Professional Documents
Culture Documents
Matthew
Research Associate.
D. Tesar
Prolessor.
University of Florida,
Gainesville, Florida
Introduction
Recent years have seen a substantial increase in the use of materials for cams which have heretofore been considered too expensive or too difficult to machine. Manufacturing of cams has improved by an order of magnitude (now better than 0.0001 in.)
during the past two decades. At the same time, ever higher operating speeds have been imposed upon practically all machinery.
As materials and innovative construction progress, the usefulness
of the cam system will continue to grow as a variable complex
motion mechanism in the industrial process. While the physical
capabilities of cam system components improve, so must the
designer's understanding of the governing dynamic theory in
order that such materials and construction be exploited to their
fullest.
The primary goal of the companion paper (Part I) was the dynamic synthesis of the required cam surface function in terms of
an assumed set of model parameters and a fixed design speed (see
also, reference [l]). 1 Should any of these parameters be in error,
the wrong cam surface will have been calculated, and the resulting output motion will be a distortion of the desired specified motion. This will induce unwanted vibrations and reduce overall
precision of operation for the system, both of which should be
kept to a desirable minimum. The goal of this paper is to derive
direct, easily used means of approximating the magnitude of this
distortion and provide the designer with clear suggestions to min-
1
Numbers in brackets designate References at end of paper.
Contributed by the MechanismsDED Division of THE AMERICAN
SOCIETY OF MECHANICAL ENGINEERS and presented at the Mechanisms Conference, New York, October 6-10, 1974. Manuscript received at
ASME Headquarters, June 18,1974. Paper No. 74-DET-44.
imize motion distortion and reduce the rate of growth of wear and
potential for noisy operation. The results given here make it possible to avoid tedious, complex, and somewhat unreliable analysis
procedures dependent on coupled differential equations.
Many early works made excellent formulations for cam synthesis and analysis [2-5]. These works were exemplary in their coverage and especially in their experimental validation. Various other
researchers [6-16] have investigated segments of the cam design
problem. Johnson [7, 8] showed how to use finite differences to
design the cam shape function. The idea of a nonconstant angular
velocity for the cam was treated in references [9, 10, 12]. Generally, the objective has been to minimize internal vibrations or unusual variations of the cam surface force. The concisely posed
problem [16] of Freudenstein treats the concept of amplitude amplification due to a cyclic reinforcement from the cam. An attempt to treat general cyclic reinforcement due to a repeated impact from a gross manufacturing error was given in reference [18].
Very recent activity by Chen [19-21] has been significant in enhancing the quality of numerical computation. The interesting
question of clearances and associated impact was studied in reference [22].
Rarely is dynamic synthesis used to design a cam. This means
that the design speed u>d of such a cam is zero; which assures that
the system always operates "off-speed" and will generate a distorted output motion relative to the specified motion. Cyclic
speed variations due to speed drift or variation of a range of speed
-as in operation of the internal combustion engine make the operating speed cc distinct from the design speed. Frequently, machinery rates of production are changed daily or increased to meet
output requirements, all implying off-speed operation.
The procedures for dynamic synthesis (Part I) immediately
infer the existence of a model. Since the modeling process involves geometry, mass content, deformation rates, extensive computation, and numerous assumptions; there always exists a po-
NOVEMBER 1975
1181
W V +^wx'
k + hr
(2)
where the analysis is established in terms of the actual system
parameters (m, k, cr, kr, w or /Zm, jic, jik) which, differ from the
modeled parameters. Here y represents the known specified output motion, x represents the actual response of the real system,
and S represents the shape factor synthesized for the modeled
system. The response condition, equation (2), shows that S acts
as the driving function for analysis of the real system.
