Gear Noise and the Making of Si' ent Gears

J. Liu, H.I. Wu,

M.H. (Jian, c.P ..Chen,

DaGan University of Technology, DaHan, China

Summary:
research group has been engaged in the study of gear
noise for some nine years and has succeeded in cutting the
noise from an average level of some 81-83 dBto 76-78 dB by
both experimental and theoretical research. Experimental
research centered on the investigation into the relation between the gear error and noise. Theoretical research centered
on the geometry and kinematics of the meshing process of
gears with geometric error. A phenomenon caned "out-ofbound meshing of gears" was discovered and mathematically proven, and an in-depth analysis of the change-over
process from the meshing of one pair of teeth to the next is
followed, which leads to the conclusion we are using to solve
the gear noise problem. The authors also suggest some optimized profiles to ensure silent transmission, and a new
definition of profile error is suggested.
QUI

Introduction
For some nine years, OUI research group has been engaging in the study of gear noise and the making of silent gears,
six papers in English have been published on the international
conferences and periodicals (see Reference 1-6). Experiments
were done on machine tool headstocks' power gears, as they
represent the sort of light-load, medium-speed gears for which
silent transmission is such a problem. A new single flank total
composite error tester (Initiated by Huang T onglian of China.
See Reference 4,) was used to find the relation between gear
geometry and the noise. The tester is able to .indicate the pitch
error, the profile error, the final transmission error and the
change-over character (from the meshing of one pair of teeth
to the next). Conclusions drawn from these experiments have
also been varified by machine testing. When we reached the
required accuracy, the gears offer the expected silent transmission. (Under 78 dB measured at 300mm away from pitch
point, minimum 73 dB.)
for theoretical analysis, a deep investigation has been made
on the meshing process of involute spur gears. A phenomenon called "out-of-bound meshing" of gears was discovered;
that is, with the presence of error, the actual contact line of
a pair of gears is much discrepant from the theoretical one.
It usually goes along a broken line linked by a sector of the
addendum circle of one gear and a part of the theoretical contact line, (4-5) With the disclosure of this phenomenon a
precise and thorough statement on the change-over process
of gear meshing (a geometry and kinematics of the changeover process of gear meshing) is made.

Gea(Technology

Summarizing from the above mentioned experiments and

analysis we fonned the following hypothesis concerning gear
noise.
Dynamic Behavior-and Change-Over
Process of Gear Transmission
Out-of-bound meshing and change-over impact of gears.
Since gear noise is largely induced from change-over impact,
thorough analysis of the change-over process must be made.
But before entering the study of it, two things must be
mentioned.
a. Out-at-bound meshing (ORM) of gears. - When a gear
is meshing with its pinion, theoretically speaking, the contact
point will move along the common tangent 1-4 (See Table Ia),
starting from 1(the intersecting point of addendum. circle of
driven gear with glg2), and ending at 4 (the intersecting point
of addendum circle of driving gear with glg2l. On most occasions, there are two adjacent pairs of teeth in mesh, but,actually, there is always a difference in normal pitches of driving and driven gears, Assume t'o>to as in the case of Table
Ia, the contact point, after passing 4, wi1l continue the
meshing and the gap at l' will be narrowed until at point 4",
the next pair of teeth come into meshing at e' and take over
the transmission. So the actual contact line is a broken line
e'44". There is a shock at the instant of take-over, called "leaving impact," of OBM. Table 1b shows the other case, t'o<to'
where the contact line is 2"2e. The impact thus produced is
called approaching impact.(4.5f
b. The single flank total composite gear error tester, - It
is the main measuring instrument we have been using in this

AUTHORS:
DR. C.P.CHEN
is a professor at Dalian University of Technology in charge of post-graduate students. His field of research is
new techniques in integrated precision engineering. He is a graduate
in mechanical engineering from [iantong University, ShanghQi.
PROF. J. UU is a professor at Dalian University of Technology.
He directs post-graduate students ana specializes in gear noise
research and advanced gear making theories. He is a gmdtlate of
Dalian University of Technology in mechanical engineering.
H.J. WU is an assistant professor and lecturer at Dalien University of Technology. In addition to his teaching responsibilities,
he works in gear noise theory and gear error measurement. He is
a graduate of Dalian Institute of Technology in mechanical
engineering,
M.H. QIAN is a post-gr-aduate student at Daiian University of
Technology working in gear noise and silent transmis.siorl.

