Académique Documents
Professionnel Documents
Culture Documents
Abstract: This paper is devoted to a phenomenon known as corner contact, or contact outside the
normal path of contact, which can occur in spur and helical gear transmission systems under certain
conditions. In this case, a change in position of the driven gear with respect to its theoretical position
takes place, thus inducing a transmission error referred to here as the transmission error outside the
normal path of contact ( TE ).
o.p.c.
The paper deals with spur gears only, but the results are directly applicable to helical gears. It
systematizes previous knowledge on this subject, suggests some further developments of the theory
and introduces the novel phenomenon of top contact. The theoretical results are compared with
experimental measurements using a single flank tester and a back-to-back dynamic test rig for spur
and helical gears, and they are in good agreement.
Convenient approximate equations for calculation of TE suggested here are important for
o.p.c.
analysis of experimental data collected in the form of Harris maps. This will make possible the
calculation of tooth stiffness values needed for use in theoretical models for spur and helical gear
transmission systems.
Keywords: gear, transmission error, path of contact, corner contact, top contact
c geometrical construction angle (Fig. 3) ments in this direction were made by various authors.
C2
c geometrical construction angle (Fig. 3) The most significant step forward in the understanding
E2
Dh and Dh angles of rotation of the first and of gear kinematics was made in the late 1950s by Harris
1 2
second gears ( Fig. 3) [2], who suggested that the TE under different loads be
h geometrical construction angle (Fig. 5) presented as a series of plots against the gear rotation.
W geometrical construction angle (Fig. 3) These plots are now known as Harris maps and have
C2
W geometrical construction angle (Fig. 3) become a very powerful tool for clear understanding of
E2
W pressure angle at the tip ( Fig. 4) gear kinematics and can help to predict their dynamic
T
behaviour. At the same time studies of spur gear dynam-
ics were set up at Cambridge University. A dynamic
1 INTRODUCTION back-to-back test rig was designed for measurements of
the transmission error by means of optical gratings and
Corner contact, or contact outside the normal path of considerable progress was made in understanding spur
contact, can occur in spur and helical gear transmission gear dynamics [3, 4]. Subsequent research has put the
systems either owing to the bending of the teeth under design of profile relief on a proper scientific basis (e.g.
working load or to adjacent pitch errors. This paper references [5] to [10].
deals with spur gears only, but the results are directly It can be concluded that the static and dynamic behav-
applicable to helical gears. This statement becomes clear iour of spur gears has been studied by researchers very
if one considers a helical gear as a series of thin staggered extensively indeed, leading to its very clear theoretical
spur gears, although clearly there are some additional understanding. One question remains open though, that
complications in the helical gear case owing to their being the choice of tooth pair stiffness value for use in
three-dimensional geometry. Corner contact is highly theoretical models. The huge amount of experimental
unwelcome for several reasons: data that has been collected at Huddersfield University
in the form of Harris maps provides an opportunity for
1. Corner contact leads to rapid tooth wear.
extracting the data needed for calculation of stiffness
2. It creates unfavourable conditions for tooth lubri-
values for particular gear pairs, and in order to do so
cation. With regard to lubrication it should be noted
an accurate and convenient mathematical relationship
that the standard tests for the scuffing resistance of
for the corner contact is needed.
oils induce corner contact. This may be a convenient
The first attempt to understand the geometry of corner
way to speed up the failure process, but it is open to
contact in detail was made by Richardson [11] in 1958,
question whether this simulates lubricant failure
i.e. before the concept of TE was suggested by Harris.
under realistic design conditions.
He calculated the value of what would now be termed
3. High-frequency harmonics of the transmission error
the transmission error outside the normal path of contact,
( TE ) take place in the case of corner contact, leading
TE (i.e. when the contact occurs at the tooth corner).
to system vibration and unpleasant noise. By defi- o.p.c.
