= (32)
nx, (0) ny, (0)
Here X~(~b) and Y,(~) represent in S~ the coordinates of point I, the instantaneous center of
rotation.
Using any of the mentioned above approaches, we obtain that the equation of meshing coincides
with the previously derived equation (17). Thus, we have

f ( O , d?) = r I s i n ( 0 + ~b) - a sin 0 = 0

900 Faydor L. Litvin and Pin-Hao Feng

Yl

02

~ I Centrodes
gQ
/

Fig. 11. Relation between parameters 0~ and 02 for two Fig. 12. Two positions of the pin with the same value of
envelopes. parameter 0.

It is easy to verify that equation of meshing provides two solutions for 0 considering as given ¢.
Using this equation, we obtain that

sin4~ (2 r~) (33)

tan 0 = ). _ cos q~ =

Equation (33) provides two solutions for 0 that differ for 180 °. This means that two envelopes
exist at any instant (excluding the case when a = r~ and ¢ = 0, see below). The geometric inter-
pretation of this statement is illustrated with Fig. 11. The drawings show that when the normal
to the pin passes through the instantaneous center of r o t a t i o n / , the pin may have two contact
points with Z2:M1 and M 2.
The relation between parameters 0 and ¢ of equation of meshing (17) depends on the ratio a/rt
(see Figs 7 and 8). In the case of a/r~ > 1, the working part of the pin circle is an arc, and during
the process of meshing the contact points perform a reciprocative motion over the circular arc. At
the position determined with Isin 01 = cos ¢ = 1/2 [see equation (21)] the velocity of the contact
point in its motion over the pin circle is zero.
It is important to recognize that in the case of 2 > 1, two values of ¢ correspond to the same
value of 0 (with an exception when 0 = 0max). This means that there are two positions of the driving
link when the same pin point generates in sequence two respective points of the conjugate profile
as shown in Fig. 12. Drawings of this figure yield that

~(2) + ~(1) = 180 ° _ 20

where ~b~2) and 4~~) designate the related positions of the driving link.

3.4. Equations of the envelope

The envelope E 2 to the family of pin circles is determined with equations (29) and (17) considered
simultaneously.

4. C U R V A T U R E OF C U R V E ~2
The curvature of c u r v e ~~2 may be determined by application of two approaches: (i) as the
curvature of the curve that is equidistant to curve a2 (see Section 2), and (ii) using the direct
 G e n e r a t i o n and design of cycloidal gearings 901

relations between the curvatures of the pin circle Y~ and the generated curve E 2 [1-3]. The second
approach requires the application of the following three relations [1-3]:
fi(2)=
r fi~l) o 02) x n (1) (34)
v~2) = v~~)+ v(~2) (35)

fo~t + f~ ~t =0 (36)

Here:
dO
Irl~1) = ~ [COS 0 sin 0 0] (37)

dO
V~1) = p ~-~ [COS 0 sin 0 0] 1138)

fo~t+f~t =[acosO-rlcos(O+49)]~t-r, cos(O+(9)~t =0 1139)

The results of investigation are as follows:

(1) Applying the first approach and using equation (9), we obtain that the curvature x2 of curve
E2 is represented by the equation
1 rl (m21 -- 1) (22 _ 22 cos ~b + 1)3/2
K2 -- (m21 - - 1)22 _ rn21 + (2m21 _ 1)2 C o s ~b + p (140)

where the upper and lower signs correspond to the internal and external tangency of the pin circle
with the envelope. It is assumed that the centrodes are in internal tangency [Fig. 5(a)].
(2) An inflection point of curve E2 = 0 means such a point of the curve where x2 = 0. The analysis
of equation (40) is based on following considerations:
(i) 1/K2~oo if
--(m21 -- 1)22 - m21 -+- (2m21- 1)2 cos q5 = 0 (41)
(ii) Equation (41) yields that
m21 + (m21- 1)2 2
cos q~ = (42)
(2m21- 1)2
Here ~b determines the position of link 1 when an inflection point of E 2 (if such a point exists) is
generated; 2 = a/rl.
(iii) It follows from equation (42) that an inflection point exists indeed if
m21 + (m21 - - 1)2 2
1 (43)
(2m21- 1)2
Then we obtain the following requirement for 2
m21
1 ~< 2 ~ < - (44)
mzl - - 1