,
I

a) Perfect in geometry

Z - Number 01' leeth

b)

b) Error in circuldJ" pitch (irUxing error)

fig. 1

research work, It is an all-round gear tester developed from

the ordinary single Aank gear transmission error tester.(4)
The result coming out is the transmission error of every individual pair of teeth (from tip to root), but overlapped into
a polar coordinate, where e of the polar coordinate represents
the transmission error, while (J of the polar coordinate
represents the angular position of gear, which detereunes the
position of contact point on the profile, expressed by its
generating angle tPc. (See Fig. 1a) The greatest merit of this
tester is its thorough exposure of the change-over
characteristic. Fig. 2 shows the expression of different errors
on the tester's record. Fig ..1 indicates the characteristic points
on the profile. T and R are the tip and root points of the whole
profile. The contact line is g1g2. The pitch point is pt. The
generating angle of point c is tPc' which also indicates the
angular position of the gear when the contact point passes c.
B C is the basic section of profile which offers a contact ratio
1. Fig. Ib shows the location of the points on the tester's
record. where 9b=Oc=360 /(2z), where z = the number of
teeth. Dot limes show the OBM. TB and CR are sections
generally considered for tip/root relief,
Now let us see how the change-over takes place. From
Table Ia, lb. one may think that the way of change-over
could be either as shown in Table 1a or Table lb. But
remember that both Table 1a and Table Ib are based on the
assumption that the profile is an involute curve with error in
pressure angle (generated from a base circle with error in
diameter), but without tip (root) relief. (TN in Fig. 3a.) Here
the change-over point falls on the DaM zone, Now let's see
Fig. 3. If the profile has tip relief 1-2IaJ'ge enough to.cover the
normal pitch error. then the change-over point e' will fall
within the profile 2lR. It iscalled inside change-over as shown
in Table IcIt is a change-over without OBM. The convex
pro. iles (See Fig. 14, Fig. 15) also belong 10 this group.
Fig. 3b shows the case with profil shape 'error. It is a
~ypical saddle form profile. The pressure angle is too small.
with a relative reHef, and the outside change-over will be
transformed into inside change-over .
Table Ic shows a profile with normal pressure angle, but
with tip relief to obtain inside change-over. Since the tip relief
should be large enough to cover the normal pitch error, a
fairly large intersecting angle 6' results, and consequenently,
little benefi't could be gained in reducing noise. Table 1d

~~

'T

T. l

T]

R]

dl Pressure angle 100 large (rool hlgher)

e) Concave profile

(OU!

of involute)

lip too high

Fig. Z - Geometric 'errors expressed on tester's record.

y~.. el./\I~
",..

.'
I

\
I

b)

TR - original profile
12 - tip relieving
e
- original change over point
e1 - chang ov r point after tip relieving
Flg. 3 - Transformation
tip

of outside change over to inside chang

over by

relieving

shows another way of transforming outside change-over into

inside change-over. It is done by adapting the pressure angle
to obtain. correct normal pitch (normal pitch to = t,c~sa,
varying a, we can get right to). expressing on the testers
record, the geometric expression of right to is !that th adjacent curves willi link up very welL So if the pressur a.nglecan
be adapted to make the adjacent curves meet very we'U,a very
small "relative relief" can guarantee the inside change-ever,
Mcrch/Ap~1I1990

Table 1 Change-over character of gears

Outside change-over

driven

Figure of

contact

~iQf- ,

4 \:..........

-.- \t>\\.

..-- .\

RI

~~
IT
I I

R2

Brief
Description

.~
0

\.

g1.

--

gl ~

driving

-----2 -2-e

e'-e

T2

r-

T3

TI~
tTl

./

IT)

\I

R\ I

"

R2 I

R3 "

Type (b)

at leaving
Change-over impact
induced by too small
pressure angle

Characteristic
parameter for
gear noise

driven

Type (a)
Impact happens

< -~
= \::::

e'4-4"

Contact line

;:~~

I\)~

driving

Tester's
record
I

,\:'i
\.