The above equations may be combined to give
JImx"
+ Mc*' + Hx
= l1^"
+ ^ '
V-*y
(3)
It is convenient to examine the output response in terms of its deviation & = y x in equation (3) to yield
(5)
or
(J-M = (1 + am)jim,
(i c = (1 + a c K , iik = (1 + ak)nk
(6)
2.
s=.v
>*
mV-nS>"
+ cP-<y' + ak^y
(1)
-1
K +Kr
+ (1 + a > c 5 ' + (1 +
(7)
1182
NOVEMBER 1975
p,q,
ft (i = 0,1,2,3,4)
actual coefficients of
the response motion for analysis in
a two degree of
freedom system
= modeled dimension-
v,\y
less
distribution
parameters for two
degrees of freedom
An, A r 2
f\, X, 7 = actual
dimensionless
distribution
parameters for two
degrees of freedom
i, /^> Py ~ model error parameters for two degrees of freedom
d = output motion distortion
ad, w = assumed and actual
operating speeds,
respectively
v, r = real and imaginary
parts, respectively, of the assumed
solution of a response equation
be
1.0
"-I.6
/
.8
j34p4
= (1 - 2 J J ) ( 1
/1
-"^06 y
-X)77X/3/J.
/^02
IBX=O
P4
= (1 -2x)(i -
(31)
v)vxfixiiJ
v,r
l=o
A/
V*
0.0
tin
(32)
0.0
0.0
ft,
.4
.6
=JJ\(1-\)(1-27J)
Fig. 3
(33)
/?,, or /9 2T
/34;i = 7 ? X ( l - j 7 ) ( l - 2 A )
are functions of the modeled distribution parameters only. Thus,
if the designer has some estimate of the probable magnitude of
errors in the mass deformation distribution (fSv,fS\), the @tn,f3t\
functions will indicate the sensitivity of the p 4 synthesis coefficient to these errors. The error distribution coefficients have been
plotted in normalized contour charts (Figs. 3 and 4) with a normalization factor of 4 in each case. Note that equation (32) is an
approximation which allows the qualitative representation of the
charts. For complete accuracy, the designer should use equation
(28).
The above formulation has been carried out for all the ft error
coefficients and is listed in Table 1. Also, all the /3( Pa error distribution coefficients have been plotted in Figs. 3-8. These show
that to minimize the system's sensitivity to distribution error, the
mass and deformation should be near the cam (?) = 0, X = 0).
The condition for best dynamic synthesis (?) = 0, X = 1) shows increased sensitivity due to |8 3 ,\(Y = 1), /?2,\, /?i,\- The significance of
this contradiction will depend on the system and the magnitude
0.0
0 0
-0.2/
i,r
i"l
V4
( .0
0.0
.4
.6
.8
X
/?4x or 0'2\
NF
B. = n X ( l - X ) ( l - 2 n )
4n
B 4A - n X ( l - n ) ( l - 2 X )
B, - nX(l-x)
3n
,
.
UXU-2A), Y-0
B3
X " ^ ( l - n ) X (1-2X), Y - 1
Approx. Form
Remarks
{(1+8
( 8 . B + 6.,B,})J 2
4n n
4X.A m
)n (1+8,)X ( l - x - B X ) ( l - n - P n)
A
A
n
-r,A(l-n)U-X)>i
m
B3P3
08 -iq
Exact Form
+0.125
(PEAK)
Fig. 4
Table 1
. ,
1 ~0AI
( f ( l + B n ) n ( 1 - Y - B Y) + (1+B )Y ( l - n - B n n ) ]
[(1+BX)X (1-X-B X X)] -
{f
Vn
+ 8
3X X>V m
In(l-Y) + Y ( l - n > ] [ A ( l - X ) ] ) p c i i n
62p2
{(1+B
)n + ( l - n - 6 n ) ( l - X - B , X ) n
n
X
n - (l-n)(l-x)}>i +
B2X -
YX(1-Y)(1-2X)
B 2 Y - YX(1-X)(1-2 Y )
B
{8
2n 8 n
2AV>V
ID
{(1+8
Vl
Y(1-Y)
{(1+B
-
)Y (1+0!)X ( l - Y - 6 YX1-A-B.X)
A
A
Y
Y -
2YSY
Wl
XU-A)}^
)Y + ( H - J v l ) ( l - H , l )
Y
A
{8
1Y 6 Y
+ 8
lX8Xll,c
lY "
U-Y)U-*)}UC
Vo
+ {6
-X(1-Y)
YX
NF = Normalization Factor
g
n
BZH
ft
. X~X
n *
X
X '
1 -1=1
Y
Y
(ftm very small), it may indeed be best to use motions which minimize (y")rmS- Given a choice, the designer should choose motion
specifications which have smaller (y") r m s values. The conclusions
of Section 2 are reinforced by the form of equation (21).
