IMPROVING THE TEETH AND BEARING RACEWAY SURFACE ROUGHNESS QUALITY

BY GRINDING AND SUPERFINISHING PROCESS MASTERING

De Looze, J.B.
Dynamic Systems Product Center
EUROCOPTER Company
France

OVERVIEW

As a consequence of the increase in EC Helicopters sales, the manufacturing workload, in Eurocopter

Dynamics System Product Center, has ramped up from 25 to 30 % every year. At the same time, EC has increased the
Dynamic Systems reliability, and consequently, now matches high quality design requirements. The annual new Gears
Boxes developments have also brought more and more integrated bearing raceway on gears, using EC practiced Deep
Nitriding process. Teeth and raceway require a high surface condition level. In addition, it has to be noted that gears and
integrated bearing raceway are parts of core competency fields in the Dynamics System Product Center.
Therefore, ramp-up and reliability requirements impose a perfect mastering of the manufacturing process, under a high
level of industrial constraints and to avoid any industrial perturbation. Matching the design surfaces condition
requirements increases the risk of burns and cracks defects, so, the key of success lies in anticipating actions.
Root causes analyses revealed that the main contributors include burns detection process realized by NitWater
etching, grinding process variability, limiting state surface of grinding used to mitigate health defects occurrences risks,
and non-efficiency of Superfinishing process. In addition, the diversity of morphology range, of functionalities complexity
and of the parts‟ dimensions, makes these issues particularly difficult to address. Finally, the quality department has
reported a highly dispersed state in the surface measurement caused by the filtering roughness method (ISO2CR Filter).
A study carried out on the roughness filter impact has shown that the „‟Gaussian‟‟ filter is, for several reasons,
the best filter on the market today and that the ISO2CR filter brings up over-shoots. Changing roughness filter therefore
means developing a new roughness characterization value, based on the „„Gaussian‟‟ filter, tested and qualified during
the new gearboxes development. It has also revealed that there is no possibility to transform the roughness
measurement from the ISO2CR filter to the „Gaussian‟‟‟ filter. Several technologies were identified for comparison tests:
an optimization of „Smuritropy process‟ used in EC work-shop, the ISF (Isotropic SuperFinishing process), the Vibro-
Finishing, the Honing and the Belt grinding process. The „‟optimised Smuritropy process‟‟ was the most efficient and
allows us to achieve the expected process robustness. Only some internal diameter of race way required honing
process. The ‟‟optimised Smuritropy process‟‟ is based on a specific ceramic media, with adapted form and size.
The defined specifications for roughness robustness are completely reached, including reducing working time
by a factor of 2. This results in reducing recurring costs and consumable costs as well as in respecting the environment.
In addition, simulation of grinding and Smuritropy process projects are now launched to perform the mastering of
finishing operation.

1 BACKROUND the grinding process robust; (iii) promoting a

As a consequence of the increase in EC common superfinishing technology for teeth and
Helicopters sales, the manufacturing workload, in race way features to cover the variety of the parts;
Eurocopter Dynamics System Product Center, (iv) minimizing the recurring cost impacts, the
has ramped up from 25 to 30 % every year. non-recurring cost and cycle impacts on the
Ramp-up and reliability requirements impose a manufacturing ramp-up; (v) optimizing the
perfect mastering of the manufacturing measurement method, the choice of filters; (vi)
process.Teeth and raceways of the gear require a taking into account the environmental needs; (vii)
high surface condition level. So, the key of capitalizing the expertise of surface roughness
success lies in anticipating actions, to perfect requirement.
superfinishing robustness process. The present work first aimed to
investigate the diversity obtained for the
2 OBJECTIVES / STRATEGIES measurements of the surface finish on the gears
The objectives defined to solve the and to propose an alternative solution, in terms of
requirements have included: (i) Giving robustness methods or surface finishing characterisation. The
to super finishing process; (ii) Increasing the work then aimed to improve the current finishing
grinding required surface condition and making process and develop simulation tools.
3 CARACTERISATION OF SURFACE This means that it reduces some of the signal
FINISHING components and allows others to pass through.
The filtering modifies some of the input signal in the
time and frequency domain.
3.1 Generality According to the theory of Fourier series, a signal
Surface finishing is an essential stage in mechanical can be considered as the sum of sinusoidal signals
engineering as it raises several issues related to (infinite number if necessary) at different
friction, wear, lubrication, leakage and fatigue frequencies.
resistance. The end roughness is directly linked to
the life duration of the part.
Part 1