Inside ~~ver

, e'

f'

Change-over impact
induced by too large
pressure angle

Type (d)
at change-over point e' e

Pressure angle normal. inside changeover obtained by tip

(root) relief which
should be great
enough to cover the
indexing error (error
in circular pitch)

r~

GearTechnology

Inside change-over
obtained by adapting
pressure angle to
obtain correct norma1
pitch (in tester's R!COrd
to link up the adjacent
curves), with slightest
tip (root) relief to
avoidOBM

Intersecting angle 9 at change-over point

10

while intersecting angle (),and hence the impact can be kept

very small. It is apparent that this way of obtaining inside
change-over (Table Id) is much more favorable than the
previous one (Table Ic),
Outside change-over is generally considered as unfavorable, This is because, in the OBM process, the tip of one
gear tooth actually does not mesh with the profile of the other
gear. There is no common tangent for them. The tip edge of
one gear tooth just scrapes the profile of other gear, and what
is more, theOBM curve in .thetestet'sl'Cord is steep. (See Fig,
2c.) This results in a very large intersecting angle, causing impad. So a gear with approaching OBM is a typically noisy
gear.
But leaving OBM acts quite differently. This is due to the
phenomenon of losing contact in gear meshing, As shown in
Fig. S, it goes along the dot line. (For detail see Fig..16c.)
Experiments show that a gear system with nonnal pitch of
the driving gear is a little bit larger than the normal pitch of
the driven. gear offers fairly silent transmission,
So there are four typical forms of change -over as listed in
Table 1. for the outside change-over, the characteristic

I ~
I

\:

Type (c)

at approaching

Normal pitch difference

.:1to= to (driving)-to(driven)
to = normal pitch of gear

~~

\
I

parameter is the normal pitch ditference .:1to; and for inside

change-over, it is the intersecting angle (J which determines the
impact and hence the noise. (See Fig. 4.)
Some discussion on low noise gear profile and the definition of profile error ..In the preceding section, we have discussed the gear noise at the instant of take-over. Now let us
see what happens during the whole process of meshing of a
pair of tooth profiles (from root to tip), They work just like
a pair of cams. According to the theory of conjugation, they
shou1d be made .into accurate involute curves (not concave
nor con vex on tester's record) to ensure smooth transmission
during this period. The characteristic parameter for noise in
this case is the reciprocal of the radius of curvature of the
curve on the tester's record, (See Reference 4.) But as mentioned above, in order to 'ensure smooth change-over,
pressure angle should be adapted to link up the adjacent
curves. So we suggest redefining the profile error, splitting it
into two items:
a) Discrepancy from involute curve, that is, draw the two
closest involute curves of the same base circle (not necessarily
the nominated base circle), The normal distance of them, M

Table 2 Gear noise and its geometric error

Il.O

....

I,

1100

ci

l!

::;j

::;j
c t:
CII

'1:;

..

......v

1100

<1

1641
1741
2251
1641
2261
0331
1331
2161
1641
3041
0341
1741
1261

2161
1641
3041
0941
1641
2261
225]
1741
3041
1741
1641
1741
1321
3041
1641
2161
0941
2161
2971
1741
2251
0341
1321
0941
2251
0951

-1.5
-1.5
-1
-1.5
-2.5
-1.5
-2.5
-2.5
-2
-2
-1.5
-3
-5
-1.5
-2.5
-2
-2
-1.5
-7.5

0151
3042
0342
1432
0312
2032
0151
2032
0332
0332
0152
0151
1332
0152
1332
0151
0152
1432
0152
0331

1331
2032
1642
2252
2032
0312
1332
0332
2032
1642
1332
2251
2031
0332
2032
0311
1331
2032
2252
2031

c:;

l!

c::

....

c:;

s ....!II

'1:

CII

Il.O

>

'1:

nj
CII

"

<1

.....0

s~

....

~
~
0 >
Z .5 ~

Fi,g.,4- Vector diagram of change over impact.

