Industrial Metrology

Springer-Verlag London Ltd. http://www.springer.de/phys/

Previous books for Springer-Verlag by the author

Advanced Machining: The Handbook of Cutting Technology (1989)

CNC Machining Technology series:

Book 1: Design, Development and CIM Strategies

Book 2: Cutting, Fluids and Workholding Technologies
Book 3: Part Programming Techniques (1993)

CNC Machining Technology: Library Edition (1993)

Graham T. Smith

Industrial
Metrology
Surfaces and Roundness

, Springer
Graham T. Smith, MPhil (Brunel), PhD (Birmingham), CEng, FIMechE, flEE
Faculty of Technology
Southampton Institute
East Park Terrace
Southampton S014 ORD
U.K.

British Library Cataloguing in Publication Data

Smith, Graham T., 1947-
Industrial metrology: surfaces and roundness
1. Mensuration 2. Surfaces (Technology) 3. Roundness measurement
1.Title
620'0044

Library of Congress Cataloging-in-Publication Data

Smith, Graham T., 1947-
Industrial metrology: surfaces and roundness/Graham T. Smith.
p.cm
Includes bibliographical references and index.

1. Surfaces (Technology) - Measurement. 2. Roundness measurement. 1.Title

TA418.7.S55 2001
620',44-dc 21001032028

Apart from any fair dealing for the purposes of research or private study, or criticism or review, as
permitted under the Copyright. Designs and Patents Act 1988, this publication may only be
reproduced, stored or transmitted, in any form or by any means, with the prior permission in writing
of the publishers, or in the case of reprographic reproduction in accordance with the terms of licences
issued by the Copyright Licensing Agency. Enquiries concerning reproduction outside those terms
should be sent to the publishers.

ISBN 978-1-84996-878-2 ISBN 978-1-4471-3814-3 (eBook)

DOI 10.1007/978-1-4471-3814-3
http ://www.springer.co.uk

© Springer-Verlag London 2002

Originally published by Springer-Verlag London Berlin Heidelberg in 2002
Softcover reprint of the hardcover 1st edition 2002

The use of registered names, trademarks etc. in this publication does not imply,even in the absence of
a specific statement, that such names are exempt from the relevant laws and regulations and therefore
free for general use.

The publisher makes no representation, express or implied, with regard to the accuracy of the
information contained in this book and cannot accept any legal responsibility or liability for any errors
or omissions that may be made.

Typeset by Florence Production Ltd, Stoodleigh, Devon

69/3830-543210 Printed on acid-free paper SPIN 10835643
To my wife Brenda, who has supported me throughout our married and
my academic life, with much love and affection.
Foreword

The subject of this book is surface metrology, in particular two major aspects:
surface texture and roundness. It has taken a long time for manufacturing engineers
and designers to realise the usefulness of these features in quality of conformance
and quality of design. Unfortunately this awareness has come at a time when
engineers versed in the use and specification of surfaces are at a premium.
Traditionally surface metrology usage has been dictated by engineers who have
served long and demanding apprenticeships, usually in parallel with studies leading
to technician-level qualifications. Such people understood the processes and the
achievable accuracies of machine tools, thereby enabling them to match production
capability with design requirements. This synergy, has been made possible by the
understanding of adherence to careful metrological procedures and a detailed
knowledge of surface measuring instruments and their operation, in addition to
wider inspection room techniques.
With the demise in the UK of polytechnics and technical colleges, this source of
skilled technicians has all but dried up. The shortfall has been made up of semi-
skilled craftsmen, or inexperienced graduates who cannot be expected to satisfy
tradition al or new technology needs. Miniaturisation, for example, has had a pro-
found effect. Engineering parts are now routinely being made with nanometre
surface texture and fiatness. At these molecular and atomic scales, the engineer has
to be a physicist.
This book is intended to bridge the gap between technology and training. Dr Smith
has approached the subject from the point of view of aprecision engineer, but in
addition has made the reader aware, whenever appropriate, of new techniques and
instruments. He is admirably suited to the task, having served an apprenticeship in
industry which has led to hirn becoming a master toolmaker. He has balanced these
practical skills by in-depth studies leading to the degree of Ph.D in manufacturing
engineering. Taking these complementary paths has given hirn a unique pedigree
that has enabled hirn to write with authority.
Dr Smith is recognised as an expert in his subject of manufacturing and has lec-
tured extensively throughout Great Britain and North America. This book is a must
for those involved in inspection and quality control and could well be used as a co re
for training courses in metrology.

Professor David]. Whitehouse

Formerly Chief Scientist,
School of Engineering,
University of Warwiek
Preface

As the UK's national measurement institute the National Physical Laboratory (NPL)
has been at the forefront of research in the field of metrology - measurement
science - since its inception in 1900. NPL has always been involved in measurements
at the highest levels of accuracy in order to suppport its role as custodian of the
national measurement standards and to address the UK industry's most demanding
measurement problems. NPL also plays a key role in the dissemination of mea-
surement good practice to industry through its wide range of knowledge transfer
activities. It was therefore a pleasure for myself and members of the Dimensional
and Optical Metrology Team at NPL to contribute to this important book, which I
believe provides a rich source of information for anyone working in the field of
surface texture and roundness - areas which have been and will continue to be
among the most important but sometimes least appreciated areas of dimensional
metrology in industry.

A measure of the importance of surface texture and roundness can be gauged by

the work carried out at NPL in this area over the past 70 years or so. As early as
the 1930s NPL was increasingly involved in developing instrumentation for surface
texture and roundness metrology. In 1940, NPL was working on the quest for a
single number or parameter that would define the texture of a surface and hence
enable comparisons between different surfaces to be made. Around this time NPL
was also producing standard artefacts for calibrating surface and roundness instru-
mentation.

In the 1970s much of the surface texture work carried out at NPL related to the
manufacture of X-ray optical components that required control of roundness to
within a few tens of nanometers, profile to a few nanometres and surface text ure
often to sub-nanometres tolerances. In order to control the manufacture of these
optical components, measurement techniques with sufficient accuracy had to be
developed. This led to an NPL-designed instument, subsequently marketed by
Rank Taylor Hobson, called "Nanostep", which was the first instrument of its time
to have sub-nanometre vertical accuracy with a large horizontal movement range.

In recent years one of NPL's key objectives has been to undertake research work
to support UK industry by providing calibrated references artefacts for surface
texture measurements. This work culminated in 2000 in a new NPL-designed
instrument, "NanoSurf IV", capable of calibrating surface texture reference arte-
facts with fully traceable measurements in all the axes, leading to a combined
expanded uncertainty of ± 1.3 nm.

Roundness measurement was revolutionised by the introduction of the "Talyrond"

in 1949 and has progressed since then. More recently Taylor Hobson and NPL have
• Preface

also developed a roundness-measuring facility that has a measurement capability

of 5 nm, making it undoubtedly among the most accurate instruments of its type.

A view 0/ NPL's new laboratory

complex

With the ever-increasing im-

portance of surface measure-
ments to industries operating in
diverse areas ranging from aero-
space to automotive and from
medical to environmental, it
should be no surprise to leam
that NPL plans to continue to
strive for further improvements
in surface measurements in its
future work programmes. It is
expected that this work will be
helped by the imminent move
of the whole of the National
Physical Laboratory into a completely new purpose-built laboratory complex. The
move is expected to be completed during 2002 and will provide NPL scientists with
new and improved facilities to enable them to provide the underpinning research
and traceable calibration services required by industry in this important field weIl
into the new century.

Professor Graham N. Peggs

Science Leader, Dimensional and Optical
Metrology Team
Centre for Basic, Thermal and Length Metrology
NPL
National Physical Laboratory
Contents

1. Surface texture: two-dimensional ............................... 1

1.1 Introduction ........................................... 3
1.2 Establishing the Ra numerical value of surface texture from the
production process ...................................... 8
1.3 Surface texture roughness comparison blocks and precision
reference specimens ..................................... 8
1.4 The basic operating principle of the pick-up, its stylus and skid . . . . 9
1.5 Filters and cut-off ....................................... 13
1.6 Measuring lengths ....................................... 15
1.7 Filtering effects (As, Ac and Aj) ............................. 19
1.8 Geometrical parameters .................................. 21
1.9 Surface profile parameters ................................. 22
1.9.1 Amplitude parameters (peak-to-valley) ................. 23
1.9.2 Amplitude parameters (average of ordinates) ............ 24
1.9.3 Spacing parameters ................................. 29
1.9.4 Hybrid parameters ................................. 29
1.9.5 Curves and related parameters ........................ 29
1.9.6 Overview of parameters ............................. 32
1.10 Auto-correlation function ................................. 32
1.11 Appearance of peaks and valleys ........................... 34
1.12 Stylus-based and non-contact systems ....................... 37
1.12.1 Pick-up .......................................... 40
1.12.2 Skid or pick-up operation ............................ 43
1.12.3 Portable surface texture instruments ................... 45
1.12.4 Surface form measurement ........................... 46
1.12.5 Non-contact systems ................................ 50
1.13 Nanotopographic instruments ............................. 63

2. Surface texture: three-dimensional ............................. 69

2.1 Introduction ........................................... 71
2.1.1 Stylus speed and dynamics ........................... 72
2.1.2 Envelope and mean systems .......................... 74
2.1.3 Three-dimensional characterisation .................... 74
2.2 Three-dimensional analysis software ........................ 79
2.2.1 Functional3-D performance ......................... 85
2.3 Portable three-dimensional measuring instruments ............. 89
2.4 Fractal techniques ....................................... 90
.. 2.4.1 Topologieal characterisation .......................... 90
2.5 Textured metal sheets .................................... 93
2.6 Surface topography characterisation by neural networks ......... 94
2.7 Non-contact measurement ................................ 96

3. Surface microscopy .......................................... 10 1

3.1 Introduction ........................................... 103
3.2 Scanning electron mieroscope .............................. 105
3.2.1 Energy-dispersive X-ray spectrometer .................. 112
3.2.2 Transmitted electron image .......................... 114
3.3 Transmission electron mieroscope .......................... 114
3.3.1 Transmission electron mieroscopy: general applieation ..... 119
3.4 Atomic force mieroscope .................................. 119
3.4.1 Criteria for using scanning probe microscopes ........... 121
3.4.2 Atomie force mieroscope: operating principle ............ 123
3.4.3 Atomie force/scanning probe mieroscope: applieations ..... 125
3.4.4 Ultrasonie force mieroscope: developments .............. 128
3.5 X-ray photoelectron spectroscopy .. . . . . . . . . . . . . . . . . . . . . . . . .. 130

4. Roundness and cylindricity ................................... 135

4.1 Introduction ........................................... 137
4.1.1 Roundness measurement: basie approach ............... 143
4.2 Roundness: measuring instruments ......................... 144
4.2.1 Types of instrument ................................ 145
4.2.2 Spindle and bearings ............................... 146
4.3 Methods of measurement ................................. 150
4.3.1 Assessment of part geometry ......................... 152
4.4 Display and interpretation ................................ 156
4.5 Roundness measurement from the display .................... 157
4.5.1 Roundness reference circles .......................... 159
4.5.2 Numerical value of roundness ........................ 160
4.5.3 Filtering and harmonies ............................. 162
4.6 Geometrie roundness parameters ........................... 171
4.6.1 Cylindricity....................................... 173
4.6.2 Cylindricity measurement techniques .................. 176
4.6.3 Cylindrieal measurement problems .................... 177
4.7 Non-contact spherical and roundness assessment .............. 178
4.7.1 Sphericity interferometer ............................ 178
4.7.2 Spherieal and roundness assessment by
error separation ................................... 180

5. Machined surface integrity .................................... 185

5.1 Introduction ........................................... 187
5.2 The machined surface .................................... 189
5.2.1 Residual stresses in machined surfaces ................. 192
5.2.2 Tribological cutting effect on surface ................... 196
 5.2.3 Micro-hardness testing ..............................
..198
5.2.4 Surface cracks and "white layers" ...................... 203
5.2.5 Machined surface topography . . . . . . . . . . . . . . . . . . . . . . . .. 211
5.2.6 Machined roundness ................................ 220
5.2.7 Power spectrum analysis of machined surfaces ........... 234
5.2.8 Manufacturing process envelopes ...................... 236
5.3 Surface engineering ...................................... 243

6. Quality and calibration techniques ............................. 249

6.1 Size and scale .......................................... 252
6.2 Predictable accuracy: its evolution .......................... 253
6.3 Traceability of measurement ............................... 255
6.4 Measurement uncertainty ................................. 262
6.5 Calibration: surface texture ................................ 274
6.5.1 Surface texture artefacts ............................. 274
6.5.2 Stylus damage ..................................... 281
6.6 Calibration: roundness ................................... 282
6.7 Probing uncertainty: roundness and form .................... 288
6.8 Nanotechnology instrumentation: now and in the future ......... 289

Appendices. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 297
Appendix A - Previous and some current surface texture parameters 299
Appendix B - Amplitude-wavelength analysis: "Stedman dia grams" 317
Appendix C - Surface texture and roundness: calibration diagrams
and photographs .................................. 320
Appendix D - Hardness conversion chart .......................... 325

Index ....................................................... 327

Acknowledgements

A book that leans heavily on current industrial practices cannot be successful

without the proactive support ofboth relevant industrial companies and research-
based organisations, to whom I am particularly indebted. It would be impossible to
name all of the people who have contributed to this book, and if I do not thank you
in name then please accept my apologies. However, certain companies have
provided unstinting support to me in this work and I would like first to thank Taylor
Hobson Limited for their major support of information on surface texture
and roundness instrumentation, Jeol (UK) Limited for information on surface
microscopy equipment and, likewise, Nikon (UK) Limited, VG Scientific Limited,
Zygo Photometrics Limited and Rubert & Co. Limited, for additional information
on their products. The National Physical Laboratory must be singled out for their
significant contribution to the book; I am particularly indebted to them for infor-
mation on the current standards.

First and foremost, I would like to offer my sincere gratitude to Dr Richard K.

Leach at the National Physical Laboratory, who read the appropriate seetions on
surfaces and standards and suggested significant ways to improve the text. Also
from the National Physical Laboratory, I would like to show my appreciation to
Professor Graham N. Peggs for agreeing to write the Preface, to Mr David Flack,
who read the roundness seetions and supplied considerable information on round-
ness standards, and to Dr Bob Angus, who along with Dr Andrew Lewis has shown
significant enthusiasm for this project.

At Taylor Hobson Limited, I am truly indebted to Dr Paul Scott, who supplied me

with information on current working standards and for numerous discussions on
their current instrumentation. Similarly, I must thank Mr Pat Kilbane (Manager)
and his calibration staff in the UKAS Calibration Laboratory for information on
certain calibration procedures. Other notable Taylor Hobson personnel that I
would like to express my thanks to include Mr Darian P.L. Mauger (Manager) in
their Cent re of Excellence, Mr Mick Garton, Mr Jon Gardiner and Mr Steve Foster
(Technical Liaison Engineers); to Dr Mike Hills (Chief Engineer) and from the Sales
and Marketing Department Mr Julian Shaw (Director of Sales and Marketing), Ms
Louise O'Reilly (formerly Marketing Manager), Mr Peter Atkinson (Sales Executive
Technicai); and to Mr Bruce P. Wilson (Managing Director) and to all the Taylor
Hobson staff I have neglected to mention.

Jeol (UK) Limited were of considerable help and support in supplying informa-
tion and producing photomicrographs for the chapter on surface microscopy. Most
notably, I would like to express my thanks to Dr Larry Stoter (Sales Executive) for
supplying instrumentation photographs and relevant information, Mr John
Critchall (Applications Manager) and Mr Andy Yarwood (Application Specialist)
111
for undertaking SEM/AFM photomicrographs and to Mr Chris Walker (General
Manager) for further technical assistance.

Other notable support came from companies that have contributed information
and photographic support to the book, including Zygo/Lamda Photometrics
Limited - Mr Joe Armstrong (Technical Sales Engineer); VG scientific - Dr John
Wolstenholme (Marketing Manager): Nikon UK Limited - Mr Robert Forster
(General Manager) and Keith Poulton (Product Manager); also Rubert & Co.
Limited - Mr Paul Rubert (Managing Director).

This book would not have been possible without the help and guidance of some
of Great Britain's leading academics and I would like to single out several people
for special mention. In the first instance, an old friend Professor David J.
Whitehouse (formerly Chief Scientist at the University of Warwick). For conver-
sations on David's immense knowledge on the topics discussed here and for
previous instruction on all manner of aspects concerning both surface texture and
roundness, furthermore, I would like to express my sincere thanks for hirn kindly
agreeing to write the Foreword for this book. Likewise to Dr Brian J. Griffiths
(Reader at Brunel University), my previous mentor and dear friend, for his instruc-
tion and support on all topics appertaining to surface integrity; and to Professor
Tom R. Thomas (Chalmers University, Sweden) for conversations concerning
surface texture and amplitude-wavelength analysis.

Numerous people at my publishers Springer-Verlag have been of great support in

this current work, most notably Ms Francesca War ren (Editorial Assistant), Mrs
Beverly Ford (Editorial Director), Mr Roger Dobbing (Production Manager) and
Mr David S. Anderson (Sales and Business Development Director). Finally, I would
like to acknowledge a long and interesting acquaintanceship with Mr Nick Pinfield
(fomerly of Springer-Verlag), who gave me considerable encouragement during the
period of the book's development and for some interesting conversations ranging
from engineering to historical topics.

Lastly, a considerable number of graphs, line and assembly diagrams were drawn
by me, and if there are any misinterpretations in their execution then they are a
reflection on the author. Further, in the body of the text some manuscript errors
may have inadvertently crept in, and if this is the case they are not to be associ-
ated with any of the companies or individuals that have supported the book.

Dr Graham T. Smith
West End,
Southampton
 Surface texture:
two-dimensional

"Errors, like straws, upon the surface flow;

He who would search for pearls must dive below."
(All Jor Love, Prologue, 25; John Dryden 1631-1700)

G. T. Smith, Industrial Metrology

© Springer-Verlag London 2002
 Surface texture: two-dimensional

Surface texture: an overview 1.1 Introduction

The concept of surface technology has previously
varied according to whether a scientist, technologist, On the macro scale the natural world mlmlCS
or engineer has inspected a surface. An individual's surfaces found in engineering. Typieal of these is
viewpoint differs depending upon their specific that shown in Figure 1, where the desert sand
interest and technical emphasis. Therefore, a surface is comprised of sand grains - roughness; ripples in
cannot be thought of in isolation from other the surface - waviness; together with the undulating
features, because each is valid and necessary and nature of the land - profile. From an engineering
needs to be considered to gain an overall impression point of view, surfaces are boundaries between two
of its industrial, or scientific usage. distinct media, namely, the component and its
Surface technology encompasses a wide range of working environment. When a designer develops a
disciplines that includes: metrology; metallurgy; feature for apart - whether on a computer-aided
materials science; physics; chemistry; tribology and design (CAD) system, or drawing board - neat lines
mechanical design. Moreover, it has become appar- delineate the desired surface condition, which is
ent that this topic, and associated characteristics that further specified by its specific geometrie tolerance.
relate to surface technology needs to be recognised In reality, this theoretical surface condition cannot
as a subject in its own right. exist, as it results from process-induced surface
Surface technology provides an important and texture modifications. Regardless of the method of
valuable insight into the practieal and theoretical manufacture, an engineering surface must have
applications of a manufactured surface, most not-
ably for the following reasons:

• the surface is the final link from the original

design concept through to its manufacture, with
the industrial engineer being the last person to
add value to the product prior to shipment;
• industry is continually attempting to improve
component power-to-weight ratios, drive down
manufacturing costs and find efficient ways of
either producing or improving surfaces - this has
become an established goal of world-class
competing companies today;
• internationally, there is growing demand to
recognise the legal implieations of a product's
performance. Product liability is directly related
to the production process technique and, hence,
to a surface's quality;
• technical data that in the past was considered
valid, has of late been found to be either unreli-
able or may give insufficient detailed information
on surface phenomena. By way of an example,
originally it was thought that the surface tex-
ture was directly related to a component's fatigue
life; however, it has been shown that surface
topography must be extended to include sub-
surface metallurgical transformations, namely
surface integrity;
• the production process selected for the manu-
facture of apart has been shown to have some
inftuence on the likely in-service reliability of
components. Figure 1. The natural world - the desert - can exhibit a large-scale
combination of surface characteristics, for instance:
• agglomeration (Le. clusters) of sand - roughness;
• ripples - waviness; and
• the undulating nature of the land - profile.
(Source: James Smith, Prayer in the desert, circa 1920s.)
some form of texture associated with it, this being operational constraints, which might be a require-
a combination of several interrelated factors, such ment for either a "smooth" or "rough" surface. This
as: then begs the question posed many years ago: "How
smooth is smooth?" This is not as superficial a state-
• the influence of the material's microstructurej ment as it might initially seem, because unless we
• the surface generation method, the tool's cutting can quantify a surface accurately, we can only hope
action, tool geometry, cutting speed, feedrate and that it will function correcdy in service. In fact, a
the effect of cutting fluidj surface's texture, as illustrated in Figure 2, is a
• instability during the manufacturing process, complex condition resulting from a combination of:
promoted by induced chatter - poor loop-stiff-
ness between the machine-tool-workpiece sys- • roughness - comprising of irregularities that
tem, or imbalance in a grinding wheelj occur due to the mechanism of the material
• inherent residual stresses within the part, pro- removal production processj tool geometry,
moted by stress patterns causing deformations in wheel grit, or the EDM sparkj
the component. • waviness - that component of the surface texture
upon which roughness is superimposed, resulting
Within the limitations resulting from the com- from factors such as machine or part deflections,
ponent's manufacture, a designer must select a func- vibrations and chatter, material strain and ex-
tional surface condition that will satisfy the traneous effectsj

Waviness spacing

Waviness

Roughness spacing

Profile - long-frequency components:

Waviness - medium-frequency components:

Roughness - short-frequency components:

Figure 2. The major eomponents that constitute a typieal surfoce texture, also exhibiting some degree of direetionality (lay).
 r
Any manufacturing process
r
Material shall be removed
r
Material shall not be removed

6
3

*The outline represents the six surfaces shown on the

3D-representation (not the front to rear surfaces).

KEY
NB: Surfaces exhibiting a defined directionality, or
a =2D parameter 1
"Lay" must be measured at right angles to this lay
b =2D parameter 2
pattern.
c =2D parameter 3
d =process
e =lay pattern
f =allowance
x = not allowed

Figure 3. The composition ofthe complete graphical symbol for surface texture. [50urce: 150 1302: 2001]

• profile - the overall shape of the surface - ignor- directional properties such as feed marks, or
ing roughness and waviness variations - is caused isotropie - devoid of apredominant lay direction.
by errors in machine too1 slideways. The lay of any surface is important when attempting
to characterise its potential functional performance.
Such surface distinctions tend to be qualitative - If the direction of the trace being produced by a
not expressib1e as a number - yet have considerable stylus instrument - re cord of the stylus motion over
practical importance, being an established pro ce- the assessed surface topography - is not taken into
dure that is functionally sound (Figure 3). The account, then a totally misrepresentative reading will
combination of these surface texture conditions, result from an anisotropie surface. This is not the
together with the surface's associated "lay", are ideal- case when measuring an isotropie surface, for
istically shown in Figure 4, where the lay of the example a "multi-directionallay" - as indicated in
surface can be defined as the direction of the domi- Figure 4. Under this isotropie condition, perhaps
nant pattern. Figure 4 also depicts the classification three different measurement paths could be scanned
of lay, which tends to be either anisotropie - having and the worst roughness trace would be utilised to
111 Industrial metrology

TYPE LAY SY 180L An anisotropie surfaee exhibiting direetionallay

Parallel
d= --/----

Perpendicular
d..L I

lAB C D E
Crossed ~ I I I I,
I I I I I
I ' / I 1/ ,
-j I I / ! ..--- - - - -
Multi - directional
I ~.I
~ M-V V

Particulate A B c D E

Figure S. Effect of the relative direction of lay and its associated

Circular
influence on the measurement of profile shape. ((ourtesy of Taylor
Hobson.)