He referred to it as ‘the no-load separation of ideal invol-
nition, TE is a deviation of the position of the driven
ute gear teeth’, and his developments led to pages of
gear from its theoretical one (which it would occupy
exact, but cumbersome formulae. In 1962, Munro [3]
as defined by the gear ratio if both gears were geo-
proposed a different approach for calculation of TE .
metrically perfect and infinitely stiff ). It can be o.p.c.
It was based on the TE concept and used some approxi-
expressed either as an angular value or linearly, i.e.
mations, thus leading to approximate, but conveniently
as a linear movement at base radius. TE continuously
simple, formulae.
changes with time (as the gears rotate) and the sign
The importance of the phenomenon was hardly recog-
convention is that it is negative when the driven gear
nized by the majority of researchers at that time, which
lags behind its theoretical position. TE is a very
has probably been the main reason for a long break in
important concept in gearing. Created by a gear
achieving new results on this subject. Only in 1975 did
pair, TE is in fact an excitation for the whole trans-
Seager [12] produce some work on analysis of Munro’s
mission system that can lead to unwanted noise and
[3] developments, and much more recently a consider-
vibrations.
able step forward was made by Lin et al. [13] who came
Corner contact can be avoided if sufficient tip and/or up with an exact version of analytical formulae for calcu-
root relief on the teeth is used. Using various types of lation of TE (still referring to it as ‘separation’) and
o.p.c.
relief has become common practice in gear design since presented a thorough analysis of its effect on spur gear
the late 1930s when the theory behind this was proposed transmission.
by Walker [1], but the choice by designers of both This paper presents more results on corner contact
amount and extent of relief needed for particular load theory, bringing all hitherto acquired knowledge on this
conditions has been by trial and error or tradition subject into systematic order. It suggests a further devel-
through most of the subsequent years. opment to the more handy approximate version of the
Meanwhile, some considerable theoretical develop- formulae for TE discussed by Munro [3] and Seager
o.p.c.
Proc Instn Mech Engrs Vol 213 Part C C02798 © IMechE 1999
GEAR TRANSMISSION ERROR OUTSIDE THE NORMAL PATH OF CONTACT 391
[12] and introduces a novel phenomenon called top con- the driving gear, Dh , TE is different in the approach
1 o.p.c.
tact, discussing possible conditions under which it can and recess cases (hence Dh ≠Dh in
1(approach) 1(recess)
occur. Theoretical calculations of TE are verified Fig. 1). This is discussed in more detail in references [12]
o.p.c.
against experimental measurements on a single flank and [13].
tester and a back-to-back dynamic test rig and results This phenomenon can also take place in the case of
are presented. normal working conditions (CR>1) if the amount of
The transmission error due to corner contact is defined tip/root relief used is not sufficient to compensate for
strictly geometrically in Section 2. The exact mathemat- tooth deflection under the load applied to a gear pair.
ical formulae for the value of TE are discussed in This can be demonstrated on the theoretical Harris maps
o.p.c.
Section 3 and the derivation of a new approximate ver- for a gear pair with no relief in Fig. 2a. TE is a straight
sion of it is presented in Section 4. Sections 5 and 6 deal line in the no-load case, but with increased loads, and
with experimental results, and finally Section 7 explains hence tooth deflections, corner contact comes into play,
the top contact phenomenon. resulting in the sharp peaks on the TE curve (this is