-...-.

tlta

(\1))
OBM

/2=12
~~

\~

1
2
3
4
5
6
7

8
9

Fig. 6

10
11

fig. 5

12
13

14

"E

IX

rig. 7

.fig. 8

is defined as the profile error. (Fig. 6). If the BTICR section

is used for tip I root relief, then only the shape error or BC section will count. Tiplroot relief should not be confused with
profile error. As shown in Fig. 7, profile error is M instead of

M'.
b) The actual pressure angle in this case is determined by
the radius of base circle Roose; that is, a = Cos -l(Roose I Rref},
(See Fig. 8.)
Experimental Research on Gear Noise
Perhaps what interests the gear manufacturers most is the
results of experimental research. They wish to know what
items of accuaracy influence the noise most and how dose
tolerances must be on errors to ensure silent transmission. The
purpose of experimental research was to find the answer for
them, but we also did some other experiments with the purpose of investigating the phenomenon of gear noise.
Influence or gear error on the noise level. The experiment
was done on m=2.5 z=SO B=20 lathe headstock gears,
They were nob bed. shaved (without shave profile correction), hardened, and then subjected to a "controllable electrochemical honing'~3J with the intention obtaining samples
with different magnitudes of errors. The profile of gear
samples thus obtained are of typical saddle form, but difJerent
in concavi ty M. It is a pity that in this batch ol gear samples.
there was no convex profile, and some 'error in pressure angle

15
16
17
18
19
20
21
22
23
24
25
26
223
224
225
226
227
228
229
230
231
232
233
234

235
2.36
237
238
239
240
241
24.2

75.7
75.8
76.8
77.S
77.6
78
78
78
78.2
78.2
78.3
78.4
78.5
78.5
78.7
78.8
78.8
79
79
79.1
79.2
79.2
79.2

79.5
79.5
79.5
84
84
84
84
84
84.1
84.2
84.2
84.3
84.5
84.5
84.6
84.6
84.7
84.8
85
85

85
85.S
85.6

1641
2161
2161
3041
22251
2161
1641
1641
2251
2161
3041
2251

1641

-1.5
+1.5
+1
+1
+4.5
-0.5
+1.5
-0.5
+4
+1
+0.5
+4
+3
-0.5
+0.5
+3
+0.5
+0.5
+3
-1
-3
+1.5
+3.5
-1
-4
+2.5

2
1.
27
42
7
18
5
15
53
43
14
25

142
171
172
110

-11.1
-11

-6
-5.5
-2
-7
+1
-1
-7
+6
-6

-13

-1

-8
-4.5

-7

-4
-2
-3
-4.5

-3
-3

-9
-5.5

-8.5
-13
-9.5
-10
-10

-8

-9.5

-11
-8.5

-8
-11

-s

-8

-]1

-5.5
-10.5

-8
-8

-4

-6
-8
-10
-10

50

11
8
21
73
6
65
68

16
95
124

49
12
34

228

227
159
231
230
180
160
139
210
184

166
217
155
211
173
220

I
I

was deliberately made. Gear noise and total transmission error were ,carefuI1y measured. Altogether 242 noise levels were
taken. They were listed on noise level sequence. Table 2
shows the two extremities, Nos. 1-26 are silent gears (below
Marchi April 1990

11

J041..3ifo ..----..3if

-1

2251 At. = +3

-0.5

--

<11= ~1.5

0151 type: (b)

...--.....

..3it. --8
Ai - -3

OJl1 type: (a)

.....--.11.- +5
Af - -5

2161 ..3ito = -O.S

0941 .110 .----

..3if-

---

- 2

..3if=0
1.331 type: (b)

0152 type: (b)

-I

':'1 --8
--3

....--.. .J

,--

.....Ar.,--2

41- -2.5

1641..3ito = -1.5
-..3if=-1

Arrow ... ad """"',Ihe direction <>I "",ving of <ontoct poinl when used ..
dnVU1881'.'
--r~ condition No;.., measured allOClO rpm. under .I ... d of 1.5 kg.m;
rmcrophoee ar lOOmm away from pnch pmnt; no lubric.itton: spred (i\tLO
I: I: goa, ledh marked

Fig. 9 ~ Testers

1432 type: (b)

0331 type: (b)

..---...lil. - -4

4/-

...

At. --5
.:of - -4.5

-3

10 ensure same meshing.

records of gears involved in the' first 5 noise tests

THE LEADER IN

GIEAR
D:EBU:R,IRIIING,

2031 type: (a)

2252 type: (a)

......

At. - +6
-~- ......
4/- -5

_ tlto - +2
t.f - -5

2032 type: (a)

..... _

At. - +3
t.f - -6

Fig. 10 - Tester's records of gears involved in the last five noise tests

GRIINDIING
SKIVVING
TOOTH R_ADIIUS
BRUSHING
ZERO SET-UP (P.C."
WHE,EL CHANGE
5SEOONDS

IB'EVEL
SPIRAL
iP,INION
SPUR
HELICAL
IRING GEARS
AUTOMATION

1817 - 18th Ave.