Radial the lay condition would be "crossed" - as shown in

Figure 4. Under this roughness state, it is usual to
measure the surface at 45°, which has the effect of
Figure 4. The indication of surface 'Iay' as denoted on engineering
averaging out the inHuence of the two directions
drawings. [Source: ISO 1302: 2001] imparted by the cross-honing operation. Where
surfaces are devoid of a lay direction - as in the case
of "particulate lay" (see Figure 4) from either shot-
specify the surface. The designer should be aware blasting or the sintering process - then under such
that the lay condition and other information can circumstances a trace will produce the same surface
be incorporated into the surface texture symbol texture reading, irrespective of the direction of
(ISO 1302: 2001) - which is often misused - see measurement.
Figure 3 for the positioning of this data and other The production of certain milled surfaces can
surface texture descriptors. The surface texture sometimes exhibit roughness and waviness condi-
symbol shown in Figure 4, indicated by "surface tions at 90° to each other - this is not the same con-
texture ticks", highlights the potential process- dition as "crossed lay" - because the relative heights
related lay type and associated directionality (ISO and spacings of the two lays differ markedly. Hence,
1302: 2001). the cusp and chatter marks of the milled topography
Returning to the theme of an isotropic lay condi- should be measured in opposing directions. When a
tion, if this surface is assessed at a direction not at turned component has had a "facing-off" operation
right angles to the lay, then a totally unrepresenta- undertaken, and a "circular lay" condition occurs
tive surface profile will result, which is graphically (see Figure 4), then under these circumstances it is
illustrated in Figure 5. As indicated in this figure, as normal to measure the surface texture in a radial
the trace angle departs to greater obliquity from direction, otherwise an inappropriate reading will
condition Athen the surface profile becomes Hat result. Conversely, if a "radiallay" occurs - resulting
once the direction has reached condition E. This from "cylindrically grinding" the end face - it is more
would give a totally false impression of the true usual to measure surface roughness at aseries of
surface topography. If employed in a critical and tangent positions with respect to the circumferential
perhaps highly stressed in-service state, with the direction.When measuring surface texture in the
user thinking that this incorrectly assessed surface majority of practical situations it can be achieved by
was Hat, then, potentially, apremature failure state direct measurement of the profile, positioning the
could arise. Normally, if the production process workpiece in the correct manner to that of the stylus
necessitates, for example, a cross-honing operation, of the surface texture instrument. When the part is
MANUFACTURINGr-__T.-__~__- r__ R_O~U~G_H_N~E~SS~~_a~)I_·n~~_m~__~--~----r-~
PROCESS: 0.005 0.1 I 0.2 I 0.4 I 0.8 I 1.6 I 3.3 I 6.3 ' 12.5 I 25

I
SUPERFINISHING ~///~

LAPPING V7mij"/////ff/~

POLIS HING ~/~W///J

HONING 'i'////- :/'///..'i'///~

GRINDING V//////////.ffff////////.l
,
BORING V////////.ß7////j /'///~
I
TURNING V////////h /'/////////h
I
DRILLING V//P//////-

EXTRUDING v/////7//h':
I
DRAWING V//////////-

MILLING V//////////.// / h

SHAPING V////////7-:/"//h:/"////////fi::

PLANING v///.. '://///7/.ff///....'lWffffA

Ra (Nominal) VALUES ROUGHNESS GRADE

in~m: NUMBER:

50
'"
~
N12

25 ~ N11
"\.
12.5 ~ N10
"-
6.3 \ N9
3.2
'"
~
N8

~
1.6 N7 INCREASING
ROUGHNESS

0.8 \ N6

0.4 '\" N5

0.2 ~
"-
N4

0.1 ~ N3
~
~
0.05 N2

0.025
~ NI

Figure 6. Anticipated process roughness and respective grades. [Source: ISO 1302: 2001.1
either too large or this is impracticable - due to face texture is closely allied to functional perform-
either stylus access to the surface or surface indina- ance, particularly when one or more surfaces are in
tion - then placing a small portable instrument on motion with respect to an adjacent surface. This sug-
the surface can allow a measurement. Occasions arise gests that the smoother the surface the better, but this
when even this is not an option and under these cir- is not necessarily true if the surfaces in question are
cumstances it may be possible to take a "cast replica" required to maintain an efficient lubrication film. The
of the surface, then present this to the instrument. apparent roughness of one of these surfaces with
Some surface texture measuring instrument manu- respect to the other enables it to retain a "holding
facturers offer "replica kits" for just this purpose. It film" in its associated roughness valleys. Yet another
should be remembered that this replica of the surface factor which may limit the designer from selecting a
is an inverted image. smooth surface is related to its production cost (see
Figure 7). If a smooth surface is the requirement then
this takes a significandy longer time to produce than
an apparendy rougher surface, which is exacerbated
1.2 Establishing the Ra if this is allied to very dose dimensional tolerancing.

numerical value of surface

texture from the production 1.3 Surface texture roughness
process comparison blocks and
precision reference specimens
Numerical data (ISO l302: 2001) to define the rough-
ness grade for surface texture has been established
(see Figure 6) which can be related to the method of In order to gain a basic "feel" for actual replica
production. Caution should be made when using surfaces in conjunction with their specific numerical
these values for control of the surface condition, value for surface roughness. Then the use of compar-
because they can misrepresent the actual state of the ison blocks allows an efficient and useful means
surface topography, being based solelyon a derived of interpreting the production process with their
numerical value for height. More will be said on this expected surface condition. A designer (Figure 8a)
later. Moreover, the "N-number" has been used to can select a surface using a visual and tactile method
establish the arithmetic roughness Ra value, this to reflect the required workpiece condition. Corres-
being just one number to cover a spread of potential pondingly, the machine tool operator can utilise
Ra values for that production process. Nevertheless, similar comparison blocks (see Figure 8b) to estab-
this single numerical value has merit, in that it lish the same surface condition, without the need to
defines a "global" roughness (Ra) value and accom- break down the set-up and inspect the part with a
panying "N-roughness grade", that can be used by a suitable surface texture measurement instrument.
designer to specify a particularly desired surface This is a low-cost solution to on-machine inspection
condition being related to the manufacturing pro- and additionally enables each operator to simply act
cess. The spread of the roughness for a specific as an initial source for adecision on the part's
production process has been established from surface text ure condition. Such basic visual and
experimental data over the years - covering the tactile inspection might at first glance seem in-
maximum expected variance - which can be modi- appropriate for any form of inspection today, but
fied depending upon whether a fine, medium or within its limitations it offers a quick solution to
coarse surface texture is required. Due to the surface roughness assessment. Tactile and visual
variability in the production process and its stoch- sensing is very sensitive to minute changes in the
astic output, such surface texture values do not surface roughness, but is purely subjective in nature
reflect the likely in-service performance of the part. and is open to some degree of variation from one
Neither the surface topography nor its associated person to another. Comparison block surface assess-
integrity has been quantified by assigning to a ment is purely a subjective form of attribute
surface representative numerical parameters. In sampling. Therefore, not only will there be some
many instances, "surface engineering" is utilised to divergence of opinion between two operators on the
enhance specific component in-service conditions, actual surface's numerical Ra value due to their
but more will be said on this subject toward the end differing sensual assessments, but also this is only a
of Chapter 5. very rudimentary form of inspection.
Previously, it was mentioned that with many Precision reference specimens are manufactured
in-service applications that the accompanying sur- to exacting and consistent tolerances, particularly
 24
23
22
21

20

19 \
18 \
17 \ .-/
Cylindrical grinding

.....
<Il 16
y- ./

U
0

:... 15 \
0
ar 14 \\
.5E-< 13
I--Surface grinding
c0
:cu 12 VI\ -----
= 11 \
"I:l
\
0
:...
Q., 10
\ \
~
.::..... 9
\ \ t - - Reamlng
.s~
8 \ \ I\- -\- End milling
Q:i
7
\ 1\ \ \/ Turning
6
1\ \ \ /\ / ' Peripheral milling
5 \ \ /- >(/ Shaping and planin
4
\ 1\ v ~ /"-. r--....
1\ 0 0-. / ' Drilling
3
2
"" ""'" ~ "'" ",'-
r--=:: -'" ~
.........
~
~K
~ ~

0.025 0.05 0.1 0.2 0.4 0.8 1.6 3.2 6.3 12.5 25.0 5.00
Ra Value (11m)

Figure 7. Relative cost, in production time, necessary to produce surface finishes by different production processes. (Courtesy ofTaylor Hobson.)

when used to calibrate stylus-based surface texture reference specimens from frequent use a hard boron
measuring instruments. A range of standard refer- nitride layer can be specified.
ence specimens can be purchased for this purpose
and will act as a "health check" on both the current
stylus condition - whether its point is worn or
partially broken - and indicate if the allied instru- 1.4 The basic operating
mentation/electronics has "drifted" since the last
calibration check. This enables the inspector to principle of the pick-up, its
speedily and efficiently remedy the situation and
bring the instrument back into calibration. Most of
stylus and skid
these calibration reference specimens are manufac-
tured from replicas of an original master production Prior to describing some of the more sophisticated
surface by nickel electroforming; this enables the instruments by which the surface is measured, it is
block to be a reproduction of the whole surface, with worth describing the basic function of a typical
the finest details being approximately 1 fLm in size. stylus instrument. In Figure 9, a schematic illu-
Each reference specimen offers exacting uniformity stration is depicted of the basic components for a
of profile shape. In order to minimise wear on these typical surface texture measuring instrument. The
..
(a)
Industrial metrology

(b)
.

Figure 8. (a) Utilising a comparator gauge to determine the surface finish for a specific manufacturing process. (b) Atypical comparator gauge,
for both visual and tactile assessment. (Courtesy of Rubert & Co Ltd.)

Surface under test

------r----l....--,
Trace Pen
unit
------'---_---l

Recorder

Figure 9. 5chematic diagram illustrating the major constituents of a stylus-type of surface texture measuring instrument. (Courtesy of
Taylor Hobson.)
 u ace texture: two- Imenslona l1li
x5000 and horizontal magnification (Vh ) of xl00,
then in this situation the respective size of the stylus
is almost indistinguishable from that of a (vertical)
straight line or point contact (see Figure llb). This
significant size differential enables the stylus to
penetrate into quite narrow valleys, although its
finite size affects the accuracywith which the surface
profile can be traced. These stylus profile and size
limitations influence the following:

• distortion of peak shape - as the spherical stylus

traverses over a sharp peak, the point of contact
shifts across the stylus - from one side to the
other (Figure llc). This results in the stylus
having to follow a path which is more rounded
Figure 10. Seanning eleetron photomierograph of a stylus (5 fLm than the peak, but due to the stylus being raised
point radius) on a surfaee. [(ourtesy of Hommelwerke GmbH.] to its full height when at the crest the true peak
height is measured;
• penetration into valleys - due to the tip's size, it
may not be able to completely penetrate to deep
stylus traverses across the surface and the transducer and narrow valley bottoms (see Figure 12);
converts its vertical movements into an electrical • re-entrant features cannot be traced - whenever a
signal. This signal is then amplified for subsequent stylus encounters a re-entrant feature (Figure 12
processing, or output to operate a pen recorder. The detail) the stylus tip loses partial contact with the
required parameter value is subsequently derived profile and as a result will remove this feature
from the filtered signal, having previously been dis- from the trace. Typical surfaces that exhibit such
played on screen. In general, the profile is the result re-entrant characteristics are powder metallurgy
of the stylus tracing the movement across the surface (PM) compacts and other porous parts, together
under test, the contacting of consecutive points of the with certain cast irons. This means that mis-
profile being spaced in time. Therefore this relation- leading surface texture parameter measurements
ship between movement and time is closely associ- may occur - although recently DIN 4776: 1988 has
ated with its "cut-off". "Cut-off" refers to the limiting described a method of obtaining"distortion-free"
wavelength at which components of the profile are assessment for PM parts. If transverse sections
passed nominally unchanged by a filter. are taken and positioned within a scanning
The stylus is normally either conical with a spher- electron microscope re-entrant features can be
ical tip (see Figure 10), or a four-sided pyramid with visually revealed.
a truncated flat tip (Figure lla). The conical type of
styli (Figure 10) have a cone angle of either 60° or If the slope of a surface texture feature has a
90°, with a tip radius ranging from less than 0.1 fJvm steeper angle than that of the half angle of the side
to 12 fLm in size. A pyramidal stylus tends to be of a conical stylus tip, then it will lose contact with
approximately 2 fJvm wide in the traverse direction, the profile, having the same effect as a re-entrant
but is normally wider transversely to the direction angle on the resulting measurement. The stylus is
of travel, giving the tip greater strength (see Figure the only active contact between the surface and the
lla left). The sharp-pointed 0.1 fJvm wide pyramidal instrument and it has been shown above that its
stylus (Figure lla right), is utilised on the Taylor shape and size affect profile trace accuracy. If the
Hobson "Talystep" and "Nanostep" instruments. force exerted by the stylus is too great, it may scratch
However, in use the edges of a pyramidal tip tend to or deform the surface, leading to erroneous results.
become rounded with wear; conversely, the spher- Conversely, the stylus force must be sufficient to
ical tip version develops a flat, so the distinction ensure that the stylus maintains continuous contact
between the two profile geometries becomes less with the surface at the adopted traverse speed.
marked with time. If a stylus tip becomes damaged Metrological instrument manufacturers incorporate
during use, this results in a considerable increase in this into the transducer design, with most surface
its width, leading to potentially very serious texture equipment having a tip radius of between
measuring errors. 0.1 and 12 fJvm, with a stylus force of "'0.75 mN.
When a comparison is made of the scale of an This stylus force changes as the tip becomes progres-
actual surface under test to that of the stylus tip sively more blunt as it bears on the surface. An
(Figure 10) against a vertical magnification (V) of instrument with very fine tip radius (Figure 10a
.. Industrial metrology

(a) Typical stylus geometries

= 0. 1 ~ m

(b) Relative size of the stylus against the surfaee topography

(e) The eurve tends to round the peaks (I) and reduee the depth of
the valleys (11), aIthough the peak height is not affeeted (111)

Figure 11. Stylus geometry and its mechanical filtering effect on surface texture. ((ourtesy of Taylor Hobson.)

right) must utilise an exceedingly low force, typically long as the skid's radius is greater than the peak
""'3 x 10- 5 N or less. The transducer assembly can spacing, the apparent line of movement of this skid
incorporate both a stylus and skid, with the skid will be virtually a straight line. Under such surface
offering a local datum for this transducer and with transducer conditions as the vertical skid moves
that of the surface. Moreover, the skid provides a from crest to crest, its relative horizontal skid
local datum for the stylus with respect to its vertical motion with respect to the test surface can be
and horizontal motion. Therefore, with the skid's ignored, as the skid's vertical motion is virtually
large curved radius to that of the relatively small insignificant. However, once the crests in the surface
surface under test, it rides along the surface being under test become more widely spaced, this hori-
measured (Figure 13), providing a "local datum". As zontal crest spacing tends to introduce a significant
 u ace texture: two- Imenslona

___ ______ Path traced by a 2.5 11 nt radiusstylu

- - - - Path traced by a 12.511111 radius stylus

Re-entrallt feature
on a surfacc Figure 14. A high-quality component's surface texture being
assessed without a skid which tends to be the "approved method", of
late. ((ourtesy of Taylor Hobson.)
Figure 12. IIlustrating how the use of a larger-radius stylus reduces
the apparent amplitude of closely spaced irregularities. ((ourtesy of
Taylor Hobson.)
1.5 Filters and cut-off
It was previously suggested that surface geometry is
comprised of roughness, waviness and profile (see
Figure 2). These interrelated factors also tend to have
different relationships to the performance of the
component. It is normally the case that they are
separated out during analysis. To obtain this sepa-
ration on surface texture measuring instruments,
selectable filters for roughness and waviness are
applied. In this section, only a superficial treatment
of filtering and its effects will be mentioned, as this
will be dealt with in greater depth later in the
chapter.
In general, selection of the desired filtering can
be established by the following "rules of thumb":
Figure 13. Surface texture of component being inspected using a
skid. ((ourtesy ofTaylor Hobson.)
• roughness filtering would be applied if control is
required of workpiece performance (for example:
resistance to stress failure; wear resistance; fric-
amount of vertical skid motion. If the skid's vertical tion; reflectivity and lubrication properties);
motion affects the subsequent surface texture • waviness filtering might be selected if control is
results, it is then preferable to use the instrument in necessary for workpiece or machine tool per-
the "skidless mode", as instruments employing a formance (for example, vibration or noise gener-
skid should be used for the measurement of rough- ation).
ness parameters (ISO 3274: 1996). To achieve skid-
less operation and minimise the unwanted affects of Once the type of filter has been chosen then the
vertical skid motion which might otherwise inter- filter cut-offs should be selected (see ISO 11562: 1996
fe re with the results, the skid must be removed, then for more detailed information) with a range, notably:
by utilising the instrument's in-built straight edge as ce••• mm; 0.08 mm; 0.25 mm; 0.8 mm; 2.5 mm;

a "sliding datum" the surface assessment can be 8.0 mm; .. :' (ISO 3274: 1996). However, this can only
undertaken, as illustrated in Figure 14. More will be be undertaken when the specific features needing to
mentioned on the influence of the skid and its asso- be measured are known. Generally, the following
ciated "phase effect" later. criteria can be applied:
.. InClustrial metrology

• a 0.08 mm cut-off length should be selected for

high-quality optical components;
. .

1. One basic feature of stylus-based instruments is

that the profile is traced by movement across the test
• a 0.8 mm cut-off length should be chosen for surface, with the consecutive points of the profile in
general engineering components; contact being spaced in time. Hence, this relation-
• longer cut-offs are necessary when very rough ship between movement and time is closely associ-
components are to be measured, or if surfaces for ated with cut-off.
cosmetic appearance are important. The frequencies present are dependent upon two
factors: traverse speed and spacing irregularities. In
NB A practical guideline is that the cut-off chosen Figure 15 (A and B), the frequencies of 100 Hz and
should be of the order of 10 times the length of the 50 Hz represent irregularity spacings of 0.01 mm
feature spacings under test. If there is any doubt and 0.02 mm respectively. At the same traverse
about the selection of cut-off, initially measure the speed, frequencies of 4.0 Hz and 1.25 Hz would be
surface in an unfiltered manner and then make a generated by spacings of 0.25 mm and 0.8 mm. If a
qualified decision from this test. high-pass electrical filter is used, this will suppress
So far, the term cut-off has been liberally used, any frequency below 4.0 Hz, enabling only those
but what is it? As an example of cut-off and assuming irregularities having spacings of 0.25 mm or less
a perfect filter characteristic, if a high-pass filter has to be represented in the filtered profile. Therefore,
a cut-off of 0.8 mm, this will only allow wavelengths this relationship is ideal to obtain a sampling
below 0.8 mm to be assessed, with wavelengths length of 0.25 mm, hence the term "cut-off" length
above this value being removed. Conversely, when a - since the response now cuts off at irregularity
low-pass filter of 0.8 mm cut -off has been selected, spacings of 0.25 mm, denoted by the international
this will only allow wavelengths above 0.8 mm to be symbol, (Ac) for cut-off length. By introducing
assessed. As previously mentioned (ISO 3274: 1996), different filters, the most suitable cut-off for the
the internationally recognised cut-off lengths for surface can be selected, where Table 1 acts as a guide
surface roughness measurement are given in Table to the selection of suitable cut-offs for various
machining pro ces ses.
Until 1990, standard roughness and waviness
Table 1. Typical cut-offs for variously manufactured surfaces
filters were analogue or 2RC (two resistors and two
capacitors) types. Since 1996, the standard (ISO
Typical finishing process Meter cut-off (mm)
11562: 1996) has specified the digital filter for surface
0.25 0.8 2.5 8.0 25.0 text ure measurements. For example, if a 0.8 mm cut-
off value is utilised, then there is a 50% transmission
Milling x x x of the profile equal to this 0.8 mm. This type of
Boring x x x filtering is analogous to a sieve, only allowing values
Turning x x below the cut-off to be passed through. The main
Grinding x x x problem with utilising an analogue filter is that it
Planing x x x assurnes that the surface is basically sinusoidal in
Reaming x x nature, but this rarely is the case. This mismatch with
Broaching x x the "real" surface under test causes the so-called
Diamond boring x x "Gibb's phenomenon". The Gibb's phenomenon
Diamond turning x x results in an overshooting of the filtered profile
Honing x x when analysing certain extreme forms of profile. The
Lapping x x
Superfinishing x x "Gibb's phenomenon" can be explained by the
Buffing x x following example.
Polishing x x Filtering is used to determine the mean line of the
Shaping x x x roughness profile - being identical to the waviness
Electro-discharge machining x x profile. An analogue filter determines a profile's mean
Burnishing x x line by a "moving average" technique. Hence, at any
Drawing x x point in time the filter averages the profile previously
Extruding x x traversed by the stylus. This has been equated to a
Moulding x x
Electro-polishing x x person walking backwards! Such a filter cannot antic-
ipate what lies ahead of it in the remainder of the
NB The cut-off selected must be one that will give a valid assessment of the profile, such as abrupt changes in the topography.
surface characteristic ofinterest. For example, although a cut-off ofO.8 mm can Under these conditions, it reacts after a significant
validly be used for almost all of these surfaces, it may not necessarily be suit-
able for assessing a particular feature of the texture. This entails examining the
change has occurred; this means that the true profile
surface and considering the purpose of the measurement before selecting the is no longer represented by the mean line, and as a
cut-off. ((ourtesy ofTaylor Hobson.) result the filtered profile will be distorted vertically
 Surface Profile Length traced by Electrical Waveform
Stylus in 1 Second
O.Olmm
A

-1 ~ M
I\/\/\/V\IVVV\,
- I.
100 peaks
~
fV\IV\/V\IVVV\
100Hz

r-
0.02mm
B

~
-1 M ~
~
50 peaks
50Hz

O.01mm
C
-1 ~
I\JVVVVV\/\/'V'
0.5mm
--1 r- ~

~
50 peaks
50Hz

O.Olmm

D
-1 t-
/\/\A/\IVVV\IV ~
I-
2mm
~I
fI/\/INVIJ\/WJV

r 200 peaks
~I 200Hz

Traverse speed: A = 1 rnrn s-l; B = 1 rnrn s-l; C = 0.5 rnrn s-l; D = 2 rnrn s-l;

Figure 15. Signal frequeney depends on irregularity spaeing (compare A and B) and traverse speed (compare A, (and D). ((ourtesy ofTaylor
Hobson.)

and horizontally shiJted. Such an action by the ana- off. Figure 16(bi) shows how the unfiltered profile
logue filter may lead to serious errors Jor either equates to an electrically filtered one - Figure 16(bii)
plateau honed types oJ surJace or PM (porous) types - for the same value of cut-off, but related to 50% of
oJ topographies. its depth. The main difference between the (PC)
If a digital filter is employed, such as that shown digital filter and the analogue filter is that the former
on the graph in Figure 16(a), then the vertical distor- type evaluates the profile mean around a point -
tions of the profile caused by abrupt changes in the enabling it to anticipate changes in the profile height
height of the topography are minimised. The prin- - whereas the latter (analogue) filter cannot. It is
ciple operation of the digital filter can be explained possible to emulate the analogue filter's characteris-
as below. tics by employing a "weighting function" for each
Digital profiles are no longer represented by a ordinate height when adetermination of the arith-
smooth curve, but instead, are described by aseries metic average or mean point is required. However,
oJ ordinates (numbers) relating to profile heights at unless there is a large variation in the surface topog-
regular intervals. In order to estimate the profile's raphy, or special processing conditions apply - such
mean line the filter's action is to calculate the average as the "Gibb's phenomenon" - then the analogue-to-
height at a given point, as the arithmetic average oJ digital differences remain slight.
the ordinate points in its vicinity. Hence, if the profile
is divided up into "windows" the mean profile height
Jor each window can be estimated. A line joins these
window averages together, which represents the 1.6 Measuring lengths
profile mean line.
Such a digital filter allows up to 100% transmis-
sion of the profile, representing wavelengths shorter In order to determine a workpiece's surface texture,
than the actual cut -off length, together with zero three characteristic lengths are associated with the
transmission of wavelengths greater than this cut- profile (ISO 4287: 1997). These are:
 (a) High-pass filter characterisation (Source: ISO 3274: 1996 - abridged graph)

AC = 0.08 0.25 0.8 2.5 8

102
....... ./'
/' / / //
.......
/
8 ~ ........
.......
~

~
6
50% " I'
" '\.
\
, \
"-
\
~

.:=
4

2 \ \ i\. \ 1\
.~'"
'"
= \ \ \ \ \
b
01

'"
"CI
10 1
8
6 \
, 1\ \ \
1\

B \ \ \ \ \
:s.Ei 4
\ \ \ \
-< 2 \ \
1 \ \ \ \ \
10-2 2 4 6 8 10-1 2 4 6 8 10° 2 4 6 8 101 2 4 6 8 102
Sine wavelength (A) mm

(bi) No filtration

Waviness depth: 100%

1I
'-I---------'-V
1 1
I (bii) With filtration

&.
k
1
_

Sine wavelength
~

~
Wavinessdepth:50%

Ac=0.8mm

Figure 16. Effeet of an eleetrieal filter on the surfaee texture. (Courtesy of Taylor Hobson.)