4.
T w o D O F S y s t e m P a r a m e t r i c Error C h a r t s
Suppose now that the designer has adequately described his
system in terms of one DOF parameters Hm, Vc, Vk but errs in his
estimation of the distribution parameters -q, X, 7, when attempting to describe the 2 DOF system. For many systems with complex geometry, such imprecise definition in the model may be the
expected situation. The real distribution parameters, if, X, 7, then
must be used in describing the 2 DOF dynamic synthesis coefficients PU i = 0, 1, 2, 3, 4 introduced in Part I. The dimensionless
distribution error factors may be defined as
*
Fig. 1
-8
I.O
1.2
Pr = Y - y
(24)
Pi ~ Pi
Pi
fit
(25)
l<uy)r
Computer simulation has shown that this result is a useful approximation even if the above assumptions are relaxed. Further,
equation (21) shows that 5rms = fim such that given (5 rms )o, Mmo
for an existing system, then the brms for any other system fim is
7jX(l-7l)(l-X)}ji,2
X (1 - X - 1
(21)
lt m a v
iiafa
(22)
where
Also, if the only error is due to off-speed operation, the RMS distortion can be written as
&P4I1
h'-mO
**Jl
\h")r
(23)
PiPi
{[(1 + 0)(1 -
(28)
(29)
+ ftp4
lBx=o
(27)
are indepen-
(26)
IB=
8)1*0
= {[(1 + /3X)(1 -
(30)
m
3
1184
NOVEMBER 1975
Fig. 2
be
1.0
"-I.6
/
.8
j34p4
= (1 - 2 J J ) ( 1
/1
-"^06 y
-X)77X/3/J.
/^02
IBX=O
P4
= (1 -2x)(i -
(31)
v)vxfixiiJ
v,r
l=o
A/
V*
0.0
tin
(32)
0.0
0.0
ft,
.4
.6
=JJ\(1-\)(1-27J)
Fig. 3
(33)
/?,, or /9 2T
/34;i = 7 ? X ( l - j 7 ) ( l - 2 A )
are functions of the modeled distribution parameters only. Thus,
if the designer has some estimate of the probable magnitude of
errors in the mass deformation distribution (fSv,fS\), the @tn,f3t\
functions will indicate the sensitivity of the p 4 synthesis coefficient to these errors. The error distribution coefficients have been
plotted in normalized contour charts (Figs. 3 and 4) with a normalization factor of 4 in each case. Note that equation (32) is an
approximation which allows the qualitative representation of the
charts. For complete accuracy, the designer should use equation
(28).
The above formulation has been carried out for all the ft error
coefficients and is listed in Table 1. Also, all the /3( Pa error distribution coefficients have been plotted in Figs. 3-8. These show
that to minimize the system's sensitivity to distribution error, the
mass and deformation should be near the cam (?) = 0, X = 0).
The condition for best dynamic synthesis (?) = 0, X = 1) shows increased sensitivity due to |8 3 ,\(Y = 1), /?2,\, /?i,\- The significance of
this contradiction will depend on the system and the magnitude
0.0
0 0
-0.2/
i,r
i"l
V4
( .0
0.0
.4
.6
.8
X
/?4x or 0'2\
NF
B. = n X ( l - X ) ( l - 2 n )
4n
B 4A - n X ( l - n ) ( l - 2 X )
B, - nX(l-x)
3n
,
.