Rockford, I'L 61:104

81:5~398-1010 :FAX 815-398-1!047

DEALERS WELCOME!
CIRCLE A-6 ON READER REPLY CARD
1,2

Gear Technology

8OdB). Nos. 223-242 are the mast noisy ones, over 84dB. The
conclusion is that for medium-module, medium-speed, lightloaded spur gears, the main errors influencing noise are the
normal pitch difference and the profile concavity. Positive
Ato (normal pitch of driving gear larger than that of driven
gear) offers "leaving impact" change-over (Table Ia), which
is apparently more favorable than negative Ato (approaching
impact change-over, Table Ib), From theaverage value of the
extremities, increasing rate of noise could be roughly
estimated as every increment of 0.66 microns in negative normal pitch difference (normal pitch of driven gear minus that
of driving gear), plus a composite concavity increment of 0.62
microns will cause an increment of Idb in gear noise. The
tester's records of the gears involved infirst five noise tests are
shown in Fig. 9 (1641, 2161, 2251, 0941,. 3041), while the
tester's records of gears involved in last five noise tests (0151,
0152, 0311, 0331, 1331, 1432, 2031,2032, 2252) are shown in
Fig. 10. Fig. 11 shows gears ground on Chinese worm wheel

Table - 3, An index to Hnd tester's record with known number of gear

Number of gear
Number of Fig.

0151
10
1432
10

,
I

0152
10
1641
09

0311
10
2031
10

0331
10

2032
10

0941
09
2161
09

702L

703L

706L

709L

13

11

12

13

11

2251
09

2252
10

3041
09

711L

1331
10

12

710L

7061. type: (d)

703L type: (b)

..--_

_'::11,,=

70Zl type: (d)

.....----

~4 =-3
M~

-1

711L type: (dial

710L type: (b)

,.....- .... ~to

=. -2

.:11- -2

Fig. 11- Ground

=/

M-O

on

Chinese worm

wheel gear grinder 7232A

.....--...

709L.o.t" -

!lt~ = +2
M =-1

: - ......
I1f

Fig. 12 - Ground on Swiss worm

wheel gear grinder NZA

gear grinder 7232A (703L 710t) .Fig. U shows gears ground

on Swiss gear grinder NZA (706L 711L).Fig. 13 shows gears
ground on Chinese gear grinder 7232Awith final free grinding. (702L,709l).Table 3 is an index to find thetester's record
of known gear number.
Comparison of noise level of gears ground on different gear
grinders. Experiments were done mainly on worm abrasive
wheel gear grinders because they are productive, and because
they show promise for manufacturing silent gears. Table 4
shows the noise level measured at 300mm from the pitch
point of the meshing gears. From these test results, one sees
that with the addition of afinal worm wheel free grinding, the
noise level was dropped froman average of 8O.7dB to 76.7dB.
Fig. 14 shows the tester's records from Swiss grinder AZA.
The machine tool is accurate, but because the wrong diamond
dressing wheel was used the profile thus obtained is a convex
form.
Some gear makers think that a convex profile may offer
silent transmission, perhaps because most gear making
methods produce a concave profile andtypically noisy gears.
Therefore, it is understandable to suppose that a.convex profile will offer silent transmission. However, we have shown
experimentally that this is not the case. A batch of some 20
gears with profiles as shown in Pig. 15 were made. They
turned out to be very noisy (over 85dB).
Experiment for investigating the phenomenon of losing
contact in gear meshing. Gear noise researchers have long
noticed the phenomenon of losing contact. We did some experiments on this phenomenon as well. The gear noise tester
was isolated between driving and driven gears. A signal

!lt~ - I
I1f - 0

- -0.5

Fig. 13 - Ground on Chinese worm wheel gear

grinder 7232A plus Rna] free grinding

QUALITY
GEARS
UP TO AGMA 15, MIL -1-45208A &. MIL -510-45062
FROM A SINGLE GEAR TO LONG RUN PRODUCTION. INCLUDING
PRDTOT'l'PE
EMERGENCY REPAIR/REBUILO SERVICE
SIZE RA_NGE' FROM UNDER 1" to 48' DIAMETER