• sampling length; NB If a shorter distance T was selected, the

• evaluation (or assessment) length; roughness patterns at C and D would not be iden-
• traverse length. tical. This would give a misleading picture of their
respective roughness heights.
What was true for the reduced distances also
Sampling length (Ip, !rand /w) applies to waviness, neglecting the roughness (Fig-
ure 17b). This is repetitive over distance U, although
The surface texture profile illustrated in Figure 17(a) if the examination is confined to a shorter distance
has both roughness and waviness present, with both V then the waviness at E and F appear to be
patterns being generally repetitive in nature. markedly different in character. Such a variation
Therefore the roughness pattern A over the distance demonstrates why it is necessary to select the length
L might be typical for the whole length of the of the profile over which a parameter is to be deter-
surface, only differing in minor detail from that mined, termed its "sampling length", to suit the
indicated at B over a similar distance. surface under test. Yet another feature concerning
 (a) Roughness and waviness

---<[Li jL i
~
I I
~
I I

C I A I I I D

I I I
I B I

T I
~ I
~
I I T
~ ko----- u ~~
(b) Waviness only

Figure 17. Effect of different sampling lengths. ((ourtesy ofTaylor Hobson.)

selection of appropriate lengths can be demon- case, the waviness would playa greater role and for
strated in Figure 17(a). Figure17(a) indicates that this reason it is undesirable to increase the sampling
while the sampling length L is sufficient to reveal the length much beyond that required to obtain a repre-
whole of the roughness pattern, the waviness has sentative assessment of roughness.
very little influence over this length. Even if the The sampling length can be defined as the length
sampling length was increased, it would simply in the direction of the x-axis used for identifying the
include more roughness detail but would have litde, irregularities that characterise the profile under eval-
if any, effect on a roughness parameter value. In this uation. By specifying a sampling length, this implies
 InClustrial metrology . .

(a) Traversing unit (i.e. skidless)

Traver ing unit

Material profile

r= Evaluation length (i.e. 5 ampling Icngths) ~

, , ,
I Vß0W&~~ff
~ "'lWßdWZI
Allowance' I 2 3 4 5
L
'I

lor 0.1., " I., "1 cl" c

I
ovcr-Iravcl
Q
run·up
Traver e length
~--- -----------------------~

(b) Derivation of the arithmetic roughness parameter Ra

Figure 18. Determining surface texture analysis roughness parameter Ra, together with its arithmetic derivation.

that structure in the profile occurring over longer and waviness profiles. The evaluation length meas-
lengths will not be applicable in this particular eval- urement is the sum of the individual sampling
uation. The sampling length for Ir - roughness - is lengths. If a different number is used for the rough-
equal (numerically) to the wavelength of the profile ness parameter assessment, ISO 4287: 1997 advo-
AC, whereas the sampling length for lw - waviness - cates that this number be included in the symbol, for
equates to that of the profile filter Af Virtually all the example Ra6. In the case of waviness, no default
parameters necessitate evaluation over the sampling value is recommended. From a practical viewpoint,
length; however, reliability is improved by taking an the selection for the correct filter can normally be
average of them from several sampling lengths as ensured when at least 2.5 times the peak spacing
depicted by the evaluation length in Figure 18(a). occurs, with two peaks and valleys within each one.
Typically, this would me an that an evaluation length
of 0.8 mm would be selected, but there are occasions
Evaluation length (In) when either a larger or smaller evaluation length
might be preferable for the surface under test. The
A simplistic diagram of the relationship of the stylus metrologist's experience and judgement will come
to the material profile/topography for a pre-selected into play here.
traverse length is indicated in Figure 18(a). The eval-
uation length can be defined as the total length in
the x-axis used for the assessment of the profile under Traverse length
evaluation (see Figure 18a). As shown, this length
may include several sampling lengths - typically five The traverse length can be defined as the totallength
- being the normal practice in evaluating roughness of the surface traversed by the stylus in making a
measurement. It will normally be greater than the the total profile, which can be adjudged to be of most
evaluation length, due to the necessity of allowing significance in a particular situation. Conversely,
run-up and over-travel at each end of the evaluation the effects of filtering can be considered as a means
length to ensure that any mechanical and electrical of removing irrelevant information, such as instru-
transients, together with filter edge effects, are ment noise and imperfections. Therefore, filters can
excluded from the measurement (see Figure 18a).On select or reject surface topographical structure
shorter surfaces, it may be necessary to confine the depending upon its scale in the x-axis, in terms
measurements to one sampling length, under these of spatial frequencies, or wavelengths. The terms
circumstances the sampling and evaluation lengths low-pass and high-pass filters refer to whether they
are identical, but the over-travel necessary to either reject or preserve surface data. For example,
contribute to the traverse length must be retained. a low-pass filter will reject short wavelengths while
retaining longer ones, whereas a high-pass filter pre-
serves shorter-wavelength features while rejecting
longer ones. The term bandpass filter refers to the
1.7 Filtering effects (As, Ac combination of a low-/high-pass filter selected to
restrict the range of wavelengths, with both high-
and At) and low-pass regions being rejected. Filter attenua-
tion (rejection) should be somewhat gradual, other-
wise significantly differing results occur from almost
In Figure 19 a flow chart is illustrated showing the identical part surfaces, the exception being when a
manner in which surface assessment occurs, with significant surface feature causes a slight wavelength
the primary profile broken down into filtered shift.
elements to obtain the waviness and roughness The term cut-off refers to the 50% transmission
profiles of the workpiece surface. Algorithms enable (and rejection) wavelength filter, this being specific
suitable characteristic functions and parameters to to the topic of surface texture. In the case of the large
be obtained, scaled to an appropriate size. As has majority of surface texture assessments, it is
been mentioned previously in Section 1.5, filters are suggested that when the width of a specific surface
vital to any form of surface texture analysis. Today, feature is significant, but its size may only be 1% of
most forms of filtering are normally electrical or the overall width, this will make it less important.
computational, although they can occasionally be Under these circumstances that affect feature trans-
mechanical for analysis of the range of structure in mission/rejection, it is suggested that bandpass

Output
scaling

Roughness Profile
algorithm recording
Characteristic
functions
Parameters of
roughness,
waviness
Waviness
and
algorithm
primary
profiles

Primary
profile
algorithm

Figure 19. Aflowehart for surfaee assessment. [Source: 1504287: 1997.]

Transmission % (low-pass) filter, with a cut-off As. The finite size of
a stylus will eliminate any very short wavelengths,
100~--~r-----~~------ __---- which practically is a form of mechanical filtering,
being in the main utilised as adefault for the As filter.
50 As the size of styli will vary and the instrument can
introduce vibration and/or supplementary noise
o into the profile signal (having equivalent wave-
AS AC lengths to that of the stylus dimensions) under these
Wavelength circumstances, it is recommended to ignore any As
filtration along the total profile.
(a)

ZtL Roughness profile

The roughness profile can be defined as the profile
X derived from the primary profile by suppressing
the long-wave component using a long-wavelength
(high-pass) filter, with a cut-off Ac. This roughness
profile provides the foundation for the evaluation of
the roughness profile parameters and will automat-
Surface profile
ically include At profile filter information, because it
is derived from the primary profile.
(b)

Figure 20. (a) Transmission characteristic of roughness and wavi-

Waviness profile
ness profiles. (b) 5urface profile. [50urce: 1504287: 1997]
The derivation of this waviness profile is the result
of the applieation of the bandpass filter to seleet the
filtering is vital to any potential analysis of surface surface structure at somewhat Ion ger wavelengths to
texture. that of roughness. The filter Af will suppress any
longer-wave components, namely the profile com-
ponent, while the filter Ac suppresses the shorter-
Profile filter wave components, such as any of the roughness
components. Hence, the waviness profile forms
The "profile filter" can be defined (ISO 4287: 1997) the basis for the evaluation of the waviness profile
as a filter that separates profiles into long- and short- parameters.
wave components (see Figure 20a). In particular,
three filters can be employed for surface measure-
ment (Figure 20a):
Roughness profile mean fine
The roughness profile mean line is a reference line
• profile filter (As) - this filter defines where the
for parameter calculation, which corresponds to the
intersection occurs between the roughness and
suppression of the long-wave profile eomponent by
the presence of shorter-wavelength components
the profile filter Ac.
in a surface;
• profile filter (Ac) - this filter defines the intersec-
tion position between the components of rough- Waviness profile mean fine
ness and waviness;
• profile filter (Aj) - filters of this type define the The waviness profile me an line is a reference line for
intersection point between waviness and the parameter calculation, whieh corresponds to the
presence of any longer wavelengths in a surfaee. suppression of the long-wave profile component by
the profile filter Af.
Primary profile
Primary profile mean fine
The primary profile forms the basis for primary
profile parameter evaluations, being defined as the The primary profile mean line is a reference line for
total profile after application of a short-wavelength parameter calculation, being a line determined by
fitting a least-squares line of nominal form through profile element is a discrimination that prevents
the primary profile. any minute, unreliable measurement features from
affecting the detection of these elements. Several of
these surface texture-related elements and associ-
Surface profile ated features are described in ISO 4287: 1997,
consisting of:
In ISO 4287: 1997, the co ordinate system is defined
by making use of the rectangular coordinate of a • profile element - the section of a profile that
right-handed Cartesian set for the surface profile crosses the mean line to the point at which it next
(Figure 20b). Here, the x-axis provides the direetion crosses this mean line in the same direction - for
of the trace; the y-axis nominally lies on the example, from below to above the mean line;
real surfaee, with the z-axis facing in an outward • profile peak - the section of the profile element
direetion from the material to the surrounding that is above the mean line, namely, the profile by
medium. Thus, the real surfaee can be defined as which it crosses the mean line in the positive
the surfaee limiting the body and separating it direction until it on ce again crosses this mean
from the surrounding medium. The lay previously line, but now in the negative direction;
alluded to in Section 1.1 is normally functionally • profile valley - as described for the "peak" above,
significant for the workpiece, so it is important to but with the direction reversed;
specify it on engineering drawings in terms of the • discrimination level - within a profile it may be
type of lay and its direction (ISO 1302: 2001 - likely that aminute fluctuation occurs whieh
Appendix B). takes the profile trace across the mean line, then
back again almost immediately. In such a condi-
tion the departure of the trace from the mean
line is not considered to be a real profile peak
1.8 Geometrical parameters or valley. In order to prevent automatie systems
from counting small trace departures, allowing
them to only consider those larger surface fea-
The following discussion is concerned with the tures, they specify partieular heights and widths
calculation of parameters from the measurement to be counted. Hence, default levels are estab-
data. These parameters are all derived after the form lished in the absence of other specifications.
has been removed. It should be stated that any para- These default levels are set such that their profile
meters selected should be appropriate for a given peak heights or valley depths must exceed 100/0 of
application and that not all of them would be neces- the Rz, Wz or pz parameter value. Furthermore,
sary in all circumstances. On an engineering the width of either a profile peak or valley must
drawing, a designer may have specified a parameter exceed 1% of the sampling length, with both
to define the part's functional behaviour, which criteria being simultaneously met;
means that this parameter is outside the control of • ordinate value Z(x) - this represents the height of
the user of the instrument. When surface texture the assessed profile at any position x, its height
parameters have been specified, the user needs to being regarded as negative when the ordinate lies
have a clear and unambiguous understanding of the below the x-axis, and otherwise positive;
manner in whieh they are calculated. • profile element width Xs - this is concerned with
In understanding and evaluating surfaces, the the x-axis segment intersecting with the profile
philosophy behind the reasons why peaks and element (see Figure 21a);
valleys are significant is important, although just • profile peak height Zp - this represents the
how and what represents a peak/valley is not always distance between the me an line on the x-axis and
clear (see further discussion on this topie in the the highest point of the largest profile peak (see
Appendix). When a surface topography shows the Figure 21a);
presence of a peak or valley that is not significant, • profile valley depth Zv - this represents is the
the decision concerning its importance for in- distance between the mean line on the x-axis and
service functional characteristics is not always a the lowest point of the deepest valley (see Figure
simple matter to establish. In the surface profile 21a);
shown in Figure 21(a), for example, are both the • profile element height Zt - this equates to the sum
peaks here of importance, or even significant? In of the peak height and valley depth of the profile
order to minimise any confusion for the user, the element, namely, the sum of Zp and Zv (see Figure
latest standards have introduced the important 21a);
concept that the profile element consists of a peak • loeal slope dZ/ dX - this represents the slope of the
and valley event. Moreover, associated with this assessed profile at position Xi (see Figure 21b).
 The loeal slope's numerieal value eritieally
depends upon the ordinate spacing and thus 1.9 Surface profile parameters
influenees paq, Raq and Waq;
• materiallength of profile at level "e" Ml( e) - this The following examples of defined parameters ean
represents the seetion lengths obtained when be ealculated from any profile. The designation of
interseeting with the profile element by a parallel letters follows the logie that the parameter symbol's
line to the x-axis at a given level "e" (see Figure first eapital letter denotes the type of profile under
21e). evaluation, for example the Ra is ealculated from the
roughness profile, while Wa has it origin from the
waviness profile, with the Pa oeeurring from the
primary profile. In the subsequent seetions, the defi-
nitions and sehematie representations of many of
these surfaee profile parameters are deseribed along
with examples of their usage.

(a) Zv

Xs

dZ(x)/dX

(b)

dZ(x)/dX

(e)

MI (e) =MI l + MI2

Sampling length

Figure 21. (<t) Profile length, (b) local slope and (c) materiallength of a surface trace. [Source: 1504287: 1997]
1.9.1 Amplitude parameters (peak- not without some merit, as they can establish un-
usual surface features such as either a burr on the
to-valley) surface or sharp spike, and will indicate the presence
of scratches or cracks - possibly indicative of low-
grade material or poor production processing.
Maximum profile peak height Hp, Wp
andPp Maximum profile valley depth Hv, Wv
This parameter is represented by the largest peak andPv
height (Zp) within the sampling length, with its
height being measured from the me an line to This parameter is represented by the largest valley
the highest point (see Figure 22a). The Rp parameter depth (Zv) within the sampling length, with its
(equating to roughness) is generally less favoured, absolute value Rp (equating to roughness) being
with preference given to parameters based on the obtained from the lowest point on the profile from
total peak-to-valley height. Often Rp and its associ- the mean line (see Figure 22b). As in the case above
ated parameters Wp and Pp are referred to as for peaks, the maximum profile valley depths are
extreme-value parameters, being somewhat un- extreme-value parameters. In a similar manner to
representative of the overall surface, because their Rp, Rv can establish whether there is a tendency to
numerical value can vary between respective sam- workpiece cracking, spikes and burrs, detecting such
pIes. In order to minimise variation in Rp, it is features on the surface.
feasible to average readings over consecutive samp-
ling lengths, although in most cases the value
obtained is numerically too large to offer practical Maximum height ofprofile Hz, Wzand pz
assistance. The Rp and its Wp and Pp derivatives are
The maximum height parameter (for example, Rz -
for roughness) is the sum of the height of the largest
profile peak height (Zp), together with the largest
profile valley (Zv) within the sampling length (see
Figure 23a). In isolation, the Rz does not provide too
much useful surface information, and therefore it is
often divided into Rp and Rv, previously described.
(a) In fact, in ISO 4287: 1984, the Rz symbol indicated
the ten point height irregularities, with many of the
previous surface texture measuring instruments
measuring this Rz parameter - see the Appendix for
more information regarding this previously utilised
parameter.
Sampling length

Mean height ofprofile elements He, Wc

andPc
Parameters of this type evaluate surface profile
element heights (Zt) within the sampling length (see
(b) Figure 23b) and to obtain this parameter it requires
both height and spacing discrimination, previously
mentioned. When these values are not specified then
the default height discrimination used shall be 10%
of Rz, Wz and Pz, respectively, with the default
spacing discrimination being stated as 1% of the
Sampling length sampling length. Both of these height and spacing
requirements must be met. Normally in practice it is
Figure 22. (a) Maximum profile peak height and (b) maximum extremely rare to utilise this parameter because of
profile valley depth (examples of roughness profiles), [Source: ISO its difficulty in interpretation and, as a result, it
4287: 1997] would rarely be used on an engineering drawing.
 (a)

Rz

(b)

Figure 23. (a) Maximum height of profile and (b) height of profile elements (examples of roughness profiles). [Source: ISO 4287: 1997]

Total heightofprofile Rf, Wtand Pt 1.9.2 Amplitude parameters

(average of ordinates)
This parameter relates to the sum of the height of
the largest peak height (Zp) and the largest profile
valley depth (Zv) within the evaluation length. It is Arithmetical mean deviation of the
defined over the evaluation length rather than the
sampling length and in this manner has no aver- assessed profile Ro, Woand Po
aging effect. This lack of averaging means that any
surface contamination (dirt) or scratches present The arithmetic mean parameter is the absolute ordi-
will have a direct effect on the numerical value of, nate value Z(x) within the sampling length The
say, Rt. numerical value of Ra is able to vary somewhat
without unduly influencing the performance of a
surface. Often on engineering drawings a designer
will specify a tolerance band, or alternatively a
maximum value for Ra that is acceptable. The math-
ematical expression of this parameter is given below
for these absolute ordinate values as:
 u ace texture: two- Imenslona

ously mentioned on several occasions, if a defined

"unidirectionallay" is present on the surface (Figure
4), then any measurements should be undertaken
perpendicularly to this "lay" if misinterpretation of
where N = number of measured points in a sampling the surface is to be avoided (Figure 5).
length (see Figure 24b). The numerical value of Ra does not impart any
In Figure 24(a) is illustrated the graphical deriva- information regarding the surface shape, nor of
tion of this parameter within the sampling length, any irregularities present. In fact, it is quite possible
with the shaded areas of the graph below the cent re to have identical Ra values for surfaces that are
line in A being repositioned above this centre line in markedly different in their profiles (Figure A in the
B. Here, the Ra value is the mean line of the resulting Appendix) and this will affect their potential in-
profile indicated in C. service performance, with engineering drawings
The absolute average roughness over one often quoting the type of production process (Figure
sampling length is given by the Ra value. This means 3c). Historica11y, in Great Britain the most widely
that the influence of a discrete but non-typical peak used of a11 the surface texture parameters has been
or va11ey on the numerical value of Ra is not rea11y Ra, but this should not dis courage industrial and
significant. Good metrological practice is to ensure research organisations from using others, as they
that a number of assessments of Ra occur over may offer more in terms of information regarding a
consecutive sampling lengths, enabling the accep- surface's functional performance.
tance of an average of these values. This sampling The mathematical derivation of both the com-
strategyensures that the numerical value of Ra is monly utilised parameters Ra and Rq are given
typical for the inspected surface. As has been previ- below (see Figure 24b):

(a)

:...:.f-------- Sampling length (L) - - - - - - - . . . ,

c
Graphieal derivation or Ra
Ccntrc linc

Derivation of the arithmetical mean deviation

(b)

I
I
I Sampli ng Icngth (L)
i- -I
Figure 24. Determination of Ra and Rq. ((ourtesy of Taylor Hobson.)
 1 1 N
Rsk=-- ~ Z3
R q3 N I tft
The skewness parameter of an amplitude distribu-
Root mean square deviation from the tion eurve as depieted in Figure 25(a) indieates a
assessed profile Rq, Wqand Pq eertain amount of bias that might be either in an
upward or downward direetion. The amplitude
The root mean square of the ordinate values Z(x) distribution eurve shape is very informative as to the
within the sampling length is established by the overall construction of the surfaee topography. If
departures from me an line of the profile and ean be this eurve is symmetrieal in nature, then it indieates
mathematieally expressed in the following way: symmetry of the surfaee profile; eonversely, an
unsymmetrieal surfaee profile will be indieative of a

Zq=~ ~ tz: skewed amplitude distribution eurve (Figure 25b

and e). As ean be seen in Figure 25(b) and (c), the
direction of skew is dependent upon whether the
bulk material appears above the mean line - termed
If a eomparison is made between the arithmetie negative skewness (Figure 25e) - or below the mean
average and the root mean square parameters, line - positive skewness (Figure 25b). Utilising this
the latter has the effeet of providing additional skewness parameter distinguishes between profiles
weighting to the numerieally greater values of having similar if not identical Ra values.
surfaee height. One of the reasons for the harmony Further, one ean deal with this amplitude or
between both the Ra and Rq parameters is mainly height data statistieally, in the same manner as one
from a historieal perspeetive. The Ra parameter is might physieally measure anthropometrie data sueh
mueh easier to determine graphieally from a profile as a person's average stature, within a specifie popu-
reeording and for this reason, it was primarily lation range. As with the statistieal dispersion of
adopted prior to automatie surfaee measurement population stature, engineering surfaces can exhibit
instruments being developed. Onee the roughness a broad range of profile heights.
parameters utilising automatie surfaee texture For example, aboring operation with a relatively
instrumentation were made available, the parameter long length-to-diameter ratio may cause defleetion
Rq had the advantage of being able to negleet the (elastie deformation of the boring bar) and oeeasion
phase effect from the eleetrieal filters, while the Ra the eutting insert to deflect, producing large peak-
parameter using arithmetie average will be affeeted to-valley undulations along the bore (waviness).
by these phase effeets and eannot be disregarded. Superimposed onto these longer wavelengths are
To eompound the problem, Ra has been virtually small-amplitude eyclieal peaks (periodic oscilla-
universally adopted for maehining specifie.ati0t.Is tions) indieating vibrations resulting from the
to the detriment of Rq. The Rq parameter IS stIll eutting proeess. The resultant surface profile for the
employed in many optieal applications for the bored hole would depiet the interaetions from the
assessment of lenses, mirrors and the like for surfaee boring bar deformations and any harmonie oscilla-
optieal quality. tions. This boring bar motion refleeted in the profile
traee would exhibit low average profile height, but
with a large range of height values.
Skewness of the assessed profile Rsk, Wsk Surfaee texture data ean be statistieally manipu-
andPsk lated, beginning with the profile traee's height, or
amplitude distribution curve - this being a graph-
This parameter represents the quotient of the me an ieal representation of the distribution of height
eube value of the ordinate values Z(x) and the eube ordinates over the total depth of the profile. The
of Pq, Rq or Wq, respeetively, within the sampling eharaeteristies of amplitude distribution eurves ean
length (Figure 25). The skewness is derive~ from be defined mathematieally by several terms ealled
the amplitude distribution eurve, representmg the "moments", but more will mentioned eoneerning
symmetry about the mean line. This parame~er this aspeet in the Appendix.
cannot distinguish if profile spikes are evenly diS- A numerieal value ean be given to Rsk and in the
tributed above or below the mean line, being eonsid- case of Figure 25(b) the "positive skewness" may
erably influeneed by any isolated peaks or valleys eventually obtain an adequate bearing surfaee,
present within the sampling length. Skewness ean although it is unlikely to have oil-retaining abilities.
be expressed in mathematieal form with Rsk, equat- This type of surfaee ean typieally be exploited for
ing to: adhesive-bonding applieations. The surfaee eharae-
 (a) Skewness bias

Negative skewness (Rsk) Positive skewness (Rsk)

(b) A "locking" surface texture

Amplitude
distribution
curve

(c) A "bearing" surface texture

Figure 25. How surface texture topography influences the amplitude distribution curve.