(iv) It is obvious that if

m21
2 > - (45)
rn21- 1
curve Z2 is a convex one. Curve E2 is a convex-concave one if inequality (44) is satisfied, and curve
Y~2 has an inflection point.
(3) A stationary point of curve E2 means such a point where the curvature radius is of an extreme
value [12] and

d--~ = 0 (46)
902 Faydor L. Litvin and Pin-Hao Feng

Equations (46) and (40) yield that a stationary point exists at positions where

sin 05[(2 - rnzl)2 2 + (2mzl - 1)~, cos 05 - (mzl + 1)] = 0 (4?)

It follows from equation (47) that stationary points of curve E2 exist at positions determined by

sin 05 = 0 (48)
and
(m~l - 2)2 2 + m21 + 1
cos 05 - (49)
(2m21- 1)2

The curvature radius at the positions 05 = 0 and 05 = 180 ° is determined by the following
equations
1 r l ( m 2 1 - 1)(2 T 1)2 _ _

- = + p (50)
~c2 +_me1 - (mzl - 1)2

The upper and lower signs of the fraction part in equation (50) correspond to positions 05 = 0
and 05 = 180 °, respectively. In the case when curve a 2 is considered (it is generated by point C,
see Fig. 6), we have to take in equations (50) that p = 0.
Equation (49) provides the third stationary point if 05 is changed in the range 0 ~< 05 ~< n.
The curvature radius of E2 at the stationary position is determined with the equation

1 (m21 -- 1)~/27(a 2 -- r~) -T-p (51)

~2 ( 2 m z l - 1)3/2

5. SINGULARITIES OF ~2
The normal to the generated c u r v e ~~2 at a singular point is indefinite, and E z may have at such
a point two branches. Curve E2 with singular points is unacceptable for application, such points
must be avoided, and this can be achieved by proper relations between the design parameters.
A singular point on E2 will occur if the velocity v~2) of the contact point in its motion over E2
becomes equal to zero [1-3]. Using the relations between the velocities of the contact point in its
motions over E~ and E2, we will obtain the following relation [1-3]

v~z) = v~1) + v(lz) = 0 (52)

The procedure of derivation of singularities is as follows:

Step 1: We represent equation (52) in coordinate system S l

dx_2 dO ,,(12) @1 dO -- ,,(12) (53)

dO dt = - t / x l ' dO dt ~Y~
Step 2: Differentiating equation of meshing (17), we obtain

Of dO Of d05
(54)
00 dt 005 dt
Step 3: Equations (53) and (54) represent a system of three linear equations in one unknown dO/dt.
The augmented matrix is

0x,
00 -v~?~]
Oyl
00
_,,(~2) I
~"yl [
(55)
Of Of O)
oo _l
 G e n e r a t i o n a n d design o f cycloidal gearings 903

and a solution for dO/dt exists if the rank o f matrix (55) is one. This requirement yields

dO
( a - - p COS 0 - - r I COS t ~ ) ( m 2 1 - - 1) + p cos 0 ~ = 0 (56)

Equations (56) and (17), if considered simultaneously, enable to determine the limiting value of
p to avoid singularities on 2;2.
Step 4: The closed form solution for the limiting value of p (considering as known m21, rj and a),
when singular points occur, is based on following considerations:
(i) We have obtained in Sections 2 and 3 that

sin~b (). : ~ )
tan 0 = ). _ cos q5 (57)

and
dO (). cos ~b -- 1) cos 2 0 ). cos q6 - 1
(58)
dO (). - cos ~b)2 ).2 _ 2). cos gb + 1

(ii) We transform equation (56) as follows

p 2 = r~(). -- cos q~)2(m2, -- 1)2

os 0(d + 1 - - m21
(59)

Then, using equations (57), (58) and (59), we obtain

p2 r~(rn2, -- 1)2(c -- d cos q~)3

(60)
(e + f c o s ~))2
where
c = ) . 2 + 1; d=2).; e =).2(1 --m21)--m21; f=(2m21--1)2 (61)

1800 -

1500

1200

p2 900

600

300

/
• I " I " I

0 3.1416 6.2832 9.4248 12,5664

¢p

Fig. 13. For derivation of singularities: 2 = 0.8.