Lubricant

Then, the role of the filter is to modify amplitude and

Part 2
phase of the sinusoidal signals.
Figure 1: Two parts separated by an oil film
3.1.2 Principle of the average line method
The base length (cut-off) is equal to the profile filter
The surface finish is defined according to three wavelength, which allows the separation of the
different parameters: the form, the waving and the roughness and waving components (Fig 3). Five
roughness. All parameters in our profile are noted P base length or cut-off values are normalised: λc =
for primary, W for waving and R for roughness. [1] 0,08mm, 0,25mm, 0,8mm, 2,5mm ou 8mm [2] (Fig
4).

Figure 2: Representation of the profile defects

Figure 3: Separation of roughness and waving
The waving and roughness analysis of a profile is
carried out via an electronic filtering of the primary
profile (adjusted total profile) according to the
standard of the ISO 4288 average line [2] –
standard currently used by aeronautical industries to Base
lenght
characterise the surface finishing of the gears and of
the bearing raceway (cf. Fig. 2).

3.1.1 Filtering Concept

The filtering of a profile is as an electronic circuit, Figure 4: Representation of the average line
which performs a signal processing operation. (Red colour)
3.1.3 Parameters linked to the average line 3.2 Filter analysis
Two roughness parameters have been used 3.2.1 ISO 2CR filter (currently used at EC)
in this study:
Ra The ISO 2CR filter is an analog and digital filter. It
The Ra is the universal roughness reacts, as a causal system, to an input signal. [3; 4]
parameter. It is defined as the sum of the surface The phase shifting is a distinctive feature of the
areas of the peaks and valleys located along the causal system.
average line on a base length “I”. This parameter is
easy to define and gives a general description of the P(x) = p.sin( x ) R(x) = r.sin( x +
variations in height. It does not however distinguish
the profile peaks and valleys.
The BODE diagram in Fig. 7 shows that the ISO
l
1 2CR filter relays 75%.
Ra y(x) dx
l 0

Figure 5 : Profile roughness

Figure 7: BODE diagram of the ISO 2CR filter
Cut-off 0.25

Bearing length rate Rmr (Tp): It has to be noted that the transmission values at the
cut-off frequencies around λc=0,25mm will be
The bearing rate allows characterising a transmitted:
surface finish profile functionally. It is the ratio of the - Greater wavelength, i.e. λ=0,8mm will be
profile bearing length at a section level c, Ml(c) on transmitted at 20%
the evaluation length ln. The graphical transcription - Smaller wavelength, i.e. λ=0,08mm will be
of the bearing rate is given by the Abbott Firestone transmitted at 99%
curve:
Fig. 8 shows the behaviour of the ISO 2CR filter on
the primary profile:
In blue, the primary profile (input profile) with the
Sinus function (2ПxT)/0,8 cut-off=0,8mm; in red, the
roughness profile (output profile) after applying the
transfer function. In order to evaluate the reaction of
Evaluation lenght the ISO 2CR filters, an analogy between the
temporal and spatial frequencies (1mm = 1 second)
is proposed. The transfer function of the ISO 2CR
Figure 6 : Roughness profile and ABBOTT curve filter is therefore represented for a probing velocity
of 1mm/second.
 P(x) = p.sin( x ) R(x) = r.sin( x +