discussed in more detail by Lin et al. [13]). It is interes-
ting to point out that at a certain load the TE curve
2 CORNER CONTACT PHENOMENON looks very similar to that in Fig. 1. Thus, the effect of
an incorrectly chosen relief profile can be very similar to
2.1 Conditions for corner contact to occur that of a CR of less than one.
Figure 2b presents measured Harris maps for a gear
In order to demonstrate corner contact, Fig. 1 can be pair with no relief and the resemblance between theoreti-
used which shows a schematic of part of the theoretical cal predictions and practical measurements is remark-
Harris map for a spur gear pair with no profile relief, able. Measured static TE curves are obtained from an
no manufacturing errors, zero load and the contact ratio optical analogue measuring system on a back-to-back
(CR) artificially chosen to be less than one (i.e. the centre dynamic test rig for spur and helical gears at
distance between two gears is artificially too large). The Huddersfield University. It is one of the sets presented
line BE represents the TE on the normal path of contact by Palmer and Munro [10] for a particular pair of gears
which, theoretically, is a straight line in this case. The with no profile relief (54 teeth each, module 3.738 mm,
normal path of contact BE is shorter than the base pitch, basic rack flank angle a=18°, outside radii r =r =
p (CR<1), and corner contact occurs beyond the a1 a2
b 105.08 mm, centre distance a=203.2 mm, face width
normal path of contact. This happens in both the 11 mm) and the applied loads were 7, 93, 176, 262, 347,
approach stage (i.e. before B, when a tooth pair is coming 430 and 560 N/mm face width. Effects of small profile
into mesh) and in the recess stage (i.e. after E, when a errors are visible in the case of the measured TE and
tooth pair is leaving mesh). As a result, a rapid change also the low-frequency level varies slightly (owing to the
in the absolute value of TE takes place with the rotation eccentricity of the gear pair), but the peaks due to corner
of the first gear, Dh . It starts decreasing again when the contact are clearly noticeable for heavier loads. Also, it
1
second pair takes over. The TE curves beyond the is clear that these peaks are asymmetric, with the steeper
normal path of contact, shown in Fig. 1, are slightly slopes occurring in approach rather than in recess, as
asymmetric because for the same angle of rotation of expected [12] and [13].
Fig. 2 Static transmission errors (Harris maps), no profile relief: (a) theoretical; (b) measured
case, contact occurs between the flank of the driven gear been ‘extended’ along the line (involute) CF is also
and the tooth corner C of the driving gear (Fig. 3). shown in Fig. 3 (dashed lines in Fig. 3). If that had been
Corner contact is also possible in the case of approach the case, the contact would have occurred at point F
(a pair of teeth in this position is also shown in Fig. 3 which lies on the common tangent line N N to provide
1 2
with thin solid serrated lines) when contact occurs a smooth conjugate motion of the driven gear. Thus the
between the flank of the driving gear and the tooth TE due to corner contact, TE , is geometrically
o.p.c.
corner of the driven gear. defined as the distance DF along the path of contact as
shown in Fig. 3.
Similarly, Fig. 3 shows TE in the approach case.
o.p.c.
2.3 Definition of transmission error outside the path of In this case the tooth of the driven gear is extended
contact (dot-dashed lines) to point A on the common tangent
N N , defining TE in approach. As discussed in
1 2 o.p.c.
In the recess case, the position that the driven gear tooth references [12] and [13], values of TE in approach
o.p.c
would have taken if the tooth of the driving gear had and recess are generally different for the same angle
Proc Instn Mech Engrs Vol 213 Part C C02798 © IMechE 1999
GEAR TRANSMISSION ERROR OUTSIDE THE NORMAL PATH OF CONTACT 393
Fig. 3 Tooth pairs of spur gears in mesh at the beginning (B), end (E) and beyond the normal path of
contact and the definition of the transmission error outside the normal path of contact, TE , in
o.p.c.
recess and approach
of rotation of the first gear even in the case of a 151 the position as defined by the tooth corner of the first
gear ratio. gear. Thus the separation as calculated by Lin et al. [13]
is the distance DF in Fig. 3 along the path of contact,
i.e. the transmission error TE in recess, and can be
o.p.c.