904 Faydor L. Litvin and Pin-Hao Feng

!,200,000 ]

900,000 -

p2
600,000

300,000

k._
"3.1416 612832 . . . . . . 9.4½48 12.5664

Fig. 14. For derivation of singularities: 2 = 1.5.

(iii) Functions p2(~b) for various coefficients of 2 are shown in Figs 13 and 14. The extreme values
o f this function we m a y obtain considering that
dp _ r~(rn2t - 1) 2 sin q~(c - d cos ~b)2(3de + 2 c f + d f c o s q~) = 0 (62)
d$ 2 p ( e + f c o s ~b)3
T h e extreme values occur if: (i) sin ~b = 0, (ii) c - d cos ~b = 0, or (iii) 3de + 2 c f + d f c o s ~b = 0.
Using the drawings o f Figs 13 and 14, we obtain that the m i n i m u m extreme value of p occurs when
~b = 2nn (n is an integral). Therefore, the limiting value o f p [see equation (61)] can be determined by
p2 = r~(m21 - l)2(a - b) 3 (63)
(c + d) 2
E x a m p l e s o f singularities of the generated curve are shown in Fig. 15. T o avoid the singularity
occurred on the generated curve, we have to choose p2 larger than provided by equation (63).

(b)

Fig. 15. Examples of singularity of Wankel's chamber. (a) 2 =1.454, p/r =0.1333, (b) 2 = 1.454,
p/r = 0.1333.
 Generation and design of cycloidal gearings 905

Fig. 16. Example of swept out space: internal tangency of Fig. 17. Examples of swept out space: internal tangency of
pin and generated curve, pin and generated curve.

6. S P A C E S W E P T O U T BY G E N E R A T E D S H A P E
We consider in this section that profile E2 of the generated shape has been determined and
represented in coordinate system $2. The goal is to determine space El that is swept out in Sj by
curve E2. The solution to this problem is based on the following considerations:
(1) We set up the same coordinate systems SI, $2, and Sf that have been applied in Section 3 for
the generation of E2. It is assumed that links 1 and 2 perform rotation about O~ and 02 with the
same ratio m2~. Thus, there will be the same centrodes and the instantaneous center of rotation
as discussed above.

,i~::.o............ ~.~.
I o ! ',,

Fig. 18. Examples of connection of pins.

906 Faydor L. Litvin and Pin-Hao Feng

(2) Curve Y~2generates in Sl a family of curves as shown in Figs 16 and 17. Using an approach
similar to that discussed above, we are able to determine the envelope E~ to the family of curves
Y,: represented in Si. Drawings in Fig. 18 show the envelopes E~ for two cases, when profile Y:
is generated by pin E1 being in internal and external tangency with E2, respectively. The envelope
consists of two branches that determine the borders of the swept out space. One of the branches
of the respective couple of curves Y"l can be used for connections of the pins as shown in Fig. 18.
It is not excluded the appearance of singular points on one of the branches of E~ as shown in Figs 16
and 17.
The analytical determination of envelope E~ to the family of curves E2 is based on the following
procedure:

Step 1: We represent curve Z2 in $2 by the equations (see Section 3)

r E r2(0 , 49), f(O, 49) = sin(0 + 49) -- 2 sin 0 ----0

= (64)
Step 2: The family of curves E 2 is represented in Sl by the equations

rl(0, 49, qJ) = Mlz(~,)r2(0, 49), f(O, 49) = 0 (65)

Here: ~O2 and ~kl represent the angles of rotation of E2 and El about centers 02 and Oi, where

4,2 = q,, m2,

Matrix Mt2(q/) is similar to matrix M12(49)= Mill(49) [see equation (4) for Mzl(49)].