The Gaussian function is dragged from point to point

in the primary profile. At each position, a result point
is calculated by applying the weighing formulae on
the neighbour points upstream and downstream the
cut-off length (Gaussian curve width). The set of
these result points corresponds to the average line
Figure 8: Reaction of the ISO 2CR filter (in red) (cf. Fig 10).
on the primary filter (in blue)
PRIMARY
The major problem with this filter is that it does not PROFILE

contain a non-linear phase that can imply important

EVARAGE
shift phases in the roughness profile. The second LINE
GAUSSIAN
issue is that this filter is linked to the filter
transmission of 75% to the cut-off length. In addition
GAUSSIAN
the sum of the roughness and waving profiles is PONDER.
different from the primary profile [5].
“The 2CR filters can be used to evaluate the
parameters Ra and Rz, however, for the other Figure 10 : Representation of a Gear Gaussian
parameters, the distortion is significant”, extracted filter average line
from the ISO 3274 1996 standard [6].

Capability study of the ISO 2CR filter Figure 11 represents, in blue, the primary
A study of the capability and repeatability of profile (input profile) with a Sinus function (2ПxT)/0,8
the bearing rate measurement is realised on a cut-off=0,8mm and in red the roughness profile
sample. The normal distribution of the ISO 2CR filter (output profile) after applying the transfer function. In
is represented on the graph below) [7] (Fig 9). It can order to compare the Gaussian filters, an analogy
be seen that the normal distribution is relatively flat, between the temporal and spatial frequencies (1mm
suggesting variability and non-robustness. = 1 second) is proposed. The transfer function of
the Gaussian filter is therefore represented for a
probing velocity of 1mm/second.

Figure 9 Capability study of the ISO 2CR filter

3.2.2 The ISO 11562 Gaussian filter

The Gaussian filter is a mathematical function which Figure 11: Reaction of the Gaussian filter (in red)
is applied to the profile data. on the primary profile (in blue)
The roughness profile relays 50% of the primary
profile, at the cut-off length. There is no shift phase, The BODE diagram in Fig. 11 & 12, shows that the
which is an important asset for the method of Gaussian filter relays 50% at the cut-off frequency
analysis [8].
(green curve).
 3.3 Comparing the ISO 2CR filter and the
Gaussian filter

The Gaussian filter is more centred on the

cut-off frequency than the ISO 2CR filter.

mm
Figure 12: BODE diagram of the Gaussian filter

The transmission values at the cut-off frequencies

around λc=0,25mm will be transmitted:
Figure 14: Comparing the BODE diagrams of the
- Greater wavelength, i.e. λ=0,8mm will be
ISO 2CR and Gaussian filters
transmitted at 7%
- Smaller wavelength, i.e. λ=0,08mm will be
transmitted at 99%

3.4 Effect of the filter on the surface

Capability study of the Gaussian filter measurement
This study mainly deals with the ISO2CR
A capability and repeatability study on the filter (existing) and the Gaussian filter (target).
bearing rate measurement has been carried out (20
measurements). Fig. 13 represents the Gaussian 3.4.1 Effect of the ISO 2CR filter
filter normal distribution. On the contrary with the 3.4.1.1 Measurement of the scratched
ISO2CR, the Gaussian filter reveals a high peak, standard
suggesting a high capability and robustness. In order to highlight the overshoot
phenomenon generated by the ISO 2CR phase shift,
measurements on a scratched standard (scratch
depth of 8,6 μm) allow to identify the excess
resulting from the filters (see Fig 15).
This stage also aims to identify the defects that may
generate one or several overshoots after filtering:

Longueur = 4.1 Longueur

µmmm Pt = 8.62 µm Echelle = µm
10 µm = 4.1 mm Pt = 12.4 µm Echelle =µm
20 µm Longueur = 4.1 mm Pt = 8.62= 4.1
Longueur µm
mm
10 0 10
-1
7.5 7.5
-2
5 -3 5
Figure 13: Representation of the Gaussian filter
-4
normal distribution
2.5 2.5
-5
0 -6 0
Following the result obtained on the Cp and Cpk
-2.5 -7 -2.5
indicators, the measurement process with the -8
-5
Gaussian filter is centred halfway to the lower -5
-9
-7.5
bearing rate limit and the specification limit. This is a -7.5
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.2
0 0.25 reliable
0.5 process
0.75 1 as it is capable and centred [9].
1.25 1.5 1.75 2 2.25 2.5 2.75
Figure 15:
3
Raw
3.25
profile
3.5
on
3.75
the left
4 mm
and filtered ISO
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 2.750 30.25 3.25
0.5 3.5
0.75 3.751 41.25
mm 1.5 1
2CR profile on the right
The filtered profile with the ISO 2CR filter completely The ST2 measurements are similar; on the contrary,
changes the raw profile by creating an overshoot the ST2 results are very dispersed.
which generates a virtual peak that does not exist on
the profile. The method of the bearing rate To better understand the non-compliance rate
calculation takes into account the maximal height of obtained with ST n°1, the roughness profile of the
the peak; the distortions resulting from the filter ISO 2CR on the part n°8 is given below (case for
highly affect this parameter. which the bearing rate is minimum):
This type of defect is specific to a standard. The
overshoot phenomenon is random and cannot be
mastered. In order to evaluate the behaviour of the
ISO 2CR filter, measurements on the gear families
have been carried out.

3.4.1.2 Specimen test measurement

The measurements have been carried out on
Specimen Test, carburized, with two different
surface finishing qualities.
The Specimens with the first surface finishing are
The circled peak represents the overshoot
named Specimen Test N°1 (ST1).
phenomenon, which is directly linked to the shift
Note: Quantity 6, 2 measurements for each
phase of the weighing function average line of the
specimen.
ISO 2CR filter. Below, the behaviour of the filter
The Specimens with the second surface finishing
average line on the primary filter in the circled zone:
are named Specimen Test N°2 (ST2).
Note: Quantity 6, 1 measurement for each
specimen.

% Specimen test ST1 ISO2CR

The ISO 2CR average line filter (in green) is shifted

compared to the valleys. This shift leads to a
% Specimen test ST2 ISO2CR different process on the roughness profile. The local
defect (valley) will be represented by a reduced
valley and an amplified peak (overshoot
phenomenon), hence a result which does not
comply with the bearing rate at a cut-off of 0.8 µm
for the highest peak.

Below, a representation of the ABOTT curves to

exemplify the bearing curve of the roughness
profiles.
Figure 16: Evolution of the bearing rate with the
ISO 2CR filter for ST N1 & ST2

Fig.16 shows the measurements carried out with the

ISO2CR filter.
 Longueur
10 = 4.1 mm Pt = 6.16 µm Echelle =µm Longueur
10 µm Longueur = 4.1 mm Pt = 8.62 µm Echelle = 4.1
= 10 µm
µm µm Longueur = 4.1 mm Pt = 8.62 µm Echelle = µm µm
mm mm
5 5
0 0
4 4
-1 -1
3 3
-2 -2
2 2
-3 -3
1 1
-4 -4
0 0
-5 -5
-1 -1
-6 -6
-2 -2
-7
-7 -3
-3 -8
-8 -4
-4 -9
-9
0 0.25 0.5 0.75 1 1.25 0 0.25 0.5 0.75 1 1.25
1.5 0 1.750.25 2 0.5 2.250.75 2.5 1 2.75 1.25 3 1.53.25 1.753.5 2 3.75 2.254 mm2.5 1.5 2.751
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.25 2.5 Figure18:
2.75 3 Raw 3.25 profile
3.5 on the
3.75 left and filtered
4 mm
profile Gaussian on the right

The filter generates a profile which is similar to

reality but with a repositioning of the average line.
The peak-valley distances are slightly different,
6.16µm instead of 8.6µm on the raw profile.