3 EXACT METHOD FOR CALCULATION OF calculated as a function of the angle of rotation of the
TRANSMISSION ERROR DUE TO CORNER driving gear, Dh , as follows (Fig. 3, where angle
1(recess)
CONTACT a is not shown to avoid additional construction lines):
C2
O E=
2 S [r sin(a −a )]2+[a−r cos(a −a )]2
a1 E1 t a1 E1 t
(3)
O C=
2 S [r sin(a −a +Dh
a1 E1 t 1(recess)
)]2+[a−r cos(a −a +Dh
a1 E1 t 1(recess)
)]2 (4)
Two lines are additionally constructed. The first line,
K K
r sin(a −a )
c =tan−1 a1 E1 t (5) CG, passes through point C and is parallel to the
E2 a−r cos(a −a ) common tangent N N . The second line, FH, is perpen-
a1 E1 t 1 2
dicular to both N N and CG. Lines FH and CG inter-
1 2
K K
r sin(a −a +Dh )
c =tan−1 a1 E1 t 1(recess) (6) sect at point H.
C2 a−r cos(a −a +Dh ) The following notations are used in these calculations
a1 E1 t 1(recess)
( Fig. 4):
r
a =cos−1 b2 (7)
E2 O E EF¬b (13)
2
W =tan a −a =inv a (8) FH¬c (14)
E2 E2 E2 E2
From the geometry of an involute, the distance b can be
r
a =cos−1 b2 (9) calculated as
C2 O C
2
b=Dh r (15)
W =tan a −a =inv a (10) 1 b1
C2 C2 C2 C2 All further developments are made with the assumption
Dh =(c −W )−(c −W ) (11) that b is small.
2(recess) C2 C2 E2 E2
The distances DF and CG are taken to be approxi-
TE =r Dh −r Dh (12)
o.p.c.(recess) b2 2(recess) b1 1(recess) mately equal, thus the TE is approximately
o.p.c.
noting that, as normal, TE is negative as the driven #CG
o.p.c. TE
o.p.c.
(16)
gear lags behind its conjugate position.
The same equations can be used to calculate TE The radii of curvature of the involutes S C and S C
o.p.c. C1 C2
in approach by changing some notations (effectively at point C are taken to be approximately the same as
‘swapping’ the driving and driven gears), i.e. those at point E, i.e. r and r :
1 2
TE can be calculated using equations (1) to
o.p.c.(approach) r =N E (17)
(12) by interchanging r and r and r and r and 1 1
b1 b2 a1 a2
substituting TE , Dh and Dh for r =N E (18)
o.p.c.(recess) 1(recess) 2(recess) 2 2
TE , Dh and Dh respect-
o.p.c.(approach) 2(approach) 1(approach) and thus the distances CH and HG can be approximately
ively. Using this method, TE is found as a
o.p.c.(approach) found:
function of Dh , but plotting TE
2(approach) o.p.c.(approach)
against Dh effectively ‘swaps’ the gears back c2
1(approach) CH# (19)
again [in this case Dh has the same value as
1(approach) 2r
1
Dh given in the original equation (11)].
2(recess) c2
HG# (20)
2r
2
4 CALCULATION OF TRANSMISSION ERROR From equations (16), (19) and (20) the equation for
DUE TO CORNER CONTACT USING AN TE can be written as
o.p.c.
APPROXIMATE METHOD
A B
c2 1 1
TE =CH+HG# + (21)
o.p.c. 2 r r
4.1 General idea of the approximate method 1 2
The method described in Section 3 [equations (1) to (12)] Seager [12] suggests that distance c be calculated
allows calculation of the exact value of TE due to corner approximately as
contact as a function of the rotation of the first gear, c=b tan W (22)
Dh . The disadvantage of this method is that it is in the T
1 where angle W is ‘the pressure angle at the tip’ which
form of an algorithm and cannot be used easily. As an T
alternative, the approximate formula suggested by can be found from the gear geometry:
Munro [3] and analysed later by Seager [12] can be used. r
A summary of the developments made by Munro and W =cos−1 b1 (23)
T r
Seager in the above documents is given below ( Fig. 4, a1
recess case). The distances r and r used in equation (21) can be
1 2
Proc Instn Mech Engrs Vol 213 Part C C02798 © IMechE 1999
GEAR TRANSMISSION ERROR OUTSIDE THE NORMAL PATH OF CONTACT 395
found: is 127.27 mm. Thus, the relative error is 10.67 per cent
but grows considerably with increase in Dh .