terised by Figure 25(c) might occur in the cases of bution curve can also relay useful information about
porous, sintered or cast-iron surface topography - the dispersion, or "randomness" of the surface pro-
having comparatively large numerical values of file, which can be quantified by means of a para-
negative skewness. The surface is sensitive to meter known as kurtosis (Rku).
extreme ordinate values within the profile under
test; this is due to Rsk being a function of the cube
of the ordinate height. As a result of this peak sensi-
tivity, it is a hin dran ce when attempting to inspect Kurtosis of the assessed profile Rku, Wku
plateau-type surfaces, although the Rsk parameter andPku
indicates a reasonable correlation with a compo-
nent's potential load carrying ability, or its porosity. The kurtosis parameter, typified by Rku, is the me an
The shape, or "spikeness" of the amplitude distri- quadratic value of the ordinate values Z(x) and the
 (a) Kurtosis is influeneed by the distribution shape

Kurtosis (Rku) > 3

Kurtosis (Rku) < 3

(b) Material distributed evenly aeross the whole of the surfaee topography

Amplitude
distribution

(e) Material distributed about the eentreline

Figure 26. Variation in surface topography influences the shape and height of the amplitude distribution curve.

fourth power of Pq, Rq or Wq, respectively, within surface topography, Rku is a useful parameter in the
the sampling length. However, unlike the skewness prediction of in-service component performance
parameters previously described, namely Rsk and its with respect to lubricant retention and subsequent
derivatives, Rku can detect if the profile peaks are wear behaviour.
distributed in an even manner across the sampling It should be mentioned that kurtosis cannot
length trace, as well as providing information on the convey differences between either peaks or valleys
profile's sharpness (see Figure 26). in the assessed profile.
Kurtosis provides a means of measuring the The kurtosis parameter indicative of Rku, can be
sharpness of the profile, with a "spiky" surface ex- mathematically expressed in the following manner:
hibiting a high numerical value of Rku (Figure 26c);
alternatively, a "bumpy" surface topography will
have a low Rku value (Figure 26b). As a consequence
of this ability to distinguish variations in the actual
 1.9.4 Hybrid parameters

Root mean square slope of the assessed

profile Ri3.q, Wi3.q and Pi3.q
The root me an square slope parameter refers to the
value of the ordinate slope dZ/dX within the
sampling length, depending upon both amplitude
and spacing information, hence the term hybrid
parameter. The slope of the profile is the angle it
makes with a line that is parallel to the mean line,
with the mean of the slopes at all points in the profile
being termed the average slope within the sampling
Figure 27. Width of profile elements. [50urce: 150 4287: 1997] length. By way of illustration of its use, it might be
necessary to determine the developed or actual
profile length, namely the length occupied if all the
Later, the parameters Rsk and Rku will be applied peaks and valleys were laid out along a single
to specific manufacturing pro ces ses to produce straight line. Hence, the steeper the slope the longer
manufacturing process envelopes describing the will be the actual surface length. In practical indus-
functional aspects of a particular generated surface trial situations, the parameter might be utilised in
condition. either a plating or painting operation, where the
surface length for keying of the coating is a critical
feature. Moreover, the average slope can be related
to certain mechanical properties such as hardness,
1.9.3 Spacing parameters elasticity or more generally to the crushability of a
surface. Further, if the root mean square value is
small, this is an indication that the surface would
Mean width ofprofile elements Rsm, have a good optical reflection property.
WsmandPsm NB In the Appendix are listed some of the
previous and current techniques for the assessment
The mean width parameter of the profile element for these hybrid parameters.
widths Xs within the sampling length relates to the
average value of the length of the mean line section
that contains adjacent profile peaks and valleys 1.9.5 (urves and related parameters
(Figure 27). The parameter needs both height and
spacing discrimination, and if not specified the Curves and their related parameters are defined over
default height bias utilised shall be 10% of Pz, Rz or the evaluation length rather than the sampling
Wz, respectively. Moreover, the default spacing length.
discrimination shall be 1% of the sampling length,
with both of these conditions requiring to be met.
The spacing parameters are particularly useful in
determining the feedmarks from a specific machin- Material ratio of the profile Rmr(c),
ing operation, as they relate very closely to that of Wmr(c) and Pmr(c)
the actual feed per revolution of either the cutter or
workpiece, depending on which production process The material ratio of a profile refers to the ratio of
was selected. the bearing length to the evaluation length and is
NB More is mentioned on spacing parameters represented as a percentage. This bearing length can
and their affects on the surface topography in the be found by the sum of the section lengths obtained
Appendix. by cutting the profile line - termed slice level - that
is drawn parallel to the mean line at a specified level.
The ratio is assumed to be 0% if the slice level is at
the highest peak; conversely, at the deepest valley
this would represent 100%. Parameters Rmr(c),
Wmr(c) and Pmr(c) will determine the percentage of
each bearing length ratio of a single slice level, or
say, 19 slice levels that are drawn at equal intervals referred to by the highest and lowest profile points.
within the Rt, Wt or Pt, respectively. In the majority of circumstances mating surfaces
NB More is mentioned on this topic in the require specific tribological functions; these are the
Appendix. direct result of particular machining operational
sequences. Normally, the initial production opera-
tion will establish the general shape of the surface -
Material ratio of the curve of the profile by roughing-out - providing a somewhat coarse
(Abbot-Firestone or bearing ratio curve) finish, with subsequent operations necessary to
improve the finish, resulting in the desired design
The material ratio curve represents the profile as a properties. This machining strategy provides the
function of level. Specifically, by plotting the bearing operational sequence that will invariably remove
ratio at a range of depths in the profile, the manner surface peaks from the original process, but often
by which the bearing ratio changes with depth leaves any deep valleys intact. This roughing and
provides a method of characterising differing shapes finishing machining technique leads to a surface
present on the profile (Figure 28a). The bearing area texture type that is often termed a stratified surface.
fraction can be defined as the sum of the lengths In such cases the height distributions are negatively
of individual plateaux at a particular height, normal- skewed; therefore this would make it otherwise diffi-
ised by the total assessment length - this parameter cult for an average parameter such as Ra to repre-
is designated by Rmr (Figure 28a). The values of Rmr sent the surface effectively for either its specification
are occasionally specified on engineering drawings, or in the case of quality control matters.
although this may lead to large uncertainties (see NB This topic is discussed in further detail in the
Chapter 6, Section 6.4) if the bearing area curve is Appendix.

C
(a)

J
I I
o 20 40 60 80 100
Ib r - - -
Evaluation length
-----..,
Rmr(c) %

(b)

.\
.....
I"-

R&[-- CO
1---- -- - --'
NI ........ ~

~
Cl
I--- -- - - ~ f-- - - \...
I ~r\
I : \

I
I : \
I ,
I
I
o 10 20 30 40 50 60 70 80 90 100
I I
RmrO Rmr

Figure 28. (a) Material ratio curve. (b) Profile section level separation. [Source: ISO 4287: 1997]
Profile section height difference RBc, with the bearing ratio at the corresponding position
here being the Rmr (Figure 28b). The reason for the
WBc undPBc reference percentage is to eliminate any spurious
peaks from the assessment, as they would tend to be
This parameter can be represented as the vertical worn away during the initial burn-in/running-in
distance between two section levels of a given period. The slice depth will then correspond to a
material ratio curve (see Figure 28b). satisfactory surface roughness, or to an acceptable
level of surface wear.

Relative material ratio Rmr, Wmrand

Pmr Profile height amplitude curve
The relative material ratio can be established at a This represents the sampie probability density func-
profile section level R8c, being related to a reference tion of the ordinate Z(x) within the evaluation length.
CO (see Figure 28b), where: The amplitude distribution curve, as it is usually
known, is a probability function giving the proba-
Cl = CO - R8c or (W8c or P8c) bility that a profile of the surface has a particular
height at a certain position. Like many proba-
CO = C(RmrO, WmrO, PmrO) bility distributions, the curve normally follows the
contours of a Gaussian - bell-shaped - distribution
Rmr refers to the bearing ratio at a predetermined (Figure 29a). The amplitude distribution curve in-
height (see Figure 28b). One method of specifying forms the user of how much of the profile is situated
the height is to move over to a certain percentage - at a particular height, from the aspect of a histogram-
reference percentage - on the bearing ratio curve, like sense.
then move down to a certain depth (slice depth), This curve illustrates the relative total lengths

+--k::~ Mean line

(a)

I Amplitude density
I
I Evaluation length
foe:::::Jl--------

+y Amplitude
Profile graph distribution

(b)

-y

Evaluation length (E)

Figure 29. Profile height and amplitude distribution curve. [Source: ISO 4287: 1997]
 Industrial metrology , .

over which the profile graph will attain any selected manufactured surfaces clearly indicate visible repet-
range of heights that are above or below the mean itive marks on either the material itself or indirectly
line (see Figure 29b). The profile's horizontallengths via the profile graph, while on other surfaces it
are included within a narrow zone width öy at a becomes difficult to distinguish any repetitive irreg-
height z and are represented in Figure 29(b) by "a, ularities from random occurrences. This visual diffi-
b, c, d and e". Therefore, by expressing the sum of culty is particularly true when the amplitudes of the
these lengths as a percentage of the evaluation repetitive irregularities are less than the random
length, this will give a measure of the relative occurrences. Moreover, the presence of repetitive
amount of the profile at height z. surface features - however small - can indicate
This graph is known as the amplitude distribu- factors such as tool wear, machine vibration or defi-
tion at height z, so by plotting density against height ciencies in the machine, with identification of them
the amplitude over the whole profile can be deter- being an important factor. Any random pattern
mined - producing the amplitude density distribu- occurring on the surface can reveal, for example,
tion curve. whether a built-up edge condition has transpired on
the cutting tool resulting in a degree of tearing of
the machined surface - this tends to be of a random
1.9.6 Overview of parameters nature and is not predictable.
The extent of this randomness of the surface can
In Table 2 are shown the more recent and previously be monitored and assessed by isolating random
utilised parameters, together with information from repetitive textural patterns, this being achieved
concerning whether the parameter has to be calcu- by auto-correlation. When a profile is perfectly
lated over the sampling or evaluation length. periodic in nature - typified by a sine wave - the
relationship of a particular group of points repeats
itself at a distance equal to the wavelength. Con-
versely, if the profile under inspection is comprised
1.10 Auto-correlation function entirely from random irregularities, the precise rela-
tionship between any specific points will not occur
at any position along the trace length, hence any
Every day of one's life, the technique of correlation repetitive feature or group of features can be identi-
is applied to compare sights, sounds, tastes, objects fied. Computers equipped with fast digital proces-
- the list is almost limitless - this being a process of sors have significantly reduced the tedious task of
comparison. The auto-correlation function (ACF) is determining a surface profile's auto-correlation.
essentially a process of determining the relationship The technique exploited by auto-correlation is to
of any point on the profile to all other points. Some compare different parts of the surface profile; in this
Table 2. Past and present parameters for surface texture

Parameters 1997 edition 1984 1997 Determined within:

edition edition
Evaluation Sampling
length length

Maximum profile height Rp Rp X

Maximum profile valley depth Rm Rv X
Maximum height of profile Ry Rz X
Mean height of profile Re Re X
Total height of profile Rt X
Arithmetical mean deviation of the assessed profile Ra Ra X
Root mean square deviation of the assessed profile Rq Rq X
Skewness of the assessed profile Sk Rsk X
Kurtosis ofthe assessed profile Rku X
Mean width of profile elements Sm RSm X
Root mean square slope of the assessed profile tlq RC.q X
Material ratio of the profile RMr(e) X
Profile section height difference FrOe X
Relative material ratio tp Rmr X
Ten-point height (deleted as an ISO parameter) Rz

Source: ISO 4287: 1997 (ElF).

 Amplitude Auto-correlation
(a) distribution:
Profile: function:
A(ß)

(b) A(ß)

Slope
Curvature
Profile
Length A(ß)
(e) Decrement

I
1----, Sampling length _I
Peakspermm
Co'

High Spot Count

Peak Count
Power Spectrum
Fourier Analysis

Figure 30. Comparison of surface parameters for typical surfaces showing those currently in use. [After Whitehouse, 1978.]

manner profile repetItlOns or similarities ean be Figure 30 shows an example of the ACF relating
discovered. When the profile of the surfaee displays to speeifie surfaee profile eonditions. As mentioned
either an isotropie or random shape then no above, the ACF eontains information relating to the
portions will be similar, exhibiting low auto-eorre- spacings of peaks and valleys in the profile. An alter-
lation funetion. On the other hand, if an anisotropie native to the ACF is the equivalent term known as
or periodic surfaee topography oeeurs - for example the power speetrum, or power speetral density, in
as in single-point turning operations - then repeti- whieh spacings are replaeed by frequencies - one
tion of the surfaee profile oeeurs and auto-eorrela- being a Fourier trans form of the other. For the ACF
tion will be high. examples shown in Figure 3D, the seale of the hori-
zontal axis can be the mean distance between peaks tion may mask significant hidden harmonie detail
or the average distance between mean line crossings. which could play an important role in its later in-
The ACF is derived from the normalised auto- service application. Results from the FFT analysis of
covariance function (ACVF) - this being as follows: the profile (Figure31 b) reveal the differing sine wave
amplitudes necessary to generate the original
A ) _ ACVF(Lix) profile. In this case, the largest sine wave amplitude
ACF.ux
( - ( 2 equals a value of 0.7 mm, this being termed the
Rq)
"dominant wave".
where It should be emphasised that the information
obtained from the FFT analysis will be exactly the
irn same as that acquired by the auto-correlation
ACVF(Lix) = lim y(x) y(x + Lix) dx method, but obviously it is displayed in aaltered
Im-----7 OO f
0 form. Due to the isolation of the harmonics into
specific bands for a particular surface profile, FFT
where 1m = assessment length,y(x) = profile height can impart valuable understanding as to how the
at position x and Rq = rms parameter. production process might have been inftuenced
The ACF does not contain information about the by variables in processing for that product. For
amplitude of the profile, with its values ranging from example, the dominant harmonie in the FFT display
plus one - giving perfeet correlation - to negative for a profile might give a valuable insight into the
one - this having a correlation of the inverted but vibrational frequency for a specific machine tool.
shifted profile. Due to the fact that the ACF is inde- Knowledge of this vibrational effect on the resulting
pendent of the amplitude of the profile it is more surface profile may enable the attenuation of this
popular than the ACVF. In the case of an ACF having vibrational signature by judicious adjustment of
an isotropie (random) profile, it quickly decays to either cutting data, or specific maintenance to the
zero (Figure 30a), with the profile trace being of an machine tool to improve the subsequent surface
unrepetitive nature. Conversely, the tendency for a profile's condition.
trulyanisotropic (periodie) generated surface is that
it does not decay to zero, with Figure 30(c) illus-
trating some periodicity in its profile trace - indi-
cating that decay here is significantly less than that
for Figure 30(a). Despite the fact that both Figure
1.11 Appearance of peaks and
30(a) and (c) have similar Rz values, their respective valleys
ACF are distinctly different and can be used as a
means to discriminate between surfaces and indicate
whether they will fulfil the desired in-service appli- For many engineering and statistical graphs no
cations, which might not otherwise be apparent. intrinsic relationship exists between the plotted
measurement units; for example, when plotting the
rise in temperature of a heat treatment oven against
Fourier analysis (FFT) elapsed time the ordinates would be represented by
both degrees Celsius and ho urs - two quite dissim-
Any surface profile will normally be of some ilar units. As a consequence of these disparate units,
complexity, this being comprised of an array of acceptance of the shape of the graph is acknow-
differing waveforms that are superimposed onto one ledged, whatever scales are applied for the ordinates.
another. An actual profile's specific composition will It is quite different when dealing with a surface
be dependent on the shape and size of the wave- profile graph, where both the horizontal and vertical
forms existing within the profile; both their ampli- ordinates are represented by the same units (i.e.,
tude and frequency, termed "harmonics", can be length). At first introduction, it may be difficult to
established by the application of Fourier analysis. appreciate that due to the difference of scale the
The technique known as a "fast Fourier trans- graph will not instantaneously give an indication as
form" - FFT for short - enables one to establish a to the shape of the irregularities. Figure 32 illustrates
series of sine waves being generated from the surface the effect, with this disparity being one of the most
profile, and when combined together contribute to difficult factors in surface texture analysis to
the original profile. Individual harmonics represent comprehend.
a specific frequency-wavelength, in combination It is very important to note that although the
with their associated amplitude. visual profile's shape is distorted due to the applica-
The surface profile depicted in Figure 31(a) was tion of different horizontal and vertical magnifica-
selected to illustrate a periodic form, which in isola- tions, measurements scaled fram the graph are
 (a) Profile
(11m )
10

-s

-10

(b) Wavelength

0.8 0.4 0.3 0.2 0.1

Figure 31. Fourier analysis of an idealised (i.e. periodic) profile.

correct. Such effects can be visually demonstrated peaks and valleys on the profile graph, this can lead
with reference to the peak angles of the profile illus- to considerable confusion when visually assessing
trated in Figure 32, indicating how the profile shape the actual graph. However, the geometrie relation-
varies as horizontallvertieal magnification is altered. ship between angles on the graph and those that
Figure 32(III) verifies this fact, as the horizontal occur on the physical profile of the component
magnification of the surface profile is increased under test are more easily determined. As an
(Figure 32II), the length x-x expands to Xl_XI and example of this interpretation, Figure 33(a), on the
the peaks; A, B, C and D appear flaUer. Increasing profile graph, depicts a symmetrieal peak with an
the horizontal magnification still further - until included angle 2a. Figure 33(b) exhibits the corre-
it equals the vertical magnification (Figure 321) - sponding peak, on the actual surface, having an
expands the sampling length y-y to yl_yl. Once angle 2ß. The ratio of the angles is determined by
again, with this still higher magnification, the peaks
E and G and valleys Fand H now have a much flaUer tan ß Vv
appearance. A point worth making is that the actual tan a Vh
difference in respective heights of corresponding
peaks and valleys in I, II and III are identical. This ratio is strictly true for symmetrical peaks,
As a result of the spiky appearance of both the although it is only slightly different for asperities of
 (I) x5000
f=J--------C=-I

(11)

f
x5000

~
j /
I

I j /
(III)
xl00
i-=----P-I je /D
~ / /
x5000

Figure 32. Diagram illustrating how profile shape varies as the horizontal magnifieation is redueed: (I) surfaee profile magnified x5000 equally
in all direetions; (11) profile with A Vv: Vh ratio of 5:1; (111) profile graph recorded with a Vv:Vh ratio of 50:1. ((ourtesy ofTaylor Hobson.)

an asymmetrical nature. For example, the data ß = 83.5°

obtained from a profile graph in which the vertical
and horizontal magnifications and peak geometry Thus
might be Vv = x5000, Vh = x50 and 2er = 10°. This
will therefore represent 2ß = 167°

er = 5° This example of peak geometry (Figure 33a) repre-

sents a fairly sharp peak on the profile graph, but
tan er = 0.0875 only a gradual slope - rising and falling at 6S from
the horizontal - on the actual component's surface
hence (Figure 33b). Due to the fact that it becomes difficult
to accurately measure the peak angle on the graph,
tan ß = 0.0875 x 5000/50 and it is rare that an accurate assessment of the
surface angles is required, an approximation of 2er :
= 8.75 2ß is normally sufficient.
Table 3 provides approximate conversions for
giving several V v : Vh ratios.
 u ace texture: two- Imenslona

(a) Peak on profile graph true for the major ridges for a surface exhibiting
marked directionallay (see Figure SA) being traced
at 90° to the direction of the dominant surface
pattern lay. Conversely, for a surface displaying ran-
dom textural influences, such as the "multi-direc-
tional" example in Figure 4, it is extremely unlikely
that a stylus's path will trace across the summit of
every peak. In these circumstances, some peaks and
valleys will be displaced from the path of the stylus
by varying distances, with the result that the stylus
will also traverse across many flanks. Stylus vertical
displacement, although somewhat less than the full
(b) Corresponding peak on the surface peak height or valley encountered, will moreover be
correct for the traced feature. Assumptions made by
many are that every peak on the graph is the summit
of a peak on the actual surface, but where random
surfaces occur there will be sufficient actual sum-
mits encountered to prevent any significant inaccu-
racy as a result.

1.12 Stylus-based and non-

Figure 33. Relationship between peak angles on the graph and the
actual surface. ((ourtesy ofTaylor Hobson.)
contact systems

It has been previously stated by way of a light- Effeet ofstylus

hearted analogy for the representation of two rough
contacting surfaces that they resemble "Switzerland It should be appreciated by now that the stylus and
on top of Austria", although this simply shows the skid, when employed, are the only parts which are
influence of the visual aspect of the profile graph. A in active contact with the surface under test.
more preferable analogy might be "The Netherlands Therefore both the shape and dimensions of this
on top of Iowa", as the peaks on the surface are in pick-up are critical factors and will strongly influ-
general of quite blunt and low aspect. ence information retrieved from the surface. The
When considering the profile with its associated geometry of a typical stylus - its shape and size -
peaks and valleys, it has been presumed that the was discussed in Section 1.4 and illustrated in Figure
peak height on a profile graph is characterised by 11 (a); it influences the accuracy obtainable from
the summit of individual peaks traced by contact the surface. Due to the finite size of the stylus some
with the stylus for a surface under test. This may be surface imperfections, such as visible scratches, are

Table 3. Peak angles (2ß, Figure 33a) on the surface corresponding to peak angles of 2a on the graph

Ratio: VJVh Peak angle on graph (2a, Figure 33)

1° 5° 10° 15° 20° 30° 400 50°

10 10° 47 0 82° 105° 121 ° 139° 1490 155°

25 25° 95° 131 ° 146° 154° 1620 167° 170°
100 82° 1540 167° 171 ° 1730 1760 177° 177°
250 131° 1690 175° 176° 1770 178° 179° 179°
500 154° 175° 177° 178° 178° 179°
1000 1670 177° 179° 179° 1790
2500 174° 179°
5000 177°

Courtesy ofTaylor Hobson.

too fine to be completely penetrated, even with For surfaces other than of this regular triangulated
the smallest tip available. This is not too much of geometry, the values tend to be only approximate,
a hindrance to assessment as the Ra is an "averaging although they do indicate the order of any error
parameter", so the omission of the finest textural likely to be occasioned. The horizontalordinates in
features will not unduly affect the resulting value. Figure 34 are the ratio of the stylus tip radius to the
To illustrate this point, many years ago aseries actual Ra value of the surface, with the vertical
of tests were conducted on ground surfaces by ordinates being the amount by which the measured
Reason et al. (1944), these sampies had four Ra value is too low. The chart demonstrates that only
increasing levels of roughness. The reduction in when the tip radius is greater than approximately 20
measured Ra value when utilising either a l.25 /-Lm times the Ra value is this error of any significance.
or 10 /-Lm stylus was 4.5 to 3.25,8 to 7.5, 11 to 10 and By traversing a stylus across a standard incorpo-
26 to 25 /-Lm, respectively. This small variation rating narrow grooves of known widths - 20, 10,5
occured despite the fact that the larger stylus was and 2.5 /-Lm - this is a technique for estimating stylus
eight times greater than the smaller. wear or damage. Obviously, the blunter the tip the
For undistorted surfaces (periodic), there is a less distance it can descend into the groove, this
predictable relationship between the stylus radius penetration depth being measured from the profile
and the resulting error introduced into the Ra graph. The effective stylus radius can be established
reading. Figure 34 charts the approximate values of from the calibration curve supplied with the arte-
this error, being applicable to regular profiles with fact. Stylus co ne angle can also be measured by this
150 included angle triangular peaks and valleys -
0
method; because the stylus wears with use, the
typical of surfaces utilised for calibration standards. change in effective tip size can be monitored.