3.4.2.2 Comparing the Gaussian filter with

ISO2CR
Fig. 19 shows the ST1 & ST2 specimen test
measurement obtained by the Gaussian and
ISO2CR filter.

Figure 17: Abbot curve between ST1 et ST2 % Specimen test ST1 ISO2CR / Gaussian

Fig 17 representThe ISO 2CR. S.T. N°1

roughness profile has a cut-off of -1.4µm at 50%
whereas the ISO 2CR S.T. N°2 profile has a cut-off
of -0.23 µm at 50%, i.e. 6 times less than the ISO
2CR S.T. N°1 filter values. [9]
In conclusion, the overshoot phenomenon augments
the roughness profile in such a way that it does not
comply with an acceptable surface finishing (ST1).
On the contrary, the shift phase will not have any
overshoot effect on a perfect primary profile (ST2). % Specimen test ST2 ISO2CR / Gaussian

3.4.2 Influence of the Gaussian filter Measuring numbers

3.4.2.1 Measurement on a scratched
standard
In order to highlight the reaction generated
by the Gaussian filter, measurements on the same
scratched standard than for ISO2CR (scratch depth
of 8,6 μm) have been carried out. Fig. 18 shows the
standard measured with the Gaussian filter: Figure 19: Evolution of the bearing rate with
ISO 2CR and Gaussian filters, ST1 & ST2

Regarding the specimen test ST1 and

measured with the Gaussian filters (cf. Fig. 18), all
parts comply with the top quarter, whereas the parts
measured with the ISO 2CR filter are located in the
bottom half of the graph.
 Regarding the specimen test ST2 and
measured with the Gaussian and ISO 2CR filters, all
results are on the top compliant.
To better understand the difference between the two
filters, the roughness profiles of the ISO 2CR S.T.
n°1 and Gaussian S.T. n°1 filters on the part n°8, are
presented (case for which the bearing rate is
minimum).

µm

mm µm
m

Figure 20: Roughness profile from the ISO 2CR

and Gaussian filters Figure 22: ABBOTT curve of ST1 ST2 with filter
The circled peak represents the overshoot directly ISO 2CR and Gaussian
linked to the shift phase of the weighing function
The ISO 2CR ST1 roughness profile from
average line of the ISO 2CR filter. The Gaussian
the ABBOTT curve has a cut-off of -1.4µm at 50%
filter gives a valley of a 1.5 µm depth. Fig. 21 shows
whereas the Gaussian ST1 profile has a cut-off of -
the behaviour of the filters average lines on the
0.50 µm at 50%, i.e. 3 times less than the ISO 2CR
primary filter in the circled zone.
filter values.
On the contrary, the ISO 2CR ST2 roughness profile
from the ABBOTT curve has the same value than
the Gaussian filter.
In addition, in order to complete the study, a
gear teeth and race way surface roughness data
base was created, for each family gear, depending
on the material, using several roughness filters and
another surface criteria characterization, in
accordance with features requirements.