Step 3: To determine the envelope E'l to the family of curves E2, we need the equation of meshing
that can be derived as follows
X2(~b ) - x2(O, 49) Y2(O ) - y2(O, 49)
- (66)
nx2(0, 49) ny2(0, 49)
Here: X2 (~O) and Y2(~k) represent the coordinates of instantaneous center of rotation I in coordinate
system $2.
Equations (66) yield the equation of meshing between E2 and El in the form

F(O, 49, qJ) = sin(0 + 491 - 492''[- I]/2) sin(0 + 49~) = 0

- - (67)

We take into account that equation (67) yields

(i) I]/2 -- 492 = 2zm (68)

(ii) sin(0 + 49~ - 49z + ~b2) = sin[(Zn + 1)z~ - 0 - 49,] (69)

where n = 0, 1, 2 . . . .
It results from equations (68) and (69) that equation of meshing (67) yields two solutions for ~2

(i) ~2 = 492+ 2~n (70)

(ii) qJ2 = 492+ (2n + 1)zt - 2(0 + 49l) (71)

Equations (70) and (71) correspond to two branches of the envelope E'l, respectively.

Step 4: Equations (65) and (67) represent the sought-for envelope E'l.

7. GENERATION OF CHAMBER WITH TOOL OF INCREASED DIAMETER

Our goal is to generate the chamber by a tool whose diameter is increased with respect to the pin.
Figure 19 shows circles of radii Pl and P2 that represent the original pin and the applied tool,
respectively. The circles are in tangency at point M, that passes through centers Cl and C2 of the
circles, and the instantaneous center of rotation I. Thus, M is the current generating point of the
pin and the tool.
 Generation and design of cycloidal gearings 907

x2

I
X!

Fig. 19. For generation of c h a m b e r with enlarged tool diameter.

The imaginary process of the generation of the chamber is based on the following consideration:
(1) We assume that coordinate system Sl with the original pin performs rotation about O~ while
coordinate system $2 with the chamber performs rotation about 02. The positions of coordinate
systems Sj and $2 are determined with angles q~ and ~b2 related as follows

/)2 rl
- (~b~ - q~) (72)
{J~l r2

The center distance E~2 is determined as Et2 = r~ - r2.

(2) Angle 0 that determines the orientation of the contact normal MI with respect to O~C~ is
determined with equations (25) and (26).
(3) Drawings of Fig. 19 yield that position vector O~C2 is determined in & as

x{ c2) = (P2 - P~ ) sin 0, y{C2)= a - (P2 - Pl ) COS 0 (73)

(4) Using coordinate transformation from S~ to $2, we obtain the coordinates of center C2 in
coordinate system $2
r~c2) = M21 (q~)r]C2)(0) (74)

where 0 and q~ are related with equation (17). Functions x~C2)(q~) and y~C2)(q~) represent in $2
the current position of the center of the generating circle of radius P2.
The limitation of the described method of generation is that the radius of the increased generating
pin cannot be larger than the smallest radius of the curvature of the generated curve. In the case
of internal tangency, the limitation of the tool radius can be obtained from equation (50) by
considering ~b = 180 ° as
rt(m21- 1)(2 + 1)2
Plimit -- q- p (75)
m21 + (m21 -- 1)2
Similarly, in the case of external tangency, the limitation of the tool radius can be obtained from
equation (50) by considering ~b = 0 ° as

rl (m21- 1)(2 - 1)2

Plimit = -- p (76)
m2~ -- (m21 -- 1)2
908 Faydor L. Litvin and Pin-Hao Feng

(a) (b)
Fig. 20. Root's blower.