Fleishauer Ground Ge'31rs

Most Type Ge_3rs Manufactured
Compl,e'e Cusl.omer Speciticalio'ns

'0

METllIC OR AMERICAN STANDAlID

SPUR. INTERNAL & EXTERNAL
HELICAL. INTERNAL & EXTERNAL
WORMS. WORM GEARS
SERRATIONS SHAFTS
SPLINES. INTERNAL & EXTERNAL
SPROCKETS. CLUSlERS
SEGMENTS. SPINDLES
RATCHETS. GEAR BOXES

-.
~

Fully IlmpllementedISP,C. andl data

communicaUons capablliUes, utilizing
state 01 tile arl CMM'sand 1M,I. M IpreclsloD
gaarchacker to 30" lei/ameter to 18'" faclI'.

Jai,.fane(jear,. !)'"Co
P.o. BOX 409,Pl

YMOUTH,

MI 48170

(3113) 4592440,
In Mich. 18004821773

FAX (313) 4592941

CIRClJE A] ON IREADER REPlV CARD'

Marchi AprU1990

Table 4 - Comparison

NO.

Type of gear grinder

Chinese horizontal gear

grinder 7132A

Chinese worm wheel gear

grinder 7232A

Swiss worm wheel gear

grinderNZA
Swiss worm wheel gear
grinder AZA
Chinese worm wheel free
grinder

4
5

of noise of gears ground on different gear grinders

Way of profile generation

and indexing

steel belt and rolling disc

with virtual rolling radius
adjustable, indexing by
change gear and worm
wheel
Combine profile generation
and indexing just like
hobbing

Average
noise

No. of Fig.
of tester's
record

83dB

(data offered
by factory)

28

BO.7dB

Fig. 15

32

78.OdB

Fig. 16

81.SdB

Fig. 18

34

76.7dB

Fig. 17

No. of
noise
test

separate
Swiss worm wheel gear
grinder AZA
'Fig, 14 - Tester's record of gear ground on AZA

(a)

76.7ms

contact

WW~~~'W)\IU1~W,~,U'W~)~\mW,'WW~I(
separate

(b)

contact
2.S!,

separate

(c)

Fig. 15 - Tesler's record of gear with convex profile

Fig. 16 - Ti me series curves of losing contact

voltage with some resistors in series was connected to the

gears. If the gears were in contact, the signal voltage was short
circuited to low voltage. If they lost contact, high voltage appeared. Our experiment not only confirmed the existance of
the phenomenon, but also defined some of its characteristics.
Three sorts of losing contact were observed .. (See Fig. 16.)
a) Jump over teeth - The driving gear loses contact with
the driven gear; then, after crossing several teeth, it comes
again into contact.
b) Frequently lose contact for short instant within the

period of one tooth, lose contact and come into contact again
several times.
c) Jump over teeth and come to frequently losing contact
alternatively.
Experiment for investigating the influence of contactratio
on noise level of gears, A pair of non-standard high precision
gears were made with very high thin teeth, which GOuld provide a contact ratio of more than two. By varying the center
distanoe on the tester, a relation between the contact ratio and
the noise level was found as shown in Fig. 17. Minimum noise