Figure 21: Primary profile + average lines of the

ISO 2CR and Gaussian filters
3.5.3 Analytical relationship between the
The position of the Gaussian filter average line is roughness criteria of the ISO 2CR
centred on the primary profile, raising the value of and Gaussian filters
the optimal roughness profile. The ISO 2CR filter
The bearing rate corresponds to the ABBOTT
average line is shifted compared to the primary
curve, which represents the ratio between the
profile, which gives an incorrect result compared to
roughness profile bearing length at a given level zc
the norm. A representation of the ABBOTT curves
(according to a reference zref) and its total bearing
exemplifying the roughness profile bearing curve is
length.
given in Fig. 22.
 4 PROCESS IMPROVEMENT
The second part of the project has
consisted in identifying the technologies capable
of ensuring a perfect mastering of the high quality
finishing process. To do so, an optimizing test
plan has been set up. These tests aimed to
assess the process capability:
- Consolidating, in manufacturing, the
quality of roughness criteria (as set by
the design specification) on the internal
roller raceways and external roller and
ball bearing raceways with or without
collar, and on the active flank of the gear
teeth.
- Unimpacting on the shape and the
circulars waving criteria of the different
When using an ABBOTT curve, some information
raceways.
may be lost (modified by the filtering) which does not
allow a generic modelling of the Tp estimator from
4.1 Technologies Comparisons
the Gaussian and the ISO 2CR filtering.
Different technologies have been compared:
A set of experiments can be carried out to determine
an optimisation of the Smuritropy process and the
the practical relations that would gather the Tp
Honing.
values obtained with a Gaussian and RC filter. It is
however impossible to ensure such relationships.
The phase shift with the ISO 2CR and Gaussian 4.1.1 New Media Evaluation
filters makes the setting of an analytical or empirical Surface finishing is currently obtained by a
bridge random. Each roughness profile comes from superfinition process: the Smuritropy.
a primary profile, which directly depends on the The functioning principle is as follows: the
machining procedures [10]. gears are fitted on a rotating spindle. They are
Changing the roughness filter therefore means immersed into a circular tank filled with abrasive
developing a new roughness characterization value, media. In addition, the gears turn around the vertical
based on the Gaussian filter, tested and qualified axis of the tank. (cf Fig 23). During the abrasion
during the new gear boxes development. In a cycle, the media is watered with water and additive
second time, a new surface roughness liquid (e.g . soap, antioxidant). At the part/media
characterization criteria will be proposed, which will contact level, the exerted pressure, associated with
bring a complementary characterization of the gears the media relative flux, creates an abrasion which
teeth roughness. improves surface finishing.

Figure 23: Smuritropy Machine / Media

In order to augment the abrasion effectiveness, the

media densities as well as its „superfinishing‟
capacity have been augmented. The first tests
allowed selecting the media composition,
dimensions and shapes as required for our study.
(cf. Fig. 24 & 25).

The tests have been carried out on spiral bevel

gears and spur gears with integrated raceways.

Figure 26: Evaluation of ‘Mixed Media 7’ on MGB

Gear range

The obtained roughness on the raceways is Ra

0.05µ for the roller raceways and Ra 0.07 for the ball
bearing.

Figure 24: Comparison of different mixed media The new media mixed fully meets the
on MGB Spiral bevel Pinion roughness requirements.

4.1.3 Honing Process Evaluation

In addition, we have assessed the honing process

which enables to reach the quality, shape and
surface requirements of the bearing raceways.
The issue is to use superfinishing with abrasive
stone or grinding belt machines on complex gears,
which have different shapes and sizes.

Figure 25: Comparison of different mixed media Circular Oscillator

on MGB Spiral, & Spur Gear Part

The best compromise between the spur and

spiral bevel gear teeth is the « mixed n°7 ».

The Mixed media n°7 is then tested on a set of gear

families. The graph on Fig. 26 shows that the Rolls
Gaussian filter measurement is on the top quarter.
(Except for the two antepenultimate cases for which
superfinishing time was weak). Figure 27: Race Way Honing Process
The tests on the parts have shown that: - The material removed by honing was around 2µm
on the radius.
- First, the internal and external roller
raceways are easily super finished. The reached - The numerical control machine allows varying the
roughness (Ra 0,04 µ) is clearly superior to the feedrate along the cylinder generatrix of the
demanded criteria. raceway. This allows impacting the shape (bumped,
concave, inclination).

The process can be used at a high level

performance (quality, cycle time). The flexibility of
the testing machine and the digital control system
are important advantages for our needs (e.g. parts
diversity, parameters mastering), ball raceway
excluded.