100
80
60
50 ~
i-'
..- -
40
V
30 V
20
'j
~

~....
10
8
1
/
..Q 6 /
I
5 5
4 I
'"'
"0
<U 3 /
~
~
2
V
....
~ /
e
<U 1 /
~ 0.8 11
0.6 I
0.5
0.4 /
0.3

0.2 /
1 2 3 4 5 6 8 10 20 30 405060 80100 200
Ratio - Stylus tip radius: Actual Ra of surface

Figure 34. Error due to finite stylus tip radius, for a specific surface. (Courtesy of Taylor Hobson.)
Illustrative example: When measuring a surface of 0.5 I-lm Ra with a stylus tip of 5 I-lm (10:1 ratio) the display will be approximately 2% low.
If the stylus has double this radius (i.e., 10 I-lm) this error increases to 8%.
 Surface texture: two-dimensional

lable 4(a). Typical (traceable) markings on pick-up

dentification marks: ±100 5 90 0.8 30/6 1/0

Stylus data:
f';1easuring range (tJ-m) x
Tip Radius (fLm) x
Tip angle (0) x
Measuring force (mN) x
Skid data:
Skid radius,
longitudinal/traverse x
Distance skid-to-stylus,
longitudinal/traverse x

lable 4(b). Configurations for some standard pick-ups

stylus
Tip angle (0) 60° ± 5° or 90° ± 5°
Figure 35. 90° conical stylus tip mechanically tracing across a
portion of a honed surface. SEM photomicrograph, X200 magnifica- Tip radius (fLm) 2± 1 5± 2 10 ± 3
tion. [Courtesy of PTß ßraunschweig/Feinpruf Perthen GmbH.] Static measuring force
(ie at electric zero, mN)
Stylus force on the surface must not be great Variation in static measuring
enough to deform or scratch the surface as this may force (mN/m) ~ 0.035 ~ 0.2 ~ 0.8
promote erroneous results. Conversely, the force
imparted by the stylus must be of a sufficient level Skid
to ensure that it maintains continuous contact with Distance from stylus dependent
upon pick-up design;
the surface at the desired traverse speed. This factor
is important in the design of the pick-up, with most Radius
(longitudinal, mm) 0.3 1.3 3 10 30 infinity
surface texture instruments having a stylus tip
(lateral, mm) depends upon design of pick-up
radius ofbetween 1 and 10 !-Lm. However, the greater
the radius of the tip, the larger the allowable force Surface roughness (tJ-m) ~ 0.1

that can bear on the surface. If a tip radius is small, Skid force (N) for:
say 0.1 !-Lm (Figure lla), then a very low force must Hard surface ~ 0.5

be used, typically around 0.75 mN. Soft surface ~ O. 25

The conieal stylus geometry was previously men- Linearity of system (%) ±1
tioned in Section 1.4 and illustrated in Figure 10,
invariably having a 90° tip angle. In action (Figure
35), the conical tip angle of 90° can easily cope with
surface features such as that shown, having the materials such as sintered parts. One problem asso-
anisotropie profile of a honed surface. When surface ciated with the chisel-type stylus is the difficulty in
features become greater than 45 ° to the horizontal, or maintaining its attitude at right angles to the lay,
the valley is smaller than, say, 5 !-Lm, then the being more sensitive to error from peaks rather than
mechanieal motion of the tip becomes to some ex- valleys. Stylus misalignment can be minimised by
tent distorted, as illustrated in Figure 11 (c). Specific- the application of "smart parameters" that mathe-
ally, under these conditions, the tip acts in a similar matically ignore certain features. NormaUy, it is
fashion to a low-pass filter with high-frequency fea- recommended to utilise a conventional stylus geom-
tures not being recorded. With a tip angle of 90°, this etry, unless particular problems are encountered
means that the stylus will foul any peak or valley that may cause the standard stylus to impair the
having an included angle of less than 45° - which validity of the resulting data.
may at first seem a considerable limitation, but in Manufacturers of stylus/skid geometries norm-
reality does not cause too much of a problem. aUy apply identifying marks on the piek-up, or at the
Chisel-edge styli (see Figure lla) are a practical very least on its accompanying transportation case.
variation on the cone-shaped stylus (Figure 35), Table 4(a) illustrates the piek-up markings, whereas
being particularly successful at entering porous Table 4(b) highlights the standard configurations.
 InClustrial metrology , .

1.12.1 Pick-up advantage of this type of pick -up is that it enables

a true recording of waviness and profile to be
The function of the transducer or pick-up is to obtained.
convert the minute vertical movements of the stylus, • Motion-sensitive pick-ups (Figure 36b) produce
as it progresses along the surface, into proportionate an output only when the stylus is in motion, with
variations of an electrical signal. The sensitivity the output being related to the speed at which the
demanded of the pick-up is such that it should be stylus is displaced, dropping to zero when
able to respond to stylus movements of approxi- stationary. If displacement is very slow, perhaps
mately 0.1 nm or better. This output is minute and due to widely spaced waviness or to change in
of necessity must undergo significant amplification form, then the output for practical purposes is
to enable stylus trace movements of between 50,000 almost zero. This low output means that varia-
and 100,000 times greater than the stylus movement tions of waviness and form are excluded from the
(Vv x50,000 or xl00,000). profile. These pick-ups te nd to be utilised if
In general, pick-ups can be classified into two instruments do not have recording facilities, such
groups, depending on their operating principle: as some portable equipment.

1. Analogue transducers (Figure 36) The variable-inductance pick-up, illustrated in

Figure 36(ai), has been widely used for many years.
• Position-sensitive pick-ups (Figure 36a) give a The stylus is situated at one end of a beam pivoted
signal proportional to displacement, even when at its centre on knife edges, while at the opposite end
the stylus is stationary. Output is independent of an armature is carried which moves between two
the speed at which the stylus is displaced and is coils, changing the relative inductance. This pick-
only related to the position of the stylus within a up's operating principle is as follows: coils are
permissible range of vertical movement. A big connected in an AC bridge circuit (Figure 36aii) such

(a) Position sensitive pick-ups

Phase dctcctor.
demodulalor and fiUer

Carrier
IV\,

(i) ariablc induclancc (ii) Schcmatic diagram 01" a modul ~ltc carricr systcm

(h) Motion-sensitive pick-ups

Magn.'

SPring 10 gtve a downward force

Slyfus
(i) Picl.O·clcctric pick-up (ii) Moving·coil pick-u!,

NB: The skid is not shown, but is

fixed to the housing near the stylus

Figure 36. Basic design elements of pick-ups of the analogue transducer configuration. ((ourtesy of Taylor Hobson.)
that when the armature is centrally positioned acteristics. The silicone fluid exhibits low stiffness
between the bridge it is balanced, giving no output. and at lower frequencies; therefore large but slow
Movement of the armature unbalances the bridge, stylus displacements allow free movement of the far
providing an output proportional to its displace- end of the crystal, hence no bending occurs and zero
ment; the relative phase of the signal depends on the output is generated. At higher frequencies the fluid
direction of movement. This signal is amplified and stiffness increases, effectively preventing the local
compared with that of the oscillator, to determine in end of the piezoelectric crystal from moving; there-
which direction it has moved from the central (zero) fore the bridge deforms generating an output.
position. It is necessary to utilise an oscillator to A piezoelectric device is a position-sensitive
produce a constant AC output, because the pick-up device. A voltage on the electrodes persists while the
- unlike a motion-sensitive pick-up (Figure 36b) - element is deformed (providing no current is drawn
does not generate any output; it merely serves to from it), due to practical considerations such as
modify the carrier. Simultaneously, the knife edges finite input impedance of the amplifier and cable
exert light pressure from a very weak spring acting losses. These considerations ensure that it can only
on the beam, enabling subsequent stylus contact be utilised in the motion-sensitive mode.
with the surface under test. The ligaments prevent The moving-coil transducer pick-up (Figure 36bii)
unwanted motion of the beam in the horizontal operates on the same principle as a DC generator or
plane, with the result that stylus movement is only an electric dynamo. In operation, as the stylus
possible normal to the surface being assessed. motion occurs the coil moves inside a permanent
This type of pick-up is also fitted to newer ver- magnet, inducing a voltage in the coil. This voltage
sions of measuring instruments, particularly where is proportional to the velo city of the coil. This design
the range-to-resolution ratio has increased from allows the transducer element (the coil) to be
around 1000 to 64,000. Improvements in electronics coupled directly to the stylus without the need for
for the latest pick-up designs of this type have obvi- an extended arm or hinged beam. Because of its
ated the need for reliance on a skid, meaning that overall size it is somewhat restrictive in use and is
precise pivot bearings have replaced the knife-edge not generally popular today.
pivots. In a similar fashion to the piezoelectric trans-
The piezoelectric pick-up (Figure 36b) has been ducer, the moving-coil pick-up does not measure
widely used in the past, but now it tends to be used displacement, but velo city. Stylus velo city has to be
for the less sophisticated hand-held measuring integrated in order that the absolute position of the
instruments. When the stylus deforms the piezo- stylus can be determined. As a result of the coil
electric crystal, it has the property of developing velo city rather than measurement of stylus displace-
a voltage across electrodes, the advantage being that ment, integration errors will increase as the signal
it is virtually instantaneous. As the stylus follows frequencies decrease. Hence, this type of pick-up is
the contours of the surface, the piezoelectric crystal unsuitable for the measurement of either profile or
distorts by bending a flexure that causes crystal waviness (low characteristic frequencies). Therefore
compression and it becomes charged. The resulting the piezoelectric transducer is only suitable for the
charge is then amplified and electronically inte- assessment of roughness. Yet another disadvantage
grated, producing signals proportional to the surface of the moving coil transducer is its poor linearity. It
profile. A piezoelectric transducer exerts a propor- should only be used in the form of a comparator by
tionally larger stylus force to that of an equivalent comparing similar component surface roughnesses
inductive transducer, with the former pick-up poss- against a known roughness calibration standard.
ibly damaging softer and more delicate surfaces.
The operating mechanism of this piezoelectric 2. Digital transducers (Figure 37)
transducer can be described in the following With stylus motion along the surface under test,
mann er: the flexible ligament interposed between pulses occur that correspond to multiples of the
the stylus arm and the piezoelectric element has transducer resolution which are fed to an up-down
enough stiffness to transmit normal vertical motion electronic counter that displays the gauge displace-
of the stylus to the crystal, but will flex as the stylus ment, with its range being determined by physical
is suddenly subjected to instantaneous shock and in constraints of the gauge. Typical range-to-resolution
this way protection of the somewhat fragile piezo- ratios are of the order of 700,000, hence the advan-
electric element is achieved. A light spring force tage is that there is no need for range switching,
provides downward force on the stylus, keeping it in enabling the maximum resolution to be available
contact with the surface. A small drop of silicone oil over the complete operational pick-up range. The
is held by capillary force between the seeuring block displayed displacement is relative to the stylus posi-
and a thin metal blade fastened to the re ar of the tion, where the counter is zeroed. This means that
housing, giving the pick-up suitable dynamic char- every time the electronics are switched off and on
 Industrial metrology , .

(a) Non-contact laser pick-up. (Courtesy ofMahr Perthen.)

collimating lens detector

--~--~------,-------------~

beam splitter
slI rfacc 1I.n d er t t

(b) Phase grating interferometric (PGI) pick-up. (Coutesy ofTaylor Hobson.)

OUTPUT
BEAMSPLITTER

r
\
LASER DIODE
STYLUS ARM & COLLIMATOR
CONVEX DETECTORS
DIFFRACTION
SPRING GRATING

Figure 37. Basic design elements of pick-ups of the digital transducer configuration.

again, the counter must be zeroed at a particular the measurement reference. Interference patterns
datum position. In a similar mann er, if a certain are detected by multiple photodiodes, enabling
maximum stylus speed has been exceeded the an output signal to be interpolated, typically
counter may lose count, requiring zeroing again at giving a basic resolution of 10 nm with an oper-
the datum position. ating range of around 6 mm. A focal point of 2
The following pick-ups are typical of those f.Lm is typical, allowing soft or delicate surfaces to
currently in general use for surface texture instru- be assessed.
mentation: • Phase grating interferometric (Figure 37b). This
pick-up has been developed to complement the
• Non-contact laser (Figure 37a): These optical laser interferometric type of pick-up as it offers
configurations vary depending upon the meas- a greater range, with a corresponding reduction
uring instrument company, but most systems in physical size, by employing a laser diode
utilise a miniature Michelson laser interferometer instead of agas (HeNe) laser source.
with a laser wavelength of 632.8 nm to provide
 The non-contact laser pick-up (Figure 37a) oper- The geometry of a skid can vary, from either a
ates on the following principle: infra-red light is large radius (Figure 38b) to a flat (Figure 38c). The
emitted by a laser diode; its path through the optical skid supports the pick-up, which in turn rests on the
system to the micro-objective is shown in Figure surface. The relative motion of the pick-up as it
37(a), with the beam being focused onto the surface traverses along the workpiece causes the skid to act
to a focal point of 2 fLm diameter. Light is reflected as a datum, giving rise to a "floating action" of rising
back by the workpiece surface, returning along an and falling of the pick-up (Figure 38b). The macro-
identical path, later being deflected onto a detector, irregularities of the surface, namely of waviness and
whereupon an electrical output is generated that profile, means that the stylus with its significantly
corresponds to the distance between the focal point smaller radius to that of the skid will follow this
and the surface feature. A powerful linear motor surface topography, sinking into valleys that the skid
continuously readjusts the hinged measuring lever bridges. However, if the crests and valleys in the
during the measuring cyde, enabling the focal point workpiece surface are too widely spaced, then this
to coincide with the surface feature(s) being distance with respect to the skid motion (Figure 38a)
measured. The focus follows the surface as its will also cause the skid to rise and fall and thereby
motion translates across that portion of workpiece lose the datum registration plane (Figure 38a). If the
under test, in a similar fashion to a stylus-based distance between the stylus tip and the skid (Figure
pick-up. The vertical movements of the measuring 38a) is half the waviness spacing, then these wavi-
lever are converted into electrical signals by an ness amplitudes are doubled. Conversely, when the
inductive transducer. Due to the relatively small size distance of the stylus tip (Figure 38a) corresponds
of the point of focus (approximately 2 fLm), quite exactly to that of the waviness spacing, then the
minute profile irregularities can be assessed. waviness is mechanically eliminated. The skid type
Some caution should be expressed in general of pick-up is only marginally affected by instrument
about non-contact pick-ups, as the surface features vibration, because the measurement reference is
being assessed can have a tendency to act as situated dose to the point of measurement. There-
secondary sources of light. When the radius of fore, the mechanical relationship of the pick-up-to-
curvature of a part feature is smaller than 10 fLm, skid system may be an important factor in its design.
diffraction effects appear which can affect the If different pick-ups are utilised for the same
beam's edge. This is not a serious problem for large measuring task, then in extreme situations these
surface features but becomes crucial for finer surface measured differences could approach that of 100%.
finish es, illustrating why optical instruments te nd to Ideally, the skid and stylus should be coincident,
produce higher values for surface texture than their thereby eliminating the "Abbe error"; the Abbe prin-
stylus-based equivalents. Other limitations for non- ciple states that the line of measurement and the
contact pick-ups are that they cannot easily measure measuring plane should be coincident. For practical
small bores, they traverse widely changing surface reasons, the stylus is slightly displaced from the skid
features, and they cannot "sweep aside" dirt on the and is situated either in front or behind the stylus.
surface, unlike stylus instruments. Phasing errors can be induced by the skid (Figure
The phase grating interferometric pick-up 38b), causing distortion of the graphical trace, while
(Figure 37b) has positioned on the end of the arm its magnitude depends on the waviness crest
of the pivoted stylus a curved phase grating, which spacing. In practice, if the waviness crest spacing is
is the moving element in the interferometer. The identical to that of the skid/stylus spacing, the skid
pitch (grating wavelength) provides the measure- will rise on one crest, with the stylus rising on an
ment reference. An interference pattern is detected adjacent peak. This relative motion completely
by four photodiodes, enabling the output signal to suppresses the peak. When the spacings differ, the
be measured, giving a basic resolution of 12.8 nm effect illustrated in Figure 38(b) occurs, where a
with a range of 10 mm - ensuring that it is a useful spurious valley is introduced into the graph at a
pick-up for form measurement. distance from the peak, this being equal to the
skid/stylus spacing. If two skids are used on either
side of the stylus, then the "phase effect" of the single
skid/stylus spacing is diminished. In some pick-ups
1.12.2 Skid or pick-up operation the line of contact of the skid is laterally offset, which
prevents the skid from burnishing the track
Two types of pick-up operation are normally (mechanically deforming asperities and giving a
employed for surface texture assessment: false reading) during repeated traverses; this is
particularly relevant when the instrument is auto-
• with a skid; matically setting the vertical magnification (auto-
• skidless with a reference plane. ranging).
 InClustrial metrology , .

(a) A rounded skid does not provide a serviceable datum,

if the surfaee irregularities are more widely spaeed

Skidmotion
=::::> Motion

(b) The "phase effeets" when using a skid

=::::> 1I'lotion

Trace

(e) Diagram iIIustrating an independent datum

An independent datullll>rovides a . traighl referent!' for the pick-UI>

TnlVcr 'e datum

Skid Illotion

r=---';~::=:=:;=;=::;-- Dlltum skid

0l>tical nat

Figure 38. Configuration of surfaee texture instrument and relative motion of the stylus with respeet to an independent datum. (Courtesy of
Taylor Hobson.)
 Surface texture: two-dimensional

The skid acts as a mechanical filter, modifying

the profile to a greater or lesser extent. Its radius
of curvature, therefore, should be suitably chosen to
minimise filtering effects. Generally, if a skid is
selected with a radius of curvature of approximately
50 mm, this is normally sufficient for general-
purpose applications.
The skidless or reference plane technique can
take the form of two distinct types:

the pick-up skid travels along a Hat surface such

as an adjacent optical Hat, or similar and not on
the surface being measured;
2 a shaft which moves in an accurate straight line,
with the pick-up being rigidly fastened to it -
usually a glass block providing an independent
datum.

The main reason for employing skidless pick-

ups are that they do not eliminate the contributions
of waviness and profile errors from the measured
topography - which the skid obviously does - and
often waviness, in particular, is important to the
manufacturing process and must be included in
the assessment. The skidless technique provides
an undistorted assessment of the actual profile, the
proviso being that the reference surface has ideal
Figure 39. Portable hand-held surface finish instrument "Surtronic
form.
Duo" can be used at any angle/attitude. ((ourtesy of Taylor Hobson.)

large and clear metric or imperial LeD display

1.12.3 Portable surface texture provides simple-to-read surface-related information
instruments and instrumental status data, with the following
specification:
As the assessment of surfaces is not confined to
those inspected in a metrology laboratory and often • "standard" cut-off (0.8 mm ± 15%);
entails the measurement to be made by both produc- • accuracy: Ra (range: 0.1-40 fLm ± 5%); Rz (range:
tion and maintenance personnei, then some easily 0.1-199 fLm ± 5%);
portable yet robust instruments are necessary. One • diamond stylus (5 fLm radius);
such instrument is that shown in Figure 39. This • traverse length 5 mm;
battery-operated surface text ure measuring instru- • minimum bore penetration 65 mm.
ment can be used across a wide range of angles
and attitudes, increasing the range of versatility of Portable and hand-held instruments of this type
the equipment. This instrument has a piezoelectric (within the specification and parameters of the
pick-up which is self-calibrating and can be used in equipment) can establish the surface condition,
remote operation - split mode - through the infra- without the need to break-down assemblies, or the
red (IrDA) link, up to a distance of 1 m without the necessity of removing the surface in question to a
inconvenience of cable supply. This is an important dedicated and more sophisticated instrument. Such
factor when attempting to measure otherwise in- versatility and portability in surface measurement
accessible component features such as bores. when combined with the confidence in knowing that
When in transit the stylus has proteetion, as the a representative roughness has been established,
"park" position fully protects the stylus assembly. albeit for either Ra or Rz, helps in the interpretation
Digital circuitry via surface mount technology pro- of component topographical information from
vides both reliability and accuracy and the stylus normally inaccessible positions, when compared to
assembly closes up (in its robust case) to fit either that of conventional desk-based equipment. They
in the pocket (weight 200 g), or on a belt, which is are, however, limited in their surface analysis and
useful for off-site surface evaluation in the field. The parameter scope.
 InClustrial metrology , .

1.12.4 Surface form measurement • as mentioned above - instead of using a straight

datum (skid) the gauge is guided by a curved
datum (Figure 41c);
Curved datums • the gauge is held stationary and the surface under
it is slowly rotated at a peripheral speed sufficient
In the previous discussion, all the surfaces have been to gather surface data information;
nominally considered to be flat with only marginal • the gauge is traversed concentrically around the
form present. There are occasions when it is neces- stationary workpiece surface;
sary to measure the surface texture of curved • a wide-range gauge is utilised to enable it to allow
surfaces such as balls, rollers, the involute curvature for the curvature of the workpiece's surface
of gear teeth and fillet radii of crankshafts (see profile as it traverses the part.
Figure 40). In the past a wide range of gauges had to
be available with sufficient resolution to assess the The inspection of a curved surface offers a par-
surface texture together with form, while setting up ticular challenge, which can be achieved by guidance
the surface under test in such a manner that the of the stylus gauge along a curved path having
geometry of its shape did not influence the actual the same radius as the part surface. The curved
surface texture measurement. surface measurement condition is attainable if
If the limiting factor in the determination of the datum kits are used (allowing curved surface paths
surface texture was due to a large radius, then this to be traced by the gauge - see Figure 41b). These
problem could be minimised by utilising a short cut- datum kits tend to be time-consuming in setting
off length and a skid fitted to the gauge to reduce up, necessitating a degree of technical expertise in
the measurement loop. The skid will induce prob- both set-up and alignment. The curved surface is
lems in obtaining a successful and representative rotated within the datum kifs conical seating against
surface texture measurement, if fitted in front or the stylus of a gauge, the stylus being incorpor-
behind the stylus - the former technique is illus- ated into the conical seat housing (Figure 41b).
trated in Figure 41(a). However, the stylus displace- The advantage of this set-up is that the tip of
ment from the measuring plane can be negated by the stylus is always aligned to the centre of the
introducing a skid which surrounds the stylus. This component, with the likelihood of tip flanking
tip projects through the skid's centre (see Figure on the part being eliminated. Datum kits are
41c), thus obeying "Abbe's principle". The solution normally only practicable when the radius is
to curved surface measurement can be addressed by constant during measurement, but if aseries of
utilising several of the following types of equipment, radii occur then under this situation they are not
or techniques: to be recommended. Today, such datum kits are
not normally used due to the necessity of long
set-up times.

Form analysis
The introduction of wide-gauge ranges on high-
resolution pick-ups and the development of various
form options (reference line fitting) has meant that
guidance of the pick-up along a curved path has
been virtually eliminated. Typical of these surface
form-fitting features that are currendy available are
the following software-based options:

• removal of a least-squares best-fit radius;

• removal of aspheric form from the raw data -
form error, surface slope error and tiIt compar-
ison with operator-designed data in the form of
a polynomial expression (see Figure 42a);
• elimination of elliptical/hyperbolic (conie) geo-
metry - provides major and minor axes values
Figure 40. Form measurement with an inductive contour pick-up, and tiIt, with residual surface texture analysis
including "customer-engineered" work-holding, for crankshafts. after removal of best-fit elliptical and hyperbolic
((ourtesy ofTaylor Hobson.) forms (see Figure 42b);
 (a) Stylus displacement solely due to surface curvature (i.e. d 1 + d2)

Pick-up traverse
---c:::o-

(b) Contour measurement-principle of rotating workpiece instrument

Light pressure
I Traverse
Weighted friction
arm

Optical flat datum Precision-ball

Stylus

(c) Side skid gauge - the stylus projects through the cylindrical form of the skid

Figure 41. Effect of curved surface measurement. ((ourtesy of Taylor Hobson.)

 InClustrial metrology , .