8. R O O T ' S BLOWER
8.1. Introduction
The Root's blower is designed as a mechanism with two rotors that perform rotation about
parallel axes (Fig. 20). The rotors are provided with two lobes [Fig. 20(a)] or with three screw lobes
[Fig. 20(b)]. In the last case the surfaces of rotors are helicoids. The main attention in this section
is paid to the design of three-lobed rotors.
The axodes of the rotors are two cylinders of equal radius r that are in external tangency. The
cross-section of three-lobe rotors is represented in Fig. 4(b). The addendum of the profile is a circle
of radius p. the dedendum is a curve that is conjugated to the addendum.

(
Y2, Yb

0~

~X2,X~
Xl,Xa

~xj
o, oI
Fig. 21. Coordinate systems applied for Root's blowers.
 Generation and design of cycloidal gearings 909

8.2. Applied coordinate systems

Movable coordinate systems, S~ and $2, are rigidly connected to rotors 1 and 2 (Fig. 21).
The fixed coordinate system Sr is rigidly connected to the frame of the blower.

8.3. Addendum and dedendum surfaces

Initially we consider the meshing of rotors in plane z~ = z2 = 0, as meshing of planar curves. The
cross-section a, of the addendum surface for a three-lobe rotor is represented in an auxiliary
coordinate system Sa by the following equations

Xa = p sin 0, Ya = a + p COS0 (77)

(a +p)2--rax//~ar~<tan_02~<(a
- +~pa)r2 - - r 2 (78)

The conjugated curve cr2 will be represented in coordinate system Sb that performs rotational
motion about 02. The rotation of coordinate systems Sa and Sb is considered as similar to rotation
of S~ and $2 (Fig. 21).

8.4. Equation of meshing

Using the approach similar to that applied in Section 3, we obtain that the equation of meshing
of a~ and a2 is represented as

f(O, 49) = r sin(0 - 49) - a sin 0 = 0 (79)

Using the coordinate transformation from S, to Sb, we obtain that curve a2 is represented in
Sb by the equations
X b = p sin(0 - 249 ) - a sin 249 + 2r sin 49

Yb = P COS(0 - - 249) + a cos 249 - 2r cos 49

r sin(0 - 49) - a sin 0 = 0 (80)

The addendum surface Z~ of the rotor 1 is a helicoid that is generated by the screw motion of
coordinate system Sa with respect to S~. A right-hand helicoid of the addendum is represented by
the equations
xt(O, Ip) = p sin(0 - ~ k ) - a sin~k
y, (0, ~O) = p cos(0 - ~b) + a cos ~b

z, (q~) =pC, (81)

Surface Z2 of the dedendum of rotor 2 is a helicoid whose direction is opposite to the direction
of the helicoid of rotor 1. Surface Z 2 is generated by the screw motion of curve tr2 about the axis
of rotor 2, and is represented in $2 as follows

r2(0, @, 49) = M2b(ff)r2(0, 49), f(O, 49) = 0 (82)

Equations (82) represent Y'2 in the three-parameter form but parameters (0, 49) are related.

8.5. Equation of meshing between Si and ~2

We obtain the equation of meshing as

N~. v~~2)=f(O, ~, 49) = 0 (83)

where N~ is the normal to surface Y.~, v~~2) is the relative (sliding) velocity. Vectors N~ and v~j2) are
represented in S, by the following equations

Or, &rl [ppsin(O-q,)]

N 1= ~ x ~ = ? p cos(0 -- ~k) / (84)
L pa sin 0 _]
MMT 3 1 1 7 ~
910 Faydor L. Litvin and Pin-Hao Feng

\
Fig. 22. Contact lines on Root's blower surface.

Table 1. Properties of Z2
Lobe number Convex Concave-convex With singularities
2 0 < 2 ~<0.5 0.5 < 2 < 0.928822 2 >~0.928822
3 0 < 2 ~<0.5 0.5 < 2 < 0.96693 2/> 0.96693

--2co[p cos(0 - ~b) + a cos ~b - r cos q~]]

V~12) = 2c0[p sin(0 - ~ ) - a sin ~b - r sin q~] [ (85)
0 J
After derivations, we obtain
f ( O , ~b, (a) = r s i n ( 0 - ~b - qS) - a s i n 0 = 0 (86)

Equations (82), if considered simultaneously, allow to obtain the instantaneous lines o f contact
between Zt and ~"~2 on surface Yl as shown in Fig. 22.