Choosing and refining honing conditions

(e.g. tools, speeds) necessitates some know-how. In
addition, this process is very sensitive to the
variations of each part (dimensional, roughness,
material).

Figure 28: Honing Process on a Gear Raceway Three years were necessary to master this
technology. In addition the specific attribute of our
Note that the centring of the part is realised by 2 gears requires honing process adaptation for each
rollers. To ensure a high quality centring, these part as well as modifications of the manufacturing
elements must rely on a precise reference of the operation mode in order to ensure the positioning of
part (e.g. teeth pitch diameter, external way; cf Fig the part (e.g. referential).
28).
In synthesis, the risks to master the
- Secondly, the ball raceways are harder to technology on complex gear and the industrial
superfinish. Besides a significant and stable costs are too high.
improvement of the roughness (Ra 0,04µ), it is Only the honing process is kept for the through
difficult to master the profile shape. After honing, the internal raceways.
circular section extremities have a subsidence of few
microns (about 3µm on the tested parts).
This is due to the increased honing contact pressure
during the output side. The risk of non-compliance is 4.1.4 Synthesis of technologies evaluation
quite high.
To conclude, the new mixed media
Remarks about the process: developed with the Smuritropy process has been
selected. It will be associated to the honing
- The changeover does not last more than one hour process for the internal raceways. The set allows
obtaining the roughness and surface finishing
and includes changing the stone clamping, adjusting
quality consolidated robustness.
the centring and position keeping and honing.

- In general, two cycles are necessary: Rough cycle

(high pressure, quick beat, low rotation and 4.2 Serial Manufacturing Implementation
vibration), finishing cycle (low pressure, low beat,
quick rotation and vibration). For our tests, the The new media mix has been integrated in the
same hone was used, but sometimes with two manufacturing at the beginning of 2012. The new
different grains. process qualification program report is not described
in this document.
Evenly, this process has required a specific
development in order to avoid impact on the etching
inspection and conformity of the gears.
To take into account the environmental needs,
a power station of filtration in closed circuit allows to
treat the waste and to avoid consequent wasting
water.

4.2.1 New Media Implementation

The performances obtained in serial

manufacturing are superior to those obtained during
the tests. The inner form of the tank (ring shape)
would increase the pressure of the media on the
parts.
st
During the 1 semester of 2012, 305 batches, 130
gear teeth and 180 raceways were controlled. Figure 30: Raceway Roughness result with New
Process

Results are very satisfactory. The

Superfinishing process robustness allows increasing
the race way and gear teeth surface roughness in
grinding process, decreasing burns and cracks risks.
In addition, new grinding wheel and specific
parameters give the right robustness for bearing
race way grinding process. Evenly, a thesis is
running on the burns and cracks phenomena
apparition understanding.

Regarding the results, all Smuritropy

working time has been divided by two.

Evenly, the target is to verify that the design

requirements are well respected, as dimensional
and geometric, surface aspect, right tooth profile.
Particularly, the non-destructive test „ NITWATER‟
Figure 29: Teeth Roughness result with New has been checked, to guaranty that changing the
Process media did not impact the surface preparation prior to
this test.
Fig. 29 shows that 82% of the results are in
the top quarter and 100% in the upper half.

Tooth profile Impact evaluation

Fig. 30 shows that more than 85% of the
roughness on raceways is in the top quarter, and To ensure that there is no impact on the
99,5 % are robust. tooth profile, some tests were carried out in serial
conditions, but increased by a coefficient of ten.
The results show a slight degradation of 4 µ (cf. Fig
31).
 Some tests were produced in serial conditions, also
increased by a coefficient of ten.

Removal
µ

2,5 5 7,5 10 Work Time coef

Figure 33: Removal result on raceway / work

time coefficient.