UXU-2A), Y-0
B3
X " ^ ( l - n ) X (1-2X), Y - 1
Approx. Form
Remarks
{(1+8
( 8 . B + 6.,B,})J 2
4n n
4X.A m
)n (1+8,)X ( l - x - B X ) ( l - n - P n)
A
A
n
-r,A(l-n)U-X)>i
m
B3P3
08 -iq
Exact Form
+0.125
(PEAK)
Fig. 4
Table 1
. ,
1 ~0AI
( f ( l + B n ) n ( 1 - Y - B Y) + (1+B )Y ( l - n - B n n ) ]
[(1+BX)X (1-X-B X X)] -
{f
Vn
+ 8
3X X>V m
In(l-Y) + Y ( l - n > ] [ A ( l - X ) ] ) p c i i n
62p2
{(1+B
)n + ( l - n - 6 n ) ( l - X - B , X ) n
n
X
n - (l-n)(l-x)}>i +
B2X -
YX(1-Y)(1-2X)
B 2 Y - YX(1-X)(1-2 Y )
B
{8
2n 8 n
2AV>V
ID
{(1+8
Vl
Y(1-Y)
{(1+B
-
)Y (1+0!)X ( l - Y - 6 YX1-A-B.X)
A
A
Y
Y -
2YSY
Wl
XU-A)}^
)Y + ( H - J v l ) ( l - H , l )
Y
A
{8
1Y 6 Y
+ 8
lX8Xll,c
lY "
U-Y)U-*)}UC
Vo
+ {6
-X(1-Y)
YX
NF = Normalization Factor
g
n
BZH
ft
. X~X
n *
X
X '
1 -1=1
Y
Y
vv
\ \ \
1.0
o.rv
/ /
N
0.6^
0.6/
\4
o.y
v, r
\0.2
A0.2/
.20.0
0.0
.4
.6
Fig. 5
0.0
ft.
Fig. 7
1.0
ft>or/
y l.o
-- y o.o
2(1 + W P j
j 6(1 :
24(1 + ,i 1 (p 1 | 120(1 + 4 ) p ,
(1 + /f>Pn | 3 d
4(1 - H ^ p ,
j 20(1 +,l,)i>.
'
(p-3)(/>-2)(p-l)pC,
0
(A-2)(/i-l)/ic,
\o.\
\
-0.2/
-0.4;
(P-l)/>c;
pc,
fftPi
(/-3)(/-2)(/-l)/CI
{ &Po
Fig. 6 ftu
(1-1)1)-UK,
(l-l)IC,
IC,
5.
S(pi-Pi)yi(0
(34)
<=0
S(l+(3 i )P ( 5 1
(n
= E^P ( y i
(=0
1186
NOVEMBER 1975
i=0
<,)
2(1 +ft) f t
0
1
0
r
6(1 + &)fc
A
0
1
12(1 + >h)fh
^
20(1 + ih)fh
30(1 + (^)p;
I IC-DK', /
which when inverted gives a matrix formulation similar to equation (14). Consequently, the fundamental vibrational amplitude
is obtained in the same manner
(38)
\p'-ct{P2y
(39)
&P2I
or
A = pWp(nmrn{V
+ (1 - i , ) ( l - X ) } < * / 2 ) - 1 | i ; X f t 1 - x a -7?)|3 X |
(40)
1.0
0.0
-ozs'
y,r
OAJT
T w o D O F System R M S Distortion
To determine the RMS distortion for the 2 DOF system, it is
again assumed that d-y is small so that its higher order properties
5i', &i", . . . go to zero. Equation (35) reduces to
(41)
{=0
(42)
(61),= \v^-i(i-v)Px\v-Jyi")rms
(43)
Combining the RMS values for the 1 DOF and 2 DOF systems
yields 4
(44)
M e a n i n g of D y n a m i c S y n t h e s i s
Suppose t h a t a cam has been manufactured by assuming the
model to be absolutely rigid (K = > and S = y). This may be
due to the economic necessity of using off-the-shelf cams or to the
-0.6,
-oa/
"(MAX. '
1.63 X I 0 " )
1.0
JOB
10.6
o.y
/OA
yl^O-Z
0.6/
I J
11
Fig. 9
OA/
Fig. 10
QZ/
NOVEMBER 1975
1187
M S l , IMPROVEMENT AS A RESULT OF
*
THE SYNTHESIS PROCESS
400
Fig. 11
800
1200
1600
2000
failure of the designer to recognize that accounting for system deformation would improve its operation. This is equivalent to synthesizing the cam for wd = 0. When the cam is placed in operation, the result is clearly an over-speed condition. Transient and
RMS deviations for this type of operation are plotted in Figs. 11
and 12, respectively. In addition, deviations for the 4-5-6-7 synthesized polynomial are also superimposed on these plots. The
improvement by reduced amplitudes for operating speed a > 0.7
u>a is quite apparent. In many cases, this reduced sensitivity to
off-speed operation may be the best recommendation for dynamic
synthesis.