(a) Aspheric form software (b) Conical form software

(c) Dual profile software

Figure 42. Screen displays of ''Windows™-based'' software options available forform assessment. ((ourtesy ofTaylor Hobson,)

• complex symmetrical profile assessment - VIa eomprehensive list ean be enhaneed still further. In
an enhaneed "Dual Profile" faeility, allows the addition to a full range of surfaee texture parame-
measurement profile of a master shape or proto- ters, form analysis software ean provide:
type eomponent to be saved as a template, en-
abling future eomponents to be measured and • form error - ealculation with referenee to best-fit
simultaneously displayed along with template eoneave or eonvex cireular are, straight line
data for immediate eomparison (see Figure 42e); measurement, including surfaee roughness detail.
• analysis of gothic areh profiles - frequently em- Alternatively, referencing to the minimum zone,
ployed on the ball traeks for recireulating ball- this being the minimum separation between two
serews; parallellines eontaining the data set;
• three-dimensional eontour analysis - a represen- • radius - using the least-squares best-fit, coneave
tative 3D visualisation of the part surfaee: axono- or eonvex cireular ares ean be automatieally
metrie projeetion, eolour height distribution, ealculated from seleeted data, with the option of
wear simulation and form rem oval, with many being able to exclude any unwanted surfaee
additional features available (this topie is a features from the seleeted data. Conversely, the
subjeet on its own and diseussed in Chapter 2). absolute radius ean be set to analyse aetual devi-
ations from a design master, with other ealculated
The number of surfaee texture parameters avail- parameters including its eentre eoordinate and
able to define a surfaee's eondition is immense for the piteh;
form-measuring instruments. Typieally, these para- • angle - using a straight edge or minimum zone
meters include 22 profile; 22 waviness; 25 rough- algorithm, surfaee tilt ean be established and then
ness; 12 R+W; 9 aspherie parameters; and this removed, prior to parameter analysis, with other
 u ace texture: two- Imenslona

calculated values including its intercept and • comparison of DFX files to the contour - if
pitch; measuring a known contour "best-fit" individual
• dimension - the linear relationship of surface geometrie elements are calculated with reference
features can be assessed and then compared, to a template and the comparison of each point
owing to the ability to calculate both the true in the contour is assessed against the nearest
X- and Z-coordinate positions. element in this template;
• geometrie element fit to an unknown eontour -
when measuring components of an unknown
Contours: profile curving, or an irregular- contour, the software finds geometrie elements
shaped figure that optimise the geometry of the profile
determining the size, position and relationship
A typical commercial system is currently available between profile features, plus other functional
for the analysis of contour measurement; assessment factors.
of features such as radii, angles, length and height
(see Figure 43 for a typical application) can be The most re cent form and surface texture
achieved with: analysis equipment has absolute positional control,
offering sophisticated high-technology instruments
• measurement maeros - these can be learnt and providing complex analytical abilities in a single
edited forming user programs for repeated traverse of the workpiece. Lately, with the high
inspection routines, which minimise repetitive thermal stability of the pick-up gauge (phase grating
operator input (achieved via aseries of definable interferometer - PGI), it typically has a large range
"fastkeys"); with around nanometric resolution, equating to a
• individual feature tolerancing - allowing indi- range-to-resolution of 780,000:1. One of the current
vidual seetions of the measurement template to developments in stylus proteetion employs a "snap-
be toleranced with a variety of values (a wide off" connector, allowing the stylus to be ejected from
tolerance on a Hat area of a surface can be the gauge if the loading stylus force exceeds safe
defined, with tighter tolerancing to the radii); limits - reducing possible damage to the gauge
because of operator error. Interchangeable styli can
be fitted to such pick-ups, for better access to the
component feature being inspected. If the part is of
a delicate nature and liable to deform, or could be
easily scratched, then non-contact gauge heads can
be utilised. These non-contacting types of gauges
comprise a focus follower with a programmable
controller for fast, simple use, replacing inductive
pick-ups. Non-contact gauges are used where the
component surface applications consist of:

• soft and touch-sensitive surfaces - printing

pastes and coatings;
• wet or dry solder pastes and printed conductive
pastes;
• contact and intra-ocular ophthalmie lenses;
• and many other industrial/scientific applications.

Due to variations in the orientation of the work-

piece under test to the traverse unit, some of the
latest versions of form analysis instrumentation can
be operated in inverted attitudes, or right-angle
attachments can be fitted for measurements between
shoulders on components, such as on crankshafts
(see Figure 40).

Figure 43. Form and surface measurement instrument "Taly-

contour" being utilised to assess a spherical bearing outer race.
((ourtesy ofTaylor Hobson.)
1.12.5 Non-contact systems they have yet to be developed for their three-dimen-
sional counterparts. Yet despite this two-dimen-
sional standards infrastructure, there is a distinct
need for both good practice and a clear under-
Contaet and non-eontact operational standing of which surface texture parameter is most
aspeets on surfaees suitable for an industrial application to pass down
to the shop floor - relating to standards.
As has been previously mentioned, in contact For non-contact instrumentation, if one con-
systems the stylus contacts the surface utilising a siders some form of optical inter action with the
precisely manufactured stylus and, today, it is invari- surface that enables local height variations to be
ably diamond-tipped. However, owing to the stylus established, then it will sampie a different surface
geometry - its shape and size - some styli on certain from that of stylus-based instruments. As a practical
surfaces will not be able to penetrate into valleys and example of this difference, if consideration is given
as a result create a distorted or filtered measure of to the optical assessment of metals, their surface
the surface texture (see Figures 11 and 12). Arecent homogeneity (different phases) could introduce
study has indicated that the actual radius of curva- apparent height changes up to 10 nm due to the
ture of a stylus can be very different from its phase change on reflection; whereas glasses or
nominal value, while the effect of stylus forces can ceramies may have local refractive index changes
significantly affect the measured results, with too and contaminant films introducing nanometric
high a force causing surface damage. changes. While phase changes at the conducting
The stylus, to enable a representative surface to surface can introduce variations as a function of
be assessed, must traverse across the surface incident angle, surface field anomalies that are the
following a parallel path with respect to the nominal result of multiple scattering, or sharp points acting
workpiece surface. The datum in this case would as secondary sources of light, together with various
most likely be a mechanical slideway of some sort edge and wall effects, will all introduce measurement
(see Figure 38). The datum comprises a skid that has uncertainty.
a large radius of curvature (spherical or different It was briefly stated that optical methods do not
radii in two orthogonal directions), fixed to the end have a standards infrastructure, unlike that of
of a hinged pick-up (see Figures 36 and 37). At the stylus-based surface texture instrumentation,
front end of the pick-up body the skid rests on the meaning that there exists no formal techniques for
workpiece surface; alternatively other designs occur, calibrating optical instruments. Therefore, as a result
such as a flat shoe being free to pivot so that it can of this lack of calibration methodology, care should
align itself to the surface, or two skids are situated be taken when optical assessment of a range of
either side of the stylus. It should be reiterated that surfaces is to be undertaken, particularly if they are
ISO 3274: 1996 does not allow the use of a skid, of differing physical surface characteristics. For
but many are still used in metrological applications example, if a glass step-height standard is utilised to
in industry. calibrate the vertical magnification factor of an
Notwithstanding the long his tory of stylus- instrument, it would be unwise to measure a metal
based instrumentation a number of problems exist, surface set at an identical calibration value. More will
associated with either their operation or the inter- be said on the topic of calibration in Chapter 6.
pretation of results. For example, none of these
instruments measure the surface alone: if an inho-
mogeneous workpiece is inspected utilising a Laser triangulation
mechanical stylus it responds to the topography and
changes in the surface mechanical properties, such The laser triangulation, or chromatic abberration
as its elastic moduli and local hardness. Another techniques can be utilised to measure surface topog-
important aspect in contact surface texture inspec- raphy, which are dependent on the sensor model
tion techniques is the scale - horizontal and vertical employed. Cut-off filters available range from 0.8
- although it is not common for instrument manu- mm to 8 mm with large measuring envelopes of up
facturers to incorporate metrology into the scanning to 150 mm by 100 mm, enabling large or multiple
axes of their instruments, with many companies arte facts to be scanned within the same measure-
failing to calibrate scanning axes. ment period. A typical instrument will have a
Stylus-based surface texture instruments have vertical range of 300 fLm and resolutions of 10 nm
an infrastructure of many specification standards, from a spot size of 2 fLm in diameter, with scanning
unlike opticallnon-contact instrumentation. This capabilities of up to 2000 points per second. The
point regarding standards for stylus instruments measuring ranges currently available are from 0.3
only embraces two-dimensional measurements and mm to 30 mm, depending on the sensor model.
 (a) Specular reflection Optical techniques: reflection and
scattering
Highly reftective surfaces are norma11y assumed to
be very smooth, so by measuring reftectivity an
indirect technique for measuring roughness can
be established. Reftectivity testing is particularly
s
beneficial where visual appearance is important
(cosmetic items), or on soft surfaces such as
Angle of paper where non-contact measurement is necessary.
reflection Advantages of optical methods include assessing and
averaging over an area and high-speed inspection.
Smooth surfaces invariably appear to be visually
Flat reflective surface "glossy" (reftective); conversely, a rough surface
has a matt appearance. In the past, the gloss meter
(b) Diffuse reflection principle was used to verify how parallel beams of
incidence light at an angle on the surface are reftected.
If a perfect1y reftective surface is present (Figure
44a), then much of the light will be specularly
reftected (angle of reftection equals angle of inci-
den ce) and enter the viewing system S, with no light
entering the viewing system at D, being positioned
perpendicularly above the surface. However, if a
perfect1y matt surface is inspected (Figure 44b), the
light is diffusely reftected (scattered in a11 direc-
tions), causing equal amounts of diffused light to
enter both Sand D. The ratio of specular-to-diffuse
reftection is a measure of the gloss of the surface.
The relative proportions of light entering Sand D
Matt, or rougher surface
can be measured by either visual means or photo-
Figure 44. Diagram illustrating the principle of surface assessment
electric comparison, incorporated into a suitable
by speeular and diffuse refleetion. ((ourtesy ofTaylor Hobson.) portable instrument. As it is the ratio of the light that
is measured, any absorption of light by the surface
will not inftuence the result, and neither will ftuctu-
ations in the irradiance distribution of the lamp, or
One such system has a tested repeatability of 0.03
more specifically today, the laser.
fLm, but this claim applies to a specimen having
Considerable interest has been shown over the
a 2.90 fLm Ra roughness, measured with a 2.54 mm
years in the use of scattering for surface texture
cut-off length. Optional equipment can be added
measurement. In essence, three techniques are
to investigate feature loeation and alignment or
generally used for scatter-based instrumental
vacuum stages can be incorporated to hold
design:
awkwardly-shaped arte facts and replication kits can
be supplied to measure inaccessible or difficult to
assess surface features. 1 light scattering theory, when used to construct an
It is claimed that these systems can measure instrument giving absolute measurements of
surface colour or texture. These instruments offer surface text ure - when a number of conditions
fast and multiple scanning facilities that reduce the are met;
time taken for each measurement, compared to 2 general theoretical approach in instrument
conventional contact instrumentation. Common to design - assumptions are made that have general
all optical equipment (as described in the following connection with a particular surface text ure
section), light scattering causes measurement accu- parameter;
racy problems in some component surfaces. If a 3 application-specific approach, where instrument
fixed resolution occurs for each sensor model, this design is developed to solve an immediate
would imply that up to six sensors would be required problem.
to enable satisfactory coverage of all surface
textures, or types of measurement needed for a In just one example, an in tegrating sphere (Figure
complete metrological range of applications. 45) can be used to measure gauge block surface
 Integrating
sphere

beam

Specimen
over
sampIe
port

Output
port

/ ~ B,m,
Detector Photodetector
port

Figure 45. Geometry and layout of an integrating sphere. [(ourtesy of Dr R. Leach/NPL.j

texture. The instrumental technique measures the Interference instruments

total intensity of light from a gauge surface, then
compares it with the light intensity that was diffusely Surface texture interferometry occurs in several
scattered. The ratio of two intensities is termed total instrumental configurations, probably the most
integrated scatter (TIS) and it is proportional to the common types being interference microscopy,
square of the RMS surface texture parameter Rq. plus full field methods. Becoming popular of la~e
This method relies on a number of assumptions that are phase-stepping techniques and swept-/multl-
must be met for the type of surface to be meas- frequency source methods. Today, there are several
ured. The diffuse component of reflected intensity commercially available phase-shifting interferomet-
is measured by collecting light in an integrating ric microscopes, giving three-dimensional surface
sphere. An integrating sphere can be practically imaging with very short measuring tim es. The
represented by a hollow sphere that is coated inter- National Physical Laboratory (NPL) has developed
nally with a highly diffuse reflecting material. Any a sub-nanometre resolution system that employs a
light reflected from the inner sphere's surface gives microscope objective with a numerical aperture of
rise to a constant flux of electromagnetic power 0.5, in conjunction with a birefringent lens providing
through an element of the sphere. This sphere has a a focused spot from a laser source and a defocused
number of ports, allowing radiation both in and out, orthogonally polarised beam 10 /-Lm in diameter.
with the gauge to be irradiated and photodetectors Fourier analysis has been obtained from surfaces,
to detect the intensity of integrated light. indicating that with this objective the system pro-
duces a surface wavelength range from 0.5 to 15 /-Lm.
The trace displayed by the system is an interfero-
metrically obtained path difference of the focused
probe beam and the defocused beam.
 (a) Basic elements of the instrument
A: mercury vapour lamp;
B: green filter;
C: iris diaphragm;
D: lens system;
E: semi-reflector;
F: objective;
H
~
'
I
m I
,
G: indined reflector plate;
H: graticule;
I: surface under test.
I
i VB A

E~~~ ~~_
D

__

Inclined reflector
(optical) plate

Figure 46. Fizeau-type surface texture interferometer. ((ourtesy of Taylor Hobson.)

As has already been mentioned, light interference cause phase changes in the light being reflected back
can be successfully used as a non-contact optical to the observer. The result of this reflection will be
technique to assess surface features. Essentially, the to promote alternate dark and light interference
principle of interference of light relates to how light fringes. The shape and spacing of these fringes will
rays are reflected between two surfaces; the differing depend on the regularity of both the surface and the
light path lengths along various points of the surface reference reflector. The surface texture irregularities
are reproduced as modifications in the interference may be slightly inclined to the work surface I, allow-
pattern and, when viewed under specific conditions, ing a wedge-shaped air gap to occur (detailed in
displacement of the fringes is a measure of the Figure 46b). Multiple reflections between the surface
roughness height. and the reflecting plate produce good-contrast,
For some interferometer instruments such as the sharp fringes (see Figure 47 for typical interfero-
Fizeau type, interferometric surface assessment metric images) wh ich are viewed in the eyepiece.
relies on the fact that a "wedge of air" occurs on the Utilising a spherical semi -reflecting surface, it is pos-
test surface, with the resulting interference lines sible to examine three-dimensional curved surfaces,
being of equal height and their respective height such as precision ball surface finishes (Figure 47c).
differences given by Multiple re fleet ion (Figure 46a) on the semi-silvered
surface E and the workpiece I means that all on-
Fringe spacing = Al2 coming beams are split into several partial beams,
which cause interference. Interference fringes that
where A = wavelength. occur are both higher in co nt rast and narrower as
Under examination, the surface roughness the reflection coefficient becomes greater. If too
produces air wedge thickness variations causing great a reflection coefficient occurs, the co nt rast will
deflection of the interference fringe, from which the degrade. Such conditions allow multiple beam reflec-
total height parameter Rt results. Hence the total tion to occur only when the distance to the reflection
height becomes surface and that of its test surface is quite small.
Large apertures can be utilised by this Fizeau
Rt= aIA(Al2) surface interferometer, enabling full exploitation of
the light microseope. When employing high aper-
where A = interference fringe, a = fringe deflection tures, the line standard (graduated scale) is depen-
and Al2 = fringe spacing. dent on the selected aperture and, considering the
The height of irregularities that can be resolved associated aperture angle, the fringe spacing is
range from 0.005 to 1 f.1m using conventional inter-
ferometric methods. The range can be extended by Fringe spacing = A/2 (2/1 + cos fL)
using the technique of oblique incidence, which
increases fringe spacings from Al2 to 2A, enabling where A is the wavelength, fL is a eosine error.
one to measure a surface roughness height four times Normally, the largest aperture (A) utilised is 0.65,
larger than that measured using a standard interfer- hence the fringe spacing equation above can be
ometric method. Furthermore, if cast replicas of the simplified to
test surface are made, then immersing them in fluids
of a suitable refractive index enables transmission Fringe spacing = 1.14A12
interferometry to assess roughnesses up to 25 f.1m.
Maximum contrast is important for optical inter- Generally, it can be said that applications of this
ferometry, and the wavelength of emitted light must particular Fizeau surface texture interferometer
remain constant to eliminate variations in the phase have limited use, although the technique does have
difference between interfering beams. Any defects in some merit:
optical surface quality within the interferometer can
promote areduction in fringe contrast, as can vibra- • it can examine a relatively large specimen surface
tions of optical components, together with parasitic area;
light caused by unwanted reflections. Due to the • it indicates any form error present;
strongly convergent light utilised in these interfero- • being a non-contact method, it may be possible
metric optical configurations (Linnick and Mirau to estimate the scratch depths.
interferometers), this can lead to obliquity effects,
which can introduce errors in estimation of, say, a An optical and practical limitation of this type of
scratch depth by as much as 12%. The magnitude of interferometry when applied to precision surfaces
error is dependent on the numerical aperture of the occurs due to visual viewing, which tends to be time-
objective and the surface's scratch depths. However, in consuming and somewhat fatiguing. By using auto-
the case of more commonly available full field tech- mated image analysis techniques for detection, this
niques this obliquity and hence error is not a problem. limitation could be overcome, but at a considerable
In essen ce, the Fizeau surface texture interfero- cost dis advantage to the user.
meter is basically a microseope, with a built-in Although three-dimensional surface texture
illumination and special-purpose optical system analysis will be discussed in Chapter 2, one instru-
(see Figure 46a). Between the objective and the mental technique has been included in this chapter.
surface a semi-reflector G can be positioned, which An automated surface profiling interferometer is
 u ace texture: two- Imenslona

surface texture is used to control the product's

performance. Until recently, microscopes tended to
be utilised purely for attribute sampling, such as
imaging surface topographical details, while surface
profilers provided accurate measurements to charac-
terise the surface details. One of the latest develop-
ments in surface texture (white light) interferometers
is illustrated in Figure 48, which combines these two
technologies, providing a fast, quantitative surface
measurement on many types of surface topogra-
phies. The surface details that can be quantified
include surface roughness, step heights, critical
dimensions and other surface topography features
within a matter of seconds. Profile heights that can be
assessed range from >1 nm up to 5000 fLm, at speci-
men translation speeds up to 10 fLm/s with a 0.1 nm
height resolution, which is independent of both the
magnification and feature height. This interferome-
ter can resolve sub-micrometre X- and Y-plane fea-
tures on profile areas up to 50 mm 2, achieving this by
image stitching. This stitching technique has been
designed to analyse surface details much larger than
is possible with a single measurement. Stitching
occurs by taking several measurements of the test
specimen as it is moved by a motorised stage and
then combines - stitches - the multiple data sets into
one surface image. In effect, stitching increases the
field of view, without compromising lateral or verti-
cal resolution. The scanning actuator is of a closed-
loop piezoelectric variety, employing low-noise
capacitance sensors to ensure that both accurate and
repeatable linear motion occurs over the whole oper-
ational range. Other notable modes include both
(b)
"Phase Stepping Interferometry" (PSI) and "White-
Light Scanning Interferometry" (WSI) techniques .

Use ofspeckle

(c)
(i) (ii) • (iii)

Figure 47. Typical interference fringes from a surface texture inter-

ferometer, showing both flat and spherical surfaces ((ourtesy of
Taylor Hobson).
This technique for surface texture inspection utilises
partially coherent light, with the reftected beam
from the surface consisting of random patterns of
dark and bright regions termed speckle. The spatial
patterns and co nt rast of the speckle depend on the
optical system configuration utilised for observation
and the condition of coherence of the surface
(a) Lapped surface of a gauge block, showing a scratch. Fringe spacing texture. Speckle has two important attributes:
is 0.25 J.1m. Due to a scratch extending across adjacent fringes, its
depth must also be 0.25 J.1m. (b) Lapped gauge block edge, indicating
• contrast;
edge rounding. (c) Interference patterns on spherical surfaces (i) pitch
polished steel ball; (ii) commercial steel ball; (iii) glass sphere.
• number of spots per unit area.

The contrast can be shown to have a relationship to

illustrated in Figure 48 that has a wide range of indus- the surface's correlation length, whereas the speckle
trial applications, from research to manufacturing density is primarily concerned with the image
process control. Of particular relevance is its use in system resolving capacity. Information on the speci-
semiconductors, disk drives, printing plates and fuel men's surface can be obtained using the contrast of
injector seals, together with applications where the speckle patterns produced in the first instance, near
 InClustrial metrology . .

(h)
(a)

(e)

Figure 48. Automated non-contact three-dimensional surface profiling interferometer. (Courtesy of Zygo Corporation.) (a) Optical configura-
tion. (b) View of interferometer. (c) Typical screen displays.
 u ace texture: two- Imenslona

(a) Optieal path of SehmaItz microseope (b) Optieal sectioning images, projeeted onto a sereen:
-

Eyc
• I. _............. t'r.~..... ,••

(i) feedmarks from machining, indication cusp heights

- - - --- -

._ ~

I
I

(ii) an engraved line on a ftat surface

(iii) embossed plastic surface

Figure 49. The "5ehmaltz teehnique" - optkal seetioning principle and associated projeeted images. ((ourtesy of Taylor Hobson.)

the image plane, and secondly, dose to the diffraction Cusp height = h x -)2
or defocus plane. Yet another technique utilises the
polychromatic speckle patterns, while others employ where h is the actual profile height (Rt).
the correlation properties of two speckle patterns. Measurements are normally undertaken using
From both of these methods, surface texture values either an eyepiece gratide or an eyepiece incorpo-
ranging from 0.01 to 25 /-Lm have been estimated. rating a micrometer for topographical feature
assessment.
An alternative technique for optical sectioning
Optical sectioning can be achieved by utilising an optical projector, as
this has the added advantage of an enlarged image
The method of optical sectioning termed the being projected onto a screen. The surface's profile
Schmaltz technique after its inventor (sometimes height can then be quickly measured with a specially
termed the light section microscope) pro duces virtu- prepared and calibrated template, which compen-
ally identical results to that of physical sectioning, sates for any distortion introduced by the viewing
but in a simpler, non-destructive manner. However, angle.
the magnification and data obtainable from the The optical sectioning technique is suitable for
projected image are quite limited, when compared surface topographies having a roughness range of Rt
with stylus-based surface texture measuring instru- between > 1 and 200 /-Lm. Moreover, it is eminently
ments. The ratio of vertical to horizontal magnifica- suitable for the inspection of soft and slightly pliable
tion is around unity, meaning that the field of view surfaces (Figure 49biii, which could be deformed by
is small when a large magnification (x400) is used, a stylus), or for estimation of depths of surface
hence for fine surfaces it is rather limited. scratches/engraved lines (Figure 49bii).
The optical sectioning principle is illustrated in The reftectance of a surface is a "sensitive func-
Figure 49(a). The surface to be examined is illumi- tion" of its relative roughness, and consequently the
nated by a thin band of light delineating a profile wavelength of light is considerably greater than
section, which is then viewed at an angle with a the root mean square value of the surface tex-
microscope. Typically, illumination and viewing ture. Hence, the reftectance will depend only on the
angles are 45° to the work surface producing the surface roughness rather than the peak slope of any
dearest profiles (see Figure 49b). For example, the irregularities. Measuring reftectance at two distinct
projected surface indicated in Figure 49(bi) illus- wavelengths enables one to determine the surface
trates an apparent machined profile cusp height of roughness and slope of asperities (peaks).
%10O't/P

X14

XI0
4

3
L_
2

.' ...-/
1

-1

-2

Figure SO. Performance curves for the Schmaltz - optical sectioning - microscope, illustrating the relative error of indication for various micro-
scopes. [After Thomas, 1974.]