The results in Fig. 33 show a slight degradation of 5

µ of diameter dimension for a work time increased
Figure 31: Teeth defect result with Cycle x 10 by ten. The measurement values are similar
concerning the „2 and 4 coefficient ‟. This could be
The 4 µ defect is located on the middle of due to the dispersion of temperature measurement.
the profile, on the right side. The follow-up of manufacturing batch, including
different gear families, in serial conditions, has never
shown any impact on dimension.

4.2.2 Impact on NITWATER

A qualification plan has been realised in first half of

2012, to evaluate the impact of new process on the
surface preparation prior to the NITWATER.
Effectively, the teeth and raceway
roughness has substantially improved, changing the
process (cf. Fig 34).

Figure 32: Teeth defect result of Cycle x 10

This study shows that in the case of a

particular gear, In fact, the contact pressure between
the tooth and the média. (cf. Fig 32). A local wear
appear after a Smuritropy consequent work time
(factor 10).

The follow-up of more than fifty

manufacturing batches, including different gear
families, in serial conditions, has never shown local
defect.

Material removing Impact evaluation

A similar study has been carried out to ensure that Figure 34: 2D & 3D roughness on sample before
there is no impact on the raceway diameter value. and after changing process
 The follow-up of more than 150 serial [6] Norme ISO 3274. 1996. Surface finishing :
manufacturing batches (1160 parts) shows that the Profile method – Nominal Characteristics of touch
surface preparation was not impacted by the new probe. Edition AFNOR.
roughness.
[7] RODRIGUEZ. 2006. The means Capability. <
http://www.ac-nancy-metz.fr/pres-
etab/juliedaubierombas/Disciplines/productique/Qualite/
Les%20capabilites.pdf >.
5 CONCLUSIONS
[8] Norme ISO 1156. 1998. Surface finishing : Profile
The defined specifications for roughness method – Metrological characteristics of the right
robustness are completely reached, including phase filter. Edition AFNOR.
reducing working time by a factor of 2. This
results in reducing the recurring costs, [9] TURKY, W. 2011. Implementation of a
consumable costs and in respecting the roughness filter. EUROCOPTER / Arts et Métier
environment. ParisTech, Aix-en-Provence. France. Unpublished
Changing the roughness filter therefore
means developing a new roughness [10] ARANDA, S. 2010. Roughness estimator
characterization value, based on the ‘‘Gaussian’’ comparison calculated with Gaussian and RC Filter.
filter, tested and qualified during the new gear EUROCOPTER / EA(MS)² . University of
Mediterranean. Aix-en-Provence. France.
boxes development.
Unpublished

Bibliography
Copyright Statement
The author confirms that the company
[1] RAPHET, B. 2008. Surface finishing -
EUROCOPTER holds copyright on all of the original
Characterizations. Technical of the Engineer.
material included in this paper. The author also
[2] Norme ISO 4288. 1997. Surface finishing : confirm that he has obtained permission, from the
Profile method – Terms, definitions et parameters of copyright holder of any third party material included
a Surface finishing . Edition AFNOR in this paper, to publish it as part of their paper. The
author confirms that he has obtained permission
[3] RAJA, J. MURALIKRISHNAN, B. SHENGYU, from the copyright holder of this paper, for the
FU. 2001. Recent advances in separation of publication and distribution of this paper as part of
roughness, waviness and form. Journal the the ERF2013 proceedings or as individual offprints
international Societies for Precision Engineering and from the proceedings and for inclusion in a freely
Nanotechnology. Pages 222-235. accessible web-based repository.

[4] TOMOV, M. KUZINOVSKI, M. 2010. Function

on Gaussian and 2RC Filters to Determine the
Roughness Profile in Real Non-periodic and Periodic
th
Surfaces. 14 International Research/ Expert,
Conference “Trends in the Development of
Machinery and Associated Technology”

[5] PEYRON, G. HAMPARSOUIMIAN, S. 2004.

Theoretical explanation of GAUSSIEN and ISO 2CR
filter. EUROCOPTER / LTDS INSA De Lyon.
France. Unpublished