r = √r2 −r2 (24) 1
1 a1 b1
r =(r +r ) tan a −r (25)
2 b1 b2 t 1 4.3 Exact formula for distance c
Distance c can be calculated exactly as
4.2 Numerical results
c=sin a SC+cos a ES (26)
A number of numerical calculations were conducted to t t
compare the approximate method [ formula (21)] used where
together with equations (15) and (22) to (25) against SC=r |sin(a −a +Dh )−sin(a −a )| (27)
a1 E1 t 1 E1 t
the exact algorithm [equations (1) to (12)]. The calcu-
ES=r |cos(a −a +Dh )−cos(a −a )| (28)
lations show that for small angles Dh ( less than about a1 E1 t 1 E1 t
1
2–3°) the relative error lies within about 10 per cent. (this comes from the geometry in Fig. 4). Using the exact
One of the considered examples was the pair of spur equation (26) [with equations (27) and (28)] instead of
gears with no profile relief, described in Section 2.1. They the approximate equation (22) [with equations (15) and
have 54 teeth each, a module of 3.738 mm, basic rack (23)] reduces the error significantly. For example, for the
flank angle a=18° (thus giving the base radii r =r = numerical example described in Section 4.2 the relative
b1 b2
95.9863 mm) and outside radii r =r =105.08 mm. error is reduced to 1.77 per cent (|TE |=125.02 mm,
a1 a2 o.p.c.
For the centre distance a=203.2 mm the approximate as opposed to the exact value |TE |=127.27 mm).
o.p.c.
method gives |TE |=113.69 mm for the angle of The disadvantage of the exact equation (26) is that it
o.p.c.
rotation of the first gear Dh =2.5°, while the exact value requires additional calculations by equations (27) and
1
C02798 © IMechE 1999 Proc Instn Mech Engrs Vol 213 Part C
396 R G MUNRO, L MORRISH AND D PALMER
(28). The one-line formula proposed in the next section Now, from DO N E:
1 1
is another version of an approximate formula for calcu-
r
lation of distance c. cos(90°−h+a )= b1 (34)
t r
a1
and thus
4.4 New version of the approximate formula for
distance c r
h−a =sin−1 b1 (35)
t r
From Fig. 5, assuming that the chord EC is approxi- a1
mately equal to the arc EC, the following can be written: Finally, substituting equation (35) in equation (33) gives
the one-line formula for distance c as a function of the
|EC|#r Dh (29) angle Dh and gear parameters only:
a1 1 1
Line O V connects the gear centre O and the centre of
A B
1 1 r Dh
chord EC, V, and thus is perpendicular to line EC. Line c#r Dh cos a sin b1 − 1 (36)
a1 1 r 2
SE is parallel to the centre line O O , therefore giving a1
1 2
Dh The approximate calculations described in Section 4.1
3
CES=h− 1 (30) were again applied for the numerical example described
2
in Section 4.2, but equation (36) was used instead of
and from right triangle DESC: equation (22). This gave |TE |=125.04 mm, meaning
o.p.c.
that the relative error was reduced from 10.67 per cent
A B
Dh
SC=EC sin h− 1 (31) (as in the method described by Seager [12]) to 1.75 per
2 cent (and it remains approximately the same with
increase in Dh ).
A B
Dh 1
ES=EC cos h− 1 (32) Thus, an approximate algorithm consists of only four
2
relatively simple equations:
Equation (29) is substituted in equations (31) and (32)
A B
r Dh
and then the resulting two equations are substituted in c#r Dh cos a sin b1 − 1 (36)
a1 1 r 2
equation (26). After trivial trigonometrical manipu- a1
lations this leads to r = √r2 −r2 (24)
1 a1 b1
A B
Dh r =(r +r ) tan a −r
c#r Dh cos h−a − 1 (33) 2 b1 b2 t 1
(25)
a1 1 t 2
A B
c2 1 1
TE # + (21)
o.p.c. 2 r r
1 2
and the new version of the equation for c [equation (36)]
improves the accuracy in calculations of TE con-
o.p.c.
siderably compared with that when equation (22) was
used.