Performance tests undertaken for the Schmaltz x30 objective - with the added advantages of a larger
(optical sectioning) microscope (illustrated in field of view and depth of focus. In the case of the
Figure 50) show that for an objective of x60 magni- x30 objective, the relative error of indication does
fication the relative error of indication does not not exceed 8.6%. Now, if one enlarges its measuring
exceed 7.7% over the whole application range (Rt range (1.5-10 /-lm), up to values of Rtof 1.5-14.5 /-lm,
values between 0.8 and 3.0 /-lm). Accepting that prac- a practical accuracy of ±9% may be obtained. The
tical accuracy limits are in the region of 9%, then objective having a magnification of x14 has a rela-
there is little point in using a x60 objective, since tive error of indication within the recommended
the necessary accuracy can be achieved with the manufacturer's guidelines, namely, from 3 to 12 /-lm,
 u ace texture: two- Imenslona

RQP materials; moreover, this technique eliminates the

Frequcnc)' inftuence from variations in intensities of LED.
distribution Other techniques have been developed, for example
angle-resalved scatter (ARS), but the discussion here
will be confined to the current scattering technique.
Variance in the scatter distribution is termed Rap
(the optical roughness parameter). This variability is
proportional to the angle of the surface rough-
ness. Rap can be utilised to compare the workpiece
LED surface roughness against a calibrated standard,
Photo-diode although this parameter can only be validated when
measurement is made under similar conditions. If
component surfaces are quite smooth, then these are
ideal conditions for obtaining the highest sensitivity
Lens from the instrument. Typically, if a component's gen-
eral surface roughness (Ra) lies between 0.05
and 0.5 f.l-m, it is appropriate for evaluation by this
technique. Any surfaces that might be manufactured
by electrical dis charge machining (EDM), electro-
polishing, or grinding (pseudo-random production
operations) are ideal for assessment by the scattered
light technique, rather than other systematic manu-
facturing processes such as turning and milling oper-
ations. Only a brief time period is necessary for an
Figure 51. Diagrammatic representation of sensing with scattered object's surface measurement by the scattered light
light. instrument - typically 300 ms - enabling measure-
ment "on the fty" (while the workpiece is in motion).
varying from 6% to 12%. Moreover, if the measuring A practical application of this technique might be
range is altered for this x14 objective to that of Rt to establish the surface roughness of an in situ
ranging from 5-65 f.l-m, the value of its relative error measurement for a cold-rolled aluminium strip as it
of indication is maintained within the limits of progresses beneath the focused beam at maximum
±5%. In a similar manner, for the x7 magnification strip passing speeds of around 70 km/ho Due to its
objective, by enlarging the measuring range from ability to tolerate relatively fast component motion
8 to 100 f.l-m the relative error of indication will still and quickly estimate surface finish, the scattered
be within a ±4% limit. light technique can achieve 100% inspection on crit-
ical surfaces, enabling almost real-time process
control for continuous processing operations.
Scattered light: Iight-emitting diode
(LED) source Optical diffraction
Utilising "scattered light" is a fast in-process surface The principle behind this instrument's operation is
roughness technique, which is schematically illus- the technique of using a surface under test as a phase
trated in Figure 51. The parallel light is emitted by grating, enabling the instrument (Figure 52) to
an infra-red LED and is projected onto the work- capture the far-field Fraunhoffer diffraction pattern
piece surface. For this instrument, the size of the on a linear CCD detector array, thus allowing statis-
source's light spot is approximately 0.8 mm at the tical assessment of the surface condition.
focal point on the surface. Any reftected light tends A schematic diagram of the instrument is
to be scattered or diffuse in nature, resulting from depicted in Figure 52, with a photo graph of the
the roughness of the workpiece surface. The lens will complete assembly illustrated in Figure 53. The
collimate the light and project it onto an array of system is based upon two laser diodes having oper-
photo-diodes, from where the variance of the distri- ational wavelengths of 670 nm and 780 nm, respec-
bution of scatter can be established. Any variance is tively. These two wavelengths enable diffraction
then normalised, which eliminates intensity varia- order numbers to be identified, in conjunction with
tions resulting from the reftected light. Normalising a low birefringence single-mode fibre terminated by
the variance compensates for the twin effects of an achromatic collimating lens. The purpose of this
different levels of reftection and that of workpiece fibre is twofold:
ml lh~~~~------------------------------1