14

'Gear Technology

seems to fall on a contact ratio a little bit less than 2 or 1. This

indicates that even under the slight load, the influence of torsional stiffness is still of some importance.
A 'tentative suggestion on the optimized geometry for low
noise gears. For the time being, we can suggest only the type
(d) in Table 1.That is, doing best to reach better indexing (circular pitch), for the existing error, adapt pressure angle ex to
ensure correct normal pitch, so that on the tester's record the
adjacent curves will meet very welt and a smooth changeover can be realized with slightest tip/root relief. Table 5
shows an example of silent gear pair,
Research on the Making of Silent Gears
From the previous sections, it is dear that gear noise is
produced:
a) at the instant of change-over,
b} during the process of meshing of a pair of gear
teeth.
The demand for accuracy is also directed at reducing noise,
that is;
a) The normal pitch diHerence.1.to as small as possible. If it is not possible to obtain this figure, a
positive at" is preferable.
b) The profile form error cf should preferably be as
dose to the theoretical invelute as possible with
minimum relief needed.
From Table 5, it can be seen that to guarantee ideal silent
gear transmission, the demand on key items of accuracy is
remarkably high.
1) To ensure final accuracy, the key technology is in the
finishing method after hardening.
2) To reach high accuracy in normal pitch, free finishing
methods aile very favorable. They can correct the normal
pitch error of the gears with the accurate normal pitch of the
tool (gear hone, worm-shaped, abrasive wheel, internal gear
hone, etc. In Table 4, it has been shown that an additional free
grinding to the worm wheel gear grinder cut down the
average noise level by 4dB. Another factory in Dalian succeeded in cutting down the gear noise by some S-6dB by intemalhoning. (The gear hone is with a shape of an internal
gear.) Further theoretical analysis can prove mathematically
that freefinishing can guarantee the type of gear shown in
Table Id, while grinding with constrained motion of wheel
and job will alwayslead to the geometry shown in Table Ic,
3) To obtain the ideal profile accuracy, a finishing method
with controllable metal removing rate along the full profile
(from tip to root) is keenly wished. The field controlled ECH
of gears is one of the new technologies we are developing with
this intention. (See Reference 3.)
Conclusion
As the introduction of this article has stated, all we are
writing here is just an idea formed in our minds by the results

Table - 5 Example of silent gear pair

Driving gear
Gear No. Fig. No.

702L
709L

Fig. 17
Fig. 17

Driven gear
Gear No. Fig. No.

709L
702L

Fig. 17
Fig. 17

Noise level
(average)
7S.6dB
74.2dB

11 -

1000 rpm

110

1.4

2.2
Conlolct ratio (a)

Fig. 17 - Relation between contact ratio and noise level

of experiments. It can basically explain the phenomenon we

can see now, but any disclosure by further experiments may
make it necessary to revise the explanation ..Besides, there is
stal a.distance between experiment and application in.production to cover But so far as we can see now, a compa:ratively
silent gear (say below 78dB) looks feasible. Far further reducing noise, other measure such as damping and absorbing, and
sound isolation, should be considered too.
ReJerences;
1. CHEN, C. P. et al. "Electrochemical Honingof
Al1nals of ClRP. VoL 30/10/1981.

Gears." OUT.

2. WEI, G.Q. et al., "An Investigation Into. The Ability of Correcting Error in ECH." OUT. Al1nals of ClRP, Vol.
351111985.
3. WEI, G.Q. et al., "Field Controlled ECH of Gears." DUT,
Preprint of 4-IPES, Precision Engil1eeril1g, No. 4, 1987.

4. LIN.

J. et al.. "Single Flank Total Composite Error

Analysis
of Gears." OUT, Preprinl of 4-IPES,
Precision Engineeril1g. No.2, 1987.

and Noise Prediction

S. LIANG, Y.R. "Out-of-bound Meshing of Gears in Shaving

and Its Influence on the Accuracy of Shaved Gears." Masters
thesis, 1982.

J. et 31. our, "Gea:rNoise &: the Making of Silent Gears"

Preprint from 1988lntemational Gearing Conference, China.

6. LlU,

7. SISIMU, KOAYIZIMY. "Gear Noise and its Pr vention." Applicational Machine Engineering. Vol. 15, No.9. (1974)
8. HUANG, L.l. AND ZHONG. X.Q. "Gear's Dynamic Composite Error Curve and Its Testing Method." Scientia Sinica,
No. 4,1973.
9. Machil1e Tool Noise - Principie and Control. Zhang Co.,
Tianjin University, 1984 ..
10.

LlU, J. Advanced Theory of Gear Meshing - A Textbook For

Postgraduate Students. 1980. (In Chinese.)

11. WU, H.J. "Spectrum Analysis of Gearing Noise." Master's

thesis, DUT, 1983.
12. QIAN. L]. "Gear Geometric Errors and Gear Noise." Master's
Theses. OUT, 1988.
13. WEI, L. "Experiment of Gear Noise." Master's thesis. OUT,
1985.

Acknowledgement: The achievement of this research is virtu !ly the sum of

the achievements of OUJ research group over the past 9 years. II should be
credited to all members and postgraduates of our I"E5eMCh
group. ,especially.
V.f. Iia,
D.L. Wang, W. U, Y.W. Llu, M.H. Wu, and V.P, Guo.

x.s.uo.

March,/ April 1990

15