Note that the approximate method can also be used
to calculate TE in approach in a very similar way
o.p.c.
to that described in Section 3, with angle Dh
1(approach)
being calculated from equation (12) rewritten as ( Fig. 3)
Dh r +TE
Dh = 2(approach) b2 o.p.c.(approach)
1(approach) r
b1
(12∞)
bearing in mind that TE is negative.
o.p.c.(approach)
various increased centre distances (under zero load ). Table 1 Theoretical and measured (on single flank tester)
This can be done in the following way. values for TE for a pair of gears with 54 teeth
o.p.c.max
each, module 3.738 mm, basic rack flank angle a=
18°, outside radii r =r =105.08 mm and face
a1 a2
5.1 Theoretical calculations width 11 mm (the experiment is described in
Section 5)
The length of the normal path of contact BE can be
|TE |(mm)
found: o.p.c.max
a Difference
(mm) Theoretical Measured (mm)
BE=[√(r2 −r2 )−r tan a ]
a1 b1 b1 t
208.465 64.583 74.054 9.471
+[√(r2 −r2 )−r tan a ] (37) 208.665 74.513 81.829 7.316
a2 b2 b2 t 208.865 85.107 93.924 8.817
where a is the working pressure angle calculated using 209.065 96.367 107.314 10.947
t 209.265 108.278 121.424 13.146
equation (1). 209.465 120.831 133.326 12.495
Provided that the centre distance is large enough for 209.665 134.035 146.045 12.010
the length of the normal path of contact, BE, to be
shorter than the base pitch, p , corner contact will take
b
place. The theoretical value for TE can be calcu- values of TE can be found in Table 1 and in
o.p.c.max o.p.c.max
lated from either equations (1) to (12) or from equations Fig. 6.
(21), (24), (25), (36) and (12∞) using the iteration
method to satisfy
5.3 Comparison of the theoretical and measured results
(Dh +Dh )r =p −BE (38)
1(approach) 1(recess) b1 b
The increase in TE with centre distance is of similar
This will correspond to when the second pair of teeth o.p.c.
character for both theoretical and measured data,
‘takes over’ ( Fig. 1). The results of these calculations for although the measured values are larger than the theor-
several values of centre distance, a (giving a contact ratio etical ones by less than about 15 mm which is due to the
of less than one), for the pair of gears described in inevitable presence of corner wear. The change in TE
Section 4.2 can be found in Table 1 in the column o.p.c.
owing to corner wear is demonstrated in Fig. 7. If the
‘Theoretical results’ and are plotted in Fig. 6. corner is ‘worn’ (rounded slightly) then the contact
occurs not at point C, but at another point somewhere
5.2 Experimental measurements near C and closer to the driving gear centre, point C∞,
say, and in this case the TE is larger than that calcu-
o.p.c.
Using the single flank tester, the TE plots were measured lated for an ideally sharp corner by the distance DD∞.
for the same increased centre distances, a. The measured For the small distance CC∞, the increase in TE can
o.p.c.
Fig. 8 Harris maps for a particular gear pair with a contact ratio of 2: (a) theoretical Harris maps with the TE outside the
normal path of contact not taken into account, constant tooth stiffness; (b) theoretical Harris maps with the TE
outside the normal path of contact taken into account, constant tooth stiffness; (c) theoretical Harris maps with the
399
TE outside the normal path of contact taken into account, variable tooth stiffness; (d) measured Harris maps
400 R G MUNRO, L MORRISH AND D PALMER
Proc Instn Mech Engrs Vol 213 Part C C02798 © IMechE 1999