8.6. Properties o f az
The planar curve a2 m a y be a convex curve, c o n c a v e - c o n v e x one, with or without singularities
depending on the ratio o f 2 = air. The results of investigation are represented in Table 1.

9. C O N C L U S I O N
The m o d e r n theory o f gearing for the design o f trochoidal gearings has been applied.
(1) Equations o f trochoidal and equidistant curves in parametric form have been derived
(Sections 2 and 3).
(2) Curvature o f trochoidal curves and inflection and stationary points o f such curves have been
derived (Section 4).
(3) Conditions for avoidance o f singularities o f generated trochoidal curves have been deter-
mined (Section 5).
(4) Space swept out by the generated curve has been determined (Section 6).
(5) Generation o f the c h a m b e r o f Wankel's engine by application o f a tool with an increased
diameter has been developed (Section 7).
(6) Design o f R o o t ' s blower with screw rotors has been developed (Section 8).

Acknowledgement--The authors are grateful to the Gleason Memorial Fund for the financial support of this research.

REFERENCES
1. F. L. Litvin, The Theory of Gearing, 2nd Edition. Nauka, Moscow (1968).
2. F. L. Litvin, Theory of Gearing. NASA RP-1212 (AVSCOM 88-C-035), Washington, D.C. (1989).
3. F. L. Litvin, Gear Geometry and Applied Theory. Prentice Hall, Englewood Cliffs, NJ (1994).
 Generation and design of cycloidal gearings 911

4. R. F. Ansdale, with a special contribution by D. J. Lockley, The Wankel RC Engine. Barnes and Co., Cranbury, NJ
(1969).
5. J. E. Beard, D. W. Yannitell and G. R. Pennock, Mech. Mach. Theory 27, 373 (1992).
6. F. Wankel, Rotary Piston Machines. Iliffe Books, London (1963).
7. K. Yamamoto, Rotary Engine. Toyo Kogyo, Tokyo (1981).
8. J. E. Beard, A. S. Hall and W. Soedel, 13th Annual A S M E Design Autom Conf., Vol. 10-2, pp. 355 (1987).
9. J. E. Beard, G. R. Pennock and M. M. Stanisic, Trans. SAE, J. Comm. Vehicles, Sec. 2, Vol. 98, pp. 217 (1989).
10. J. R. Colbourne, Mech. Mach. Theory 9, 421 (1974).
11. I. A. Sakun, Screw Compressors. Mashinostrojenie, Leningrad (1970) (in Russian).
12. E. A. Dijksman, Motion Geometry of Mechanisms. Cambridge Univ. Press, Cambridge (1976).

PE3IOME--PaccMaTprmaeTca o6pa3oBanne aarnonaaabm.lx (TpoxorI/Ia~bablx) 3atterLqe.nfi,

Hatue~tunx np.Meaen.e a JaBaraTe0aax BanKe0as (Wankel), Boa~ayxoayaKax PyTa (Roots) n nacocax.
CoaepxaH.e CTaTb. nocaameno: (a) o6paaoBannio a reoMeTpnn naocKnx unK~oaaabHbIX 3aaenaenn~,
BaaTOablX poTopoB UOa~tyxoayBoK Pyva, (6) ycoBeptueHcTaoBannro npoeKTnpoBanna, noaBoaaloxuero
H36e)KaTb Oco6eHHOCTe~ o6paayeMbIX rpnBblX 14 noBepx.ocTe~. Paapa6oTanbi roMnbloxepnbie
npoFpaMMbl ~Jl~l qHc.qeHHblX pactt~TOB B ~OrlOJlnenlle K npe/l~ox~eHnbiM ana.~nTnqecKtlM pe~ttertHSM.
lqpnBo21nTC~ ~iHc.rtenHbie npHMepbl /LT1~rI.~.~lOcTpalll~IH npe~a~oxe.Hb~X MeTO,~OB.