1188
NOVEMBER 1975
Fig. 12
9. Deviations are significantly reduced for systems where dynamic synthesis has been used above 70 percent of design speed.
This is recommended especially where wide speed ranges or anticipated increases in production rates may occur.
10. The above rules may be extended to a large class of medical devices made up of gearing, cams, and linkages where relatively small motion ranges exist. For example, high fidelity pickups and speakers are in this class. The inverse is the filtering system which is designed not to transfer any information. Inversion
of the above rules could prove fruitful for those systems.
References
1 Matthew, G. K., and Tesar, D., "One Degree-of-Freedom Cam System Synthesis and Analysis," submitted to the Journal of Machines in
March, 1974.
2 Dudley, W. M., "New Methods in Value Cam Design," SAE Quarterly Transactions, Jan. 1948, Vol. 2, No. 1, pp. 19-33.
3 Thoren, T. R., Engemann, H. H and Stoddart, D. A., "Cam Design
As Related to Valve Train Dynamics," SAE Quarterly Transactions, Jan.
1952, Vol. 6, No. 1, pp. 1-14.
4 Stoddart, D. A., "Polydyne Cam Design," Machine Design Magazine, Three Parts, Jan., Feb., March, 1953.
5 Barkan, P., "Calculations of High-Speed Valve Motion With Flexible Overhead Linkage," Transactions of the SAE, 61,1953, pp. 687-700.
6 Hartman, J. B., "Anticipating Dynamic Behavior," Transactions of
the 4th Conference on Mechanisms, October, 1957, pp. 9-15.
7 Johnson, R. C., "A New Point of View on Minimizing Cam Vibrations," Machine Design, August 9,1956, pp. 103-104.
8 Johnson, R. C., "The Dynamic Analysis and Design of Relatively
Flexible Cam Mechanisms Having More Than One Degree of Freedom,"
Journal of Engineering for Industry, November, 1959, pp. 323-331.
9 Barkan, P., "High-Speed Spring-Actuated Cams," Transactions of
the 5th Conference on Mechanisms, October, 1958, pp. 64-76.
10 Rothbart, H. A., "Cam Torque Curves," Transactions of the 5th
Conference on Mechanisms, October, 1958, pp. 36-41.
11 Nourse, J. H., Dennis, R. C , and Wood, W. M., "Recent Developments in Cam Design, "SAE Summer Meeting, June 5-10,1960.
12 Johnson, R. C, "Analysis and Design of Cam Mechanisms Having a
Varying Input Velocity," Transactions of the 7th Conference on Mechanisms, October, 1962, pp. 190-201.
13 Weber, T "Filter Theory Applied to Cam Dynamics," Transactions
of the 6th Conference on Mechanisms, October, 1960, pp. 48-54.
14 Roggenbuck, R. A., "Designing the Cam Profile for Low Vibration at
High Speeds," Transactions of the SAE, 61,1953, pp. 701-705.
15 Johnson, A. R., "Motion Control for a Series System
" ASMS
Paper No. 64-Mech-7, 8th Conference on Mechanisms, 1964.
16 Freudenstein, F., "On the Dynamics of High-Speed Cam Profiles,"
Int. J. Mech. Scl, Pergamon Press, Vol. 1,1960, pp. 342-349.
17 Mackowiak, J. F., "Dynamic Synthesis of Direct Contact Mechanisms," Master's Thesis, University of Florida, April, 1967.
18 Rennak, J. R., "Steady State Analysis of Linear Systems Subject to
Periodic Excitations," Master's Thesis, University of Florida, March 1972.
19 Chen, P. Y., "A Refined Algorithm for Finite Difference Synthesis of
Cam Profiles," Journal of Mechanism and Machine Theory, Vol. 7, 1972,
pp.453-460.
20 Chen, P. Y., "Analysis and Design of Cam-Driven Mechanisms With
Industry, Trans. ASME, Vol. 95, No. 2, May 1973, pp. 519-524.
22 Winfrey, R. C, Anderson, R. V., and Guilka, G. W., "Analysis of
Elastic Machinery With Clearances," Journal of Engineering for Industry,
Trans. ASME, Vol. 95, No. 3, Aug. 1973, pp. 695-703.
NOVEMBER 1975
1189