--- - ----- - - - - - - - - -- ---- --- ,

Optical Fibre

I~,--C-C-D-LA-rr-a-y--',~
Je::::::: / ;;;1
I Ä-' \

1\ Foc~sing ~+
~----8
;' " Ophcs ~ , Figure 53. Assessment of surface texture on a diamond-turned
~
I
I
I
I
0:--
~ -=-::::.:P
I
=> I
component, within the range 5-10 nm Rq, utilising a non-contact
optical instrument. ((ourtesy ofTaylor Hobson.)
I I I I
, , Beam I I
: : Splitter \ i
I ' I eomponents that might be inspected with this

~~~~: :~:~~:::::---~+- :!
instrument include diamond-machined brass (Naval
non-Ieaded type), aluminium (grade: 6061 T6) and
copper (OFHC) flats, germanium flats or similar
I I
Mirror I1 I 1 precision surfaces.
1
Cylindrical Lens 'i I :1 One of the principal reasons for utilising laser-
[
Filter
----...""'r:z:;:I IN based diffraetion techniques is their ability to ehar-
I I1
11
'I
, acterise low Rq surfaces, in particular the advantages
I- - - - - - - - - - -- - - -- - - II- - - r - - -
~ I
obtained from directly filtering the diffraction
samPle ~ pattern. Moreover, this instrument has the capability
of measuring a substantially periodic surfaee to an
Rq of approximately 140 nm. In addition, measure-
Figure 52. Schematic representation of the Talyfine instrument. ments of Rq on a rotating surfaee can be achieved
((ourtesy of Taylor Hobson.) with no distinetion between whether the surface is in
motion or statie (unlike a stylus-based instrument).
• to act as a spatial filter, by selectively transmitting The surface diffraction physics is rather complex
the Gaussian TEM 00 mode; and is therefore beyond the scope of the current
• providing a eonvenient method for the produe- discussion on the general operational and perform-
tion of spatially coincident wavefronts. ance characteristics of this laser-based instrument.

The Gaussian beam for both laser wavefronts is

directed at normal incidence through a cylindrical Microscope app!ications
lens onto the workpieee surfaee. It forms a li ne image
at the beam waist, so that a restrietion of diffraction A cursory examination of a surfaee with a micro-
occurs only on one plane, enabling a range of metro- scope equipped with a graticle will reveal any promi-
logical parameters to be found such as Ra, etc, and nent spacing irregularities. The height of any
the surface power spectrum. Surfaces can be dominant peak cannot be estimated, although if
assessed whether they are random or unidirectional suitable illumination is utilised (see Figure 54 for
in nature. various inspection mo des that can be employed)
Diffraction patterns are produced from the phase some detail can be visually assessed.
modulation of the surface and will be focused onto In metallographical inspection of sampIes, if
a 2048 CCD deteetor. Lastly, data processing is destructive means can be praetised to visually assess
undertaken on a PC, enabling the visualisation of surface features at approximately right angles to any
diffraction patterns, power spectra and surface para- interesting irregularities, this form of visual inspec-
meters. tion reveals considerable information. Rather than
With this instrument the surfaces under test must section the material perpendicularly to the surface,
be quite smooth, in order to obtain a satisfactory a much better and informative strategy is to take a
visual performance from the equipment. Typical taper section of the surface. With taper sectioning,
 u ace texture: two- Imenslona

Op"ca' pa.h
d'.gtlm 0' CF
..
In'lnlfy Correclecl
Ophel l Sy'lem

Tube lens
Adjus'able

!
~
GI
R
._.
E Analyzer. OIe

mmOf bkx::k
~
:5 OtCpnsm
Objee1rve

P,nhOle
apenu'e
doaphrag
,mage
(ex,' pu ptI)
L--.._ .........-J

INSPECTION MODES

B,ightfield Observation Oarkfield Observation Noma,skl DIC Observation

Bnghlf,eld m,croscopy uses d,tfe,ences ,n Oa"<!leld m,c,oscopy detects t,ny flaws , subtle D,ffe,enl,al ,nlerle,ence cont,asl m,c,oscopy
'ellec"on 10 enhance Ihe nalu,al colo, and form ,rregu lan"es. ,mpunhes and delecis on 'he reveals the t,ny lIaws In walers and the subtle
of spec,mens surface 01 wafers or mas S 11 perm,ts hoghly "'egulantres 01 pholO reSlSI panerns as an
accu'a'e ,nspectlons ,nterlerenee Colo, lhe,ebyallowlnglh'ee·
dlmenSlonalobserval,on Die 'ep'esenls
subtle ,nel,natrOn (differential coefftClenl) Wl'h
sharp d erenltal conlraSI , enabl,ng deteclrOn
of very small d,fferences In he'ghl

Fluorescence Observation Simple Polarlzatlon Observation

The epl·f'uo,escence f,lIe, blocks fac"ila'e Th,s methcrd '5 apphed whon Sludy,ng 0'
de'echon and analYSIS 01 resldua' pholO res,s'S analyz,ng speclhc opt'cal charac,onSltCS-
,n Ihe semrconduclor manulaclunng process such as oplocal ,SOIr0PY and an,solJopy-01 a
The 1,lle' delecIS par1'cles as fluo'escenl glven matenal An ,nlerlerence colOr can be
,mages and de,ormlnes lhelf wavelenglh It ,s applted to allow polanzed contrast lfIum,nallon us'ng p,nho!e aper1ure
,deal lor loca"ng phOIO 'es,s' ,rregulantles and observaIrOns Th,s melhod 's 'deally sUlted lor
'he cause of par10cle generalIon observ,ng crystal cond,ltons and detecllng the
slress on a waler

Figure 54. Microscope applications for surface assessment. ((ourtesy of Nikon UK Ltd.)
 Industrial metrology , .

engineering solutions as to reasons why a surface

behaved in a particular manner in service. Engineer-
ing surfaces may fail (in service) for a variety of
unanticipated reasonSj these might include high sur-
face residual hardness, the introduction of unstable
metallurgical changes sub-surface grains exposed at
the surface occuring under conditions of load or
tribological action. Together with a combination of
these factors (see Figure 55a), their failure mode
might need to be investigated.
A stereoscan electron photomicrograph offers a
more informative visual three-dimensional image of
the surface area than can be obtained from a conven-
tional optical technique (see Figure 55b), particu-
(a) larly when additional aids of soft imaging with
topographical height profiles are utilised, as dis-
cussed later in Chapter 3 on surface microscopy.
Optical microscopes have wide-ranging capabili-
ties and offer superb yet subtle techniques to
enhance surface image quality, including the follow-
ing inspection modes (see Figure 54):

• Brightfield - the most frequently used method of

observation, employing differences in reflection
to obtain natural colour and shape of surfacej
• Darkfield - this method becomes of some signif-
icance when observing and photomicrographing
surface irregularities, minute flaws, differences in
height levels and sampies with low reflection
(b) levels (paper, plastics, composites and fibres)j
• Normarski - this technique can capture the sur-
Figure 55. Images fram a scanning electron microscope (SEM) for
face's subtle irregularities and flaws as interfer-
optical assessment of fracture and machined surfaces. ((ourtesy of
Jeol UK Ltd.)
ence colours, indicating them as three-
dimensional forms. Owing to the Normarski's
(a) Mild steel ductile fracture - photomicragraph from an SEM. ability to represent very small tilt with sharp dif-
Magnification x650; ace. V 15 kV; signal SEI; WD 21 mm; SS 28; pres- ferential contrast, this observational technique
sure Pa, (b) Stereo image (left-hand SEM stage tilt - l.5°) of an
can detect minute height differences and subtle
austenitic stainless steel (grade 316) by high-speed machining
(milling with a 0.5 mm feedrate). Magnification x50; ace. V 20 kV;
irregularities in metals, crystalline structures,
signal SEI; WD 20 mm; 55 30; pressure Pa. integrated circuits (IC) and large-scale integrated
(LSI) circuitsj
• Fluorescence - an epi-fluorescence attachment
the surface is prepared at a shallow angle (typically allows for detection of positive and negative
11 0) perpendicular to the surface (see Figure 161). photo resists. Selecting from different filter
This slight taper does not unduly modify the shape combinations offers an optimum configuration
of the surface irregularities, but increases the for any application, being ideal when trying to
apparent magnification of the surface under test, locate res ist irregularities and the cause of
revealing a considerably greater area for visual or particle generationj
micro-hardness assessment (more will be said • Simple polarisation - specific optical characteris-
on this latter topic in Chapter 5). Surface prepara- tics can be investigated with this method of
tion is a difficult task, particularly during the polarised light microscopy. The technique is
polishing phase prior to suitable etching (normally widely utilised for geological research, together
necessary for metallographical inspection of metals) with evaluation and examination of minerals,
as surface profile distortion may occur if sufficient plastics, crystal conditions and stress detection
care is not taken. Sectioning reveals modifications in on IC wafersj
sub-surface microstructural details that might not • Pinhole aperture - allows high-intensity/-resolu-
otherwise be apparent and this will be discussed tion images at large depth of focus, which is
in Chapter 5 on surface integrityj this may offer ideal for surfaces having multi-layer film three-
 u ace texture: two- Imenslona

Confocal microscope
Confocal microscopes gene rally utilise the principle
Harne of depth discrimination. Essentially, the imaging
and receiving optics are identical, offering excellent
properties. In a similar manner to the so-called
"flying spot technique", the optics project a point
onto the surface, with the reflected light being picked
up by a point detector (see Figure 56). The resulting
Source ~ signal is utilised to modulate the spot brightness on
Dichroie the CRT screen, which is scanned in synchronisa-
tion with the object under test. Essentially, the lens
focuses a diffraction-limited spot onto the object,
with the lens collecting light only from the same
vicinity of the surface that was previously illumi-
nated. As a consequence of employing the single-
point detector, both the image forminglreceiving
systems contribute to the signal at the detector.
Thus, by rotating the variable aperture over the spec-
imen, which itself in some systems rotates, it will
Collimaling
build up an optical image (via software data capture)
lens
A
of the sample's surface. By this means, the "flying
spot" covers the complete specimen's surface to the
same optical magnification and resolution of a
conventional microscope, but with the advantage of
"
a significant improvement of the field of view. These
',. images can be optical slices or sections through the
" ' I>ecimen
sm'face surface, which can be processed providing non-
contact three-dimensional information about the
Figure 56. Schematic diagram of an optical profiler. ((ourtesy of Dr surface.
R. Leach/NPL.)

dimensional structures. Typically, the aperture

1.13 Nanotopographie
diaphragm can be "stopped-down" to around instruments
30% of the exit pupil diameter using a x150
objective, providing large depths of field for In the mid-1960s, out of the growing requirement to
surface inspection. measure very smooth surfaces and thin films, an
instrument was developed having aresolution of
Many of the latest microscope systems can be approximately 0.5 nm combined with a maximum
used for image capture and processing, enabling the magnification of 1,000,000, a measuring range of l3
instrument's software to "grab" live images from the /-Lm and an arcuate traverse of ± 1 mm. Originally
microscope via video cameras, scanner computer such instruments were developed in order to
files and archive packages. Such image analysis measure minute step heights, typically around 30
systems enable image enhancement and manipula- nm as illustrated in Figure 57. From this instrument
tion (flip and rotate the surface image), with the and others having enhanced design and perform-
ability to change contrast/brightness, simple thresh- ance refinements came the current genre of nano-
olds, sharpening and smoothing, morphology, technology equipment, typified by the "Nanostep"
component separation, retouching and masking, shown in Figures 58 and 59 (pioneered by the NPL).
together with scaling and filtering. Many software- Instruments of this level of accuracy and precision
enhancing techniques allow operators to take up must be virtually free from thermal expansion
to 36 geometrical and analytical measurements, effects; therefore the majority of the components are
together with reference images for comparison, with manufactured from thermally stable materials, typ i-
Windows TM-based menus for ease of image and data cally glass ceramic "ZerodurTM".
manipulation. Such systems can be networked for The kinematic (translational component mem-
multi-user applications and sophisticated archiving bers) and operation al features of the instrument are
facilities, allowing access by different users. such that the workpiece is supported on a levelling
 InClustrial metrology . .

Step height: aO.03 11m

I

Trace

Silver deposit
Glass substrate

Figure 57. A typical "Talystep" graph of silver deposit on a glass substrate. Thickness of the deposit is approximately 0.03 f.lm. (Courtesy of
Taylor Hobson.)

Figure 58. Constructional detail of the "Nanostep" and its associated Figure 59. Detail of the interchangeable "Nanostep" stylus.
precision slideway in low coefficient of expansion (Le., glass ceramic: (Courtesy ofTaylor Hobson.)
"Zerodur"). (Courtesy of Taylor Hobson.)

table, which in turn is mounted on a carriage. This ment, producing ultra -low instrument noise down to
carriage is equipped with kinematically positioned 0.03 nm under optimum conditions. The motor can
dry polymerie pads, forming an interface with a drive the micrometer at measurement speeds vary-
highly polished precision slideway, giving a linear ing from 0.005 to 0.5 mm/s. The stylus/transducer
translational motion of 50 mm. The horizontal assembly illustrated in Figure 60 can be vertically
micrometer (shown in Figure 58) displaces (via a positioned on the workpiece and in contact with it
"slave carriage") the slideway along its desired via the vertical micrometer (depicted in Figure 58),
length of travel, providing minimal disruption to the giving a vertical step height range up to 20 ILm.
measurement process.Anti-vibration mounts isolate Finally, measurement control and signal output are
the DC motor/gearbox that drives the micrometer provided by a suitably configured Pe.
from the main structural elements of the instru- The software offers various levels of surface
 Surface texture: two-dimensional

Transducer a sClllbly indicating material

Column uscd in measurCl11cnt loop.
Adaptor plate Ligamcnt pair X

",:."
'" ' ~~------~----~--r---~~
Thbe

" ',',
:.'. . ...... .
" ..

.. : "
..
'.'
,' , .. ' "',' , ,, "
',.

,'. " .: :;<:' .:. "

.;, ' . ..
. ~ :.: "
"
Coil holder

....
.':'. : "':"
.' '. . . . .. ' '

:<:,'-:(.' ' .'

,','
Dialllond stylus

\---/
Rollers I
" /
Material:
Transduccr stern G lass cerulllic

Leaf s pring
\ luminium alloy

X) Invar

~.~
Stylus holder
/ tylus as cmbly "howing kincmatic IOClltion.

Contact poinl~ 01'

tem with holder

Vicw in dircction X - X

Figure 60. Internal construction of the "Nanostep" instrurnent. ((ourtesy ofTaylor Hobson,)

texture analysis, assessed relative to the best-fit or least-squares are) and then removed, enabling
referenee line, eovering a diverse range of inter- the texture of the surfaee to be analysed;
national parameters. Additionally, the form software • angle - providing a faeility to remove any eom-
provides for pound surfaee tilt prior to parameter analysis;
• dimension - provides for the linear relationship
• form removal - best-fit form is ealculated (least- of surfaee features, whieh ean be assessed and
squares straight line, minimum-zone straight line eompared.
 Software has also been developed for this instru-
ment enabling it to assess an enhanced dual profile.
References
Ihis provision enables one to measure the profile of
a "master component", which can then be saved as Journal and conference papers
a template; any subsequent components can then
be both measured and simultaneously displayed Bell, T. Surface engineering: past, present and future. Surface
with this template for immediate comparison. A Engineering 6(1),1990,31-40.
statistical process control software package can be Bjuggren, M., Krummenacher, 1. and Mattsson, 1. Noncontact
surface roughness measurement of engineering surfaces by
employed, along with other specialised software total integrated infrared scattering. Precision Engineering 20(1),
packages. 1997,33-45.
A number of interchangeable styli (see Figure 59) Bousefield, B. and Bousefield, T. Progress towards a metallography
can be fitted to the instrument to suit a range of standard. Metals and Materials March 1990,146-148.
different measuring applications, ranging approxi- Brown A.J.C. and Gilbert, F. Industrial applications for optical pro-
filometry. Sensor Review January 1990,35-37.
mately from 1-10 /-Lm conispherical shape to 0.1 x Caber, P.J. An interferometric profiler for rough surfaces. Applied
0.5 /-Lm pyramidical geometry, the latter being used Optics 32,1993,3438-3441.
for high-resolution work. Ihe stylus force can be Downs M.J., Mason, N.M. and Nelson, J.C.c. Measurement of the
varied between 10 and 700 /-LN, allowing it to non- profiles of super-smooth surfaces using optical interferometry.
Proceedings 0fSPIE-1009, 1989, 14-17.
destructively measure delicate or easily scratched Drews, W.E. Surface measurement: an advanced technology.
components. Once the instrument has been set to the Quality Progress April 1987,43-46.
required depth of measurement, an automatic Dyson, J. Interferometry as a Measuring Tool. Machinery Pub. Co.,
facility provides for the raising and lowering of the 1970.
Garratt, J. and Bottomly, S.c. Technology transfer in the develop-
stylus when repositioning or changing sampies. For ment of a nanotopographic instrument. Nanotechnology 1,
the measurement of ultra-fine surfaces, the instru- 1990,38-43.
ment offers a nominal gauge resolution down to 31 Garratt J. and Mills, M. Measurement of the roughness of super-
pm (with the 2 /-Lm gauge range). Iypical perform- smooth surfaces using a stylus instrument. Nanotechnology 7,
1996,13-20.
ance of the 10 /-Lm conical stylus with an applied
Gee, M.G. and McCormick, N.J. The application of confocal scan-
force of 40 /-LN can produce a surface roughness Rq ning microscopy to the examination of ceramic wear surfaces.
value of 4.9 nm. Such nano-instrumentation has a Journal of Physics D: Applied Physics 25, 1990,230-235.
component capa city of diameter 100 mm, with a Harvie, A. and Beattie, J.S. Surface texture measurement.
depth of 22 mm, covering a large range of precision Production Engineer, 1978,25-29.
Hongai, Z., Zhixiang, C. and Riyao, C. The control of roughness
components. with expert system controller. Proceedings of the First Inter-
Instrument applications are both wide and varied national Mach. Mon. and Diagnostics, Las Vegas, 1989,813-817.
and it might typically be employed for Kuwamura, S. and Yamaguchi, I. Wavelength scanning profilome-
try for real-time surface shape measurement. Applied Optics 36,
1997,4473-4482.
• ultra-precision bearing and surface defect meas- Leach, R.K. Measurement of a correction for the phase change on
ure me nt - surface finish/integrity; reflection due to surface roughness. Proceedings of SPIE 3477,
• silicon wafer measurement; 1998,138-151.
• integrated circuit production - process monitor- Leach, R.K. Traceable measurement of surface texture at the
ing; national physicallaboratory using Nanosurf IV. Measurement
Science and Technology 11,2000,1162-1172.
• magnetic tape measurement; Leach, R.K. Traceable measurement of surface texture in the optics
• diamond-turned components - small diameter; industry. Large Lenses and Prisms Conference, University
• calibration standards; College London, 27 -30 March 2001.
• precision optics; Leach, R.K. and Hart, A.. Investigation into the shape of diamond
styli used for surface texture measurement. NPL Report
• mirrors for laser gyroscopes; CBTLM 10, April 2001, 1-12.
• laser and X-ray mirrors. Mansfield, D. Surface characterisation via optical diffraction. SPIE
(International Society for Optical Engineering) Vol. 1573:
In this chapter litde or no reference has been Commercial Applications of Precision Manufacture at the Sub-
Micron Level, 1991, 163-169.
made to the role of expert systems or neural
Prostrednik, D. and Osanna, P.H. The Abbott curve: weil known in
networks in surface texture analysis. Ihis omission metrology but not on technical drawings. International Journal
was intentional, as the chapter length would other- of Machine Tools Manufacture 38(5-6), 1998, 741-745.
wise have simply been too long and less specific in Schaffer, G.H. The many faces of surface texture. American
nature. In Chapter 2 a review of three-dimensional Machinist, June 1988,61- 68.
Schneider, u., Steckroth, A. and Hubner, G. An approach to the
surface texture measurements will be made, then evaluation of surface profiles by separating them into function-
later on contour/fractal effects will be discussed, ally different parts. Surface Topography 1, 1988,71-83.
which are now becoming an important technology Stout, K. How smooth is smooth? Surface measurements and their
in describing surfaces. relevance in manufacturing. Production Engineer May 1980,
17-22.
Thomas, T.R. Trends in surface roughness. International Journal of
Machine Tools Manufacture 38(5-6), 1998, 405-41l.
Books, booklets and guides
Trumpold, H. and Heldt, E. Why filtering surface profiles. Inter-
national Journal of Machine Tools Manufacture 38 (5-6),1998, Bennett, J.M. and Mattsson, L. Introduction to Surface Roughness
639-646. and Scattering. Washington, DC: Optical Society of America,
Ulbricht, R. Die Bestimmung der mittleren räumlichen Licht- 1989.
intensität durch nur eine. Messung Electrotechnische Zeitschrift Busch, T. and Wilkie Brothers Foundation. Fundamentals of
29, 1900, 595-60l. Dimensional Metrology. Delmar, 1989.
Vorburger, T.V. and Teague, E.C. Optical techniques for online Dagnall, M.A. Exploring Surface Texture. Taylor Hobson Precision,
measurement of surface topography. Precision Engineering 3, 1997.
1981,63-83. Dyson, J. Interferometry as a Measuring Tool. Machinery Pub. Co.,
Westburg, J. Opportunities and problems when standardising and 1970.
implementing surface structure parameters in industry. Inter- Gayler, J.F. W. and Shotbolt, C.R. Metrology for Engineers. Cassell,
national Journal of Machine Tools and Manufacture 38(5-6), 1990.
1998,413-416. Haycocks, J.A. Nove/ Probes for Surface Texture Metrology. NPL
Whitehouse, D.J. Some ultimate limits on the measurement of Report MOM105, July 1991.
surfaces using stylus techniques. Measurement and Control 8, Hume, K.J. Engineering Metrology. Macdonald, 1970.
1975, 147-15l. Hume, K.J. A History of Engineering Metrology. Mechanical
Whitehouse, D.J. Beta functions for surface topology? Annals of the Engineering Pub., 1980.
CIRP27,1978,491-497. Leach, R.K. NPL Good Practice Guide No. 37, Measurement of
Whitehouse, D.J. Surfaces: a link between manufacture and func- Surface Texture using Stylus Instruments. NPL, 2001
tion. Proceedings of IMechE 192, 1978, 179-187. Leach, R.K. and Hart, A. Investigation into the Shape of Diamond
Whitehouse, D.J. Conditioning of the manufacturing process using Styli used for Surface Texture Assessment. NPL Report CBTLM
surface finish. Proceedings of the Third Lamdamap Conference, 10, April 200 l.
Mainsah, E., Greenwood, J.A. and Chetwynd, D.G. Metrology and
Computational Mechanics July 1997, 3-20.
Properties ofEngineering Surfaces. Kluwer Academic Pub., 200 l.
Whitehouse, D.J. Some theoretical aspects of surface peak para-
Mummery, L. Surface Texture Analysis: The Handbook. Hommel-
meters. Precision Engineering 23, 1999,94-102.
werke GmbH, 1990.
Whitehouse, D.J. Surface measurement fidelity. Proceedings of the Nicolls, M.O. The Measurement of Surface Finish. DeBeers
Fourth Lamdamap Conference, University ofNorthumbria, WIT Technical Service Centre, DeBeers Industrial Diamond Divi-
Press, July 1999,267-276. sion, Charters, UK, 1980.
Whitehouse, D.J. Characterizing the machined surface condition Reason, R.E. The Measurement of Surface Texture. Cleaver-Hume
by appropriate parameters. Proceedings of the Third Industrial Press, 1960.
Tooling Conference, Southampton Institute, Molyneux Press, Sander, M. A Practical Guide to the Assessment of Surface Texture.
September 1999,8-31. Feinpruf GmbH, 1991.
Wyant, J.C. Computerized interferometric measurement of sur- Stover, J.c. Optical Scattering. McGraw-Hill, 1990.
face microstructure. Proceedings of SPIE-2576, 1995, 122-130. Thomas, G.G. Engineering Metrology. Butterworths, 1974.
Zahwi, S. and Mekawi,A.M. Some effects of stylus force on scratch- Thomas, T.R. Rough Surfaces. Imperial College Press, 1999.
ing surfaces. Eighth International Conference on Metrology and Whitehouse, D.J. and Reason, R.E. The Equation of the Mean Line
Properties of Engineering Surfaces, University of Huddersfield, ofSurface Texture Found by an Electric Wave Filter. Rank Taylor
26-28 April 2000. Hobson Pub., 1965.
Whitehouse, D.J. Handbook of Surface Metrology. Institute of
Physics, Bristol and Philadelphia, 1994.
Whitehouse, D.J. Surfaces and their Measurement. Hermes Penton
Science, 1991.
Williams, D.C. Optical Methods in Engineering Metrology.
Chapman & Hall, 1993.
 Surface texture:
three-dimensional

"Brevis esse labara,

Obscurus fia."
Translation
"It is when I struggle to be brief
that I become obscure"
(Ars Poetica, 25, Horace, 65-8 Be)

69

G. T. Smith, Industrial Metrology

© Springer-Verlag London 2002
Outline on three-dimensional
Surface texture: three-dimensional

factors relating to three-dimensional parameters

and their characterisation are discussed, together
•
surfaces with some of the more esoteric and contentious
areas of the subject, namely, the research based on
Invariably the ability of a two-dimensional trace to fractal and topological feature recognition. With
satisfactorily characterise a surface has been some- these latter techniques, one is looking into means of
what limited, because in reality surfaces are quantifying three-dimensional surfaces by their
three-dimensional in nature rather than two. Three- geometry and form, of similarities to those found in
dimensional techniques have been developed to the natural world. As a result, they can be considered
answer this criticism, but at present there are only as experimental approaches to the surface topolo-
proposed standards of assessment - by surface gies and still need to be refined if they are to be
research groups - that are now being discussed, with utilised in an industrial context. The same reasoning
many of these potential parameters mirroring their could be said to be true for the discussion on neural
two-dimensional counterparts. The areal nature of networks decision-making abilities of surface char-
three-dimensional surfaces relates to their texture, acterisations. These somewhat controversial issues
directionality and lay, and this can create problems have been included to show current thinking in
in surface interpretation. For example, if one were to surface developments.
inspect a highly polished copper component, then
both our visual and tactile sens es would establish
that it was shiny and smooth, respectively. Con-
versely, a shot-peened steel surface will appear to 2.1 Introduction
have a matt or dull optical appearance and feel
somewhat rough to the touch. These three-dimen-
sional visual and textural attributes create a problem The ability to undertake surface texture measure-
of surface interpretation, but an even more funda- ments has been refined and developed continuously
mentally challenging question is: "What parameters for many years; the basic principles were mentioned
does one select to determine a representative surface in Chapter 1. Simplistically, the object is to move the
in three dimensions?" stylus, or probe in a straight line and at a fixed speed,
The anisotropie and isotropie nature of surfaces with measurement of the "z altitude" in relation to
could, in the main, be disregarded for two-dimen- the "X axis", this being the basis of two-dimensional
sional surface characterisation, but they become of surface texture assessment (see Figure 18). In the
fundamental importance for surfaces that are three- past there was not a choice as to whether two- or
dimensional in nature. Not only is considerably three-dimensional measurements were to be taken,
greater computer power necessary to process the as only the former existed. A criticism levelled at
significantly greater number of data points, but this two-dimensional surface texture assessment was its
will tend to slow the whole data-processing time inability to represent a surface's functional charac-
down, although this latter point is becoming of less teristics, thereby causing a proliferation and uncon-
importance with the recent improvements in a PC's trolled expansion in two-dimensional parameters.
computing power. Many of these newly developed parameters proved
Despite the objections raised above to the adop- to be inadequate in providing essential information
tion of a three-dimensional surface texture inspec- relating to a surface's functional aspects. More will
tion strategy, the surface characterisation technique be said on this topic in Section 2.1.3.
has some real merit, in that: Previously, interest in three-dimensional surface
texture was always the objective when attempting a
• the practice enables the user to gain an improved full understanding of topographical surface details,
visualisation of manufactured surfaces; but instrumentation limitations combined with inad-
• it enables engineers to classify and subsequently equate data processing restricted its development.
control the properties of three-dimensional Today, three-dimensional measurements can be
surfaces, as these features affect its in-service carried out, consisting of measuring many parallel
performance; and regularly spaced profiles in order to reconstruct
• three-dimensional surfaces overcome the current the surface topography from a rectangular area of the
problem of inadequately describing a surface by surface. In many circumstances three-dimensional
two-dimensions. surface measurement has become an indispensable
tool to today's metrologist seeking a clearer represen-
This chapter is predominantly concerned with tation of the surface's topographical details.
some of the three-dimensional instrumentation and An anisotropie surface, typified by Figure 10,
software that is currently available. Moreover, the results in radically different properties dependent on
the direction from which the two-dimensional where A = amplitude of sinusoidal surface; w =
measurements are obtained. Clearly, if readings are angular frequency of oscillation of tip as it moves
taken on topographical surface features such as across surface.
machined cusps or ridges, then the roughness The frequency of tip oscillation is related to the
measurement obtained would be significantly spatial wavelength A of the sinusoidal surface and
different due to "lay effects" (as illustrated in Figure traverse speed v by
4), depending on whether the trace is parallel
or perpendicular to the ridges. It is only when three- (iii) w = 27Tvl A
dimensional measurements and analysis are under-
taken that a true conceptual understanding of the By substituting equations (ii) and (iii) into equation
properties of a surface are apparent. (i), it can be shown that the stylus will lift off the
surface when

2.1.1 Stylus speed and dynamics

Conventional stylus contact instruments would Namely, for a sinusoidal feature of given ampli-
typically have operating traverse speeds of 0.5 mml s. tude A and spatial period A, there will be a
maximum speed with which the stylus can be
If such speeds were used for just a small surface
traversed across any feature yet still remaining in
area, then it could take from 20 minutes to several
contact with the workpiece surface. At higher speeds,
hours to completely map the surface, depending on
the stylus will be prone to lift off, thereby no longer
the number of data points and track width distances
representing with any accuracy the surface profile.
required to obtain a representative 3-D surface.
So that the dynamic performance of the stylus
The problem cannot be simply remedied by increas-
can be optimised for high-speed operation, the
ing the stylus speed; in particular, at high traverse
right-hand side of equation (iv) should be maxim-
speeds a conventional stylus loses contact with the
ised. This maximisation can be accomplished by
workpiece surface as it moves over sharp surface applying a static torque to the stylus via an external
features. This aspect of stylus performance is often transducer (in this case by an electromagnetic actu-
termed "trackability" , being dependent on the ator) - see Figure 61(c) - where it is attached to part
following: A. Increases in traverse speed are limited by the
square root of the static pressure, requiring large
• inertia of the pivoting system; increases in actual static stylus pressure applied to
• static stylus force on the workpiece surface; the surface, if high er speeds are to be employed.
• stylus traverse speed; However, this increase in the stylus pressure is some-
• amplitude and wavelength of surface features. what undesirable and generally should be avoided,
if surface damage promoted by it is to be minimised
To portray this "trackability effect", Figure 61(a) as it traverses across the part.
illustrates a schematic representation of a simple In order to rectify the unwanted stylus's static
stylus system. The equations of motion are pressure mentioned above and simplify the model,
a small lightweight spring can be utilised to
(i) I(Z/l) = T s - Tr distribute the mass evenly between the pivot point
and the tip, which provides static force on the
where I = moment of inertia of the system; Z = surface. This factor can now be expressed by modi-
vertical position of tiPi 1= length of the stylus from fying equation (iv) as follows:
pivot to tiPi Ts = net sum of static torques acting
about stylus point (i.e., due to mass of stylus and any
external applied force); Tr = torque due to re action
of stylus on surface (i.e., stylus loses contact with where Tspll = static force of stylus on surface due to
surface when Tr = 0). the lightweight spring; m = mass of stylus; g = accel-
By way of example, consider a stylus motion eration due to gravity; w = frequency oscillation of
across a simple sinusoidally varying surface, such tip, in motion on surface.
that the stylus tip's position (when in contact with From equation (v) it can be gleaned that this
particular stylus can have its dynamic response opti-
the surface) can be expressed as
mised by reducing the stylus's mass, while main-
taining an identical force on the surface. An
(ii) Z(t) = A sin (wt)
advantage accrues from this modification, in that the
dynamic response of the system can be optimised.
 (a) Diagram of a simple pivoting stylus

Stylus
Stylus pivot tip
z

Surface

(b) High-speed stylus gauge

Bearings

Photodiode Hairspring
Optical shutter fin

Pivot support Diamond tip

stylus
PLANVIEW SIDE VIEW - SECTION X - X

(c) One of the compound flexure spring mechanisms, showing position of sensor and actuator
used in feedback control system
Flexu re spring c1amp
A

Electromagnetic
- ---, .-- J
actuator \ Flexure spring

-----, r==-=l ~f

Moving carria ge_

- ~
B

LED -
/--
-n tr]::~
I-- - Flag
Ph otodiode

Optical position
- I-- --r --
Figure 61. Conventional stylus, with a comparison to the high-speed variety general arrangement fitted to a "Talystep" instrument. (Courtesy
ofTaylor Hobson.)
As a consequence, this now allows a shorter, light- separation of the average peaks. Ihe formation of
weight stylus to be fitted, which ensures that any the artificial two-dimensional envelope curve is
mechanical resonances in the stylus arm are at a obtained by shifting down the loci of circle centres by
high frequency and as a result do not unduly a linear distance equal to its radius (illustrated
compromise its performance. in Figure 62a). With the latest development of this E-
Ihe arrangement of the scanner, which is based system technique, both the computational time and
on aprecision compound flexure spring (Figure 61c) ensuing processing speed have been considerably
allows the stylus gauge to be rapidly and precisely enhanced by previously identifying prominent
moved across a surface in two orthogonal directions. peaks.
Being of compact design, this has the benefit of Ihe three-dimensional envelope system is simply
ensuring that the measuring head is both robust an extension of that for two-dimensions (see Figure
and portable, being ideally suited for industrial 62b), with the envelope surface resulting from
applications. Furthermore, the instrument's higher shifting down the radial distance of the locus of the
sampling rates ensure that three-dimensional spatial sphere's centre as it rolls over the three-dimensional
resolution can be maintained. Hence the require- data points. A "mean surface" can be obtained, as in
ment was to ensure that the stylus gauge had suffi- the "M-system", by shifting the envelope surface
cient bandwidth to cope with the increased data down below the highest contact point (peak) in
rate, together with minimal mechanical resonance in order that an equal volume of material to air occurs.
the stylus system at the desired bandwidth, which Ihis new "M -system" envelope approximates the
might otherwise lead to distortion of the measured previous envelope surface of the original rolling
profile. Recently, such instruments (typified by the sphere radius. In this mann er a form of three-
"Ialystep") have allowed stylus traverse speeds to dimensional filtering can be attained, with the filter
be increased lO-fold by this careful redesign of the being controlled by the radius of the rolling sphere.
stylus gauge. Ihis enables stylus speeds of approxi- In practice, the envelope surface consists of just
mately 5 mm/sec to be used without losses in either portions of these rolling spheres, with each having a
tracking or signal fidelity as it measures areas up to radius equal to that of the rolling radius. Complex
500 x 500 /-Lm, with a sampling rate of 1 /-Lm, illus- mathematical definitions for the envelope system
trating that the operation can be completed in less would be extremely difficult to derive; therefore data
than aminute. is stored numerically and processed by suitable algo-
rithms to speed up processing time.

2.1.2 Envelope and mean systems

2.1.3 Three-dimensional
In the mid-1960s it was proposed that the "envelope characterisation
system" (E-system) might offer advantages in assess-
ment of both two- and three-dimensional surfaces. Ihe goal in any three-dimensional characterisation
Ihis E-system approach was at that time somewhat of surface topography is to integrate the surface
impracticable, as the hardware and software had not features in a representative manner as accurately as
been developed to exploit this technique. Although possible. Many methods have been utilised to obtain
the E-system procedure provides a rigorous and a degree of surface visual characterisation, with
sound approach to such data computation, it could notably the best technique at present being to
not be developed further at the time. With the intro- describe the surface condition by a predefined series
duction of fast and inexpensive PCs to complement of parameters which can be quantifiably measured
the hardware developments in surface texture in- then related to practical operational performance.
strumentation, this has virtually eliminated the In recent years, many practitioners have developed
requirement for hardware execution of surface filter- new parameters and even some that are inappro-
ing and characterisation. Ihis E-system is a specific priate - which has led to the much-quoted term
example of the application of a morphological filter: "parameter rash" - to describe the unchecked evolu-
this being a dilation of the surface/profile by the tion of particular topographical conditions, or
application of a spherical/circular element. In two- specific functional applications. Moreover, up to the
dimensional E-system filtering (Figure 62a) it can be present, nearly all measurements related to surface
assumed that instead of a stylus being used a hypo- topography are based on two-dimensional tech-
thetical circle is rolled over the profile under test. As niques, although a certain number of three-dimen-
the circle rolls along the profile the locus of its circle's sional parameters have been developed and at this
centre approximates to that of the profile, with its time have not been formally adopted into specifica-
path dependent upon the circle radius relative to the tion standards. As a result of this uncertainty in
specifying three-dimensional parameters, they offer Chapter 1, whereas their three-dimensional coun-
only marginal correlation, with the important in- terparts are designated by the logical choice of the
service requirements of properties typified by: letter S. This S prefix was assigned to indicate the
"mean surface" of the profile, which has been
• tribological factors - frictional and wear charac- derived from a previously EU-funded project co-
teristics; ordinated by the University of Birmingham. This
• fluid retention - lubrication performance; work developed a range of three-dimensional
• resistance to galling - minimising surface delam- surface descriptors and has since been known as the
ination (i.e., tearing) through wear. Birmingham 14 - Primary Set. This "set" has since
been expanded and is shown in Table 5. Other para-
Two-dimensional roughness parameters are in meters have also been developed for specific produc-
general denoted by the lead letter R, as discussed in tion requirements, most notably in the steel-

(a) 1Wo-dimensional envelope

Loci centre

Profile

(b) Three-dimensional envelope

Geometrically ideal surface

Envelope profile

A - Roughness
B - Waviness Surface profile

Figure 62. The two- and three-dimensional evaluation of a surface, according to the "E-system". (After Haesing, 1964, Tholath and
Radhakrishnan, 1999.)
 InClustrial metrology , .

Table 5. Three-dimensional parameter characterisations

Amplitude parameters Spatial parameters Functional parameters Hybrid parameters

'The Original Binningham 14 Oata Ser (After Stout et al., 1993)

RMS deviation Density of summit Surface bearing index RMSslope

Sq Sds Sbi Stlq
Ten -point height Texture aspect ratio (ore fluid retention index Mean summit curvature
Sz Str Sei SSc
Skewness Fastest decay autocorrelation Valley fluid retention index Developed area ratio
Ssk length Sv; Sdr
Sal
Kurtosis Texture direction
Sku Std

Taylor Hobson "Talymap'

Additional parameters include:
SaStSpSv Spc STPSHTp
Smvr Smmr > Volume parameters

Derived fram ISO 13565-2:

SkSpkSvk
SR7 SR2

rolling/texturing industries, particularly for fluid characterisation will also be developed in due
retention applications, but more will be said on this course. A range of amplitude or height distribution
topic later in the chapter. However, some confusion parameters can be selected for the examination of a
exists in either parameter classification into appro- three-dimensional surface and include the
priate subgroupings as well as their associated desig- following:
natory letters and prefixes, because the continuity of
groupings has as yet not been fully established. • Dispersion (5q) - this is the root mean square
Generally, three-dimensional parameters can be deviation of the surface which relates to surface
classified by the following groups, which are departures within the sampling area, being statis-
assigned by their mathematical derivation: tically termed the standard deviation of the
height distribution:

f-± t
• amplitude characterisation;
• spatial characterisation;
• functional characterisation; 5q = Z2 (x, y) dx dy
• hybrid characterisation;
• fractal characterisation. where domain of integration is the measurement
area A = dx dy.
• Extremes (5z) - this is the ten point height of the
Amplitude characterisation surface, being an extreme parameter which can
be defined as the average value of the absolute
Ihe characterisation of surfaces by statistical heights of the five highest peaks and deepest
methods has been widely exploited by industry and valleys - by the "eight neighbours method" -
academia. Because surfaces tend to be represented within the sampling area.
by random data, it seems natural that statistical tech- • Asymmetry (5sk) - this is the skewness of topog-
niques can be applied to them for appropriate raphy - height distribution, which measures the
analysis. In two-dimensional parameter characteri- asymmetry of surface deviations above the mean
sation, ISO 4287: 1996 describes surface texture and plane. Ihis parameter describes the shape of the
the techniques for defining terms and parameters, topography height distribution. For example, for
while ISO 4288: 1996 deals with the rules and proce- a "Gaussian surface" - having symmetrical shape
dures for surface texture assessment. It is anticipated to its surface height distribution - this would
that the equivalent three-dimensional parameter equate to zero skewness, whereas an asymmetric
 distribution of surface heights may produce a length and is used to determine aspects relating
negative skewness, when it has a longer tail on the to a surface texture's uniformity via the auto-
lower portion of its mean plane. This skewness correlation function. Principally, the texture
parameter provides an estimation of the exist- aspect ratios vary between 0 and I, with values
ence of "spiky topographical features": <0.5 indicating that no defined lay occurs and
ones of >0.5 exhibiting a pronounced lay. In some
1 m n
Ssk = -
mn
Sq L L Z~,j
i=! j=!
instances with a finite size of the sampling area,
the possibility exists that for certain anisotropic
surfaces the AACF's slowest decay will not reach
• Sharpness (Sku) - this is the kurtosis of topog- 0.2. Under such conditions, the surface's slowest
raphy - height distribution, which measures the decay direction should be employed;
sharpness of the distribution, characterising its • Sal - this is the fastest decay auto-correlation
spread. For example, a "Gaussian surface" equates length, defined as the horizontal distance of the
to a kurtosis value of 3; conversely, a surface AACF having the fastest decay to 0.2. This value
topography that appears centrally distributed of 0.2 is the shortest auto-correlation length that
would produce a kurtosis value in excess of 3, the AACF decays to in any direction. For example,
while a more evenly spread topography would with an anisotropie surface, Sal occurs perpen-
have a value less than 3. By combining both the dicular to the lay surface and a large value of Sal
Ssk and Sku parameters, this may allow identifi- denotes that the surface is dominated by long-
cation of surface topographies having reasonably wavelength components. Conversely, a small
flat tops with deep valleys. value indicates short wavelengths;
• Std - this is the texture direction of the surface,
whieh determines the most conspicuous direc-
Spatial characterisation tion of surface texture (lay) with respect to the
Y-axis within the frequency domain (Fourier).
Parameters relating to the a surface's spatial proper-
ties offer some difficulty in their characterisation,
due to their general wavelength randomness Functional characterisation
combined with multi-wavelength variations, which
in turn are coupled to their high sensitivity to the The classification of three-dimensional surfaces by
sampling interval. Techniques utilised for spatial its "bearing area" implies that a set of techniques are
characterisation include the following: engaged relating to specific functional properties of
a surface. Unfortunately, due to the diverse range
• Auto-correlation - aversion of this technique in of functional aspects demanded by industrial appli-
surface texture characterisation is that of areal cations, no single functional characteristic can
auto-correlation function (AACF), where the adequately cover all of these techniques. However, it
dependence of one position of data values is is possible to utilise the following parameters to
dependent upon another. Typieal parameters that establish the potential in-service performance of a
originate from the AACF, might be: surface:
• Sds - this is the density of peaks, being the
number of peaks of a unit sampling area, by the • Sbi - this is the surface bearing index, which
"eight neighbours method": relates to the Sq parameter over the surface's
height at the value of 5% bearing area, with a
Sds = (~) (::)
greater surface bearing index indicating superior
bearing properties;
• The RMS deviation Sq is defined by analogy with
where its two-dimensional form;
• Sei - this is the core fluid retention index, being
2[R(O) - R(Llx)] the ratio of void volume of the unit sampling area
m2 = (LlX)4 at the co re zone (i.e., 5-80% bearing area) over
parameter Sq. Good fluid retention occurs with
2[3R(O) - 4R(Llx) + R(2Llx)] greater values of Sei and for a "Gaussian surface"
m4 = (LlX)4 the index is approximately 1.56:

• Str - this is the isotropy index or texture aspect of

the surface, defined as its ratio of fastest-to- Sei
1
= - Sq
[fz*o.oS
1- p(z*) dz*
j
slowest decay (0.2) with respect to correlation
A z*0.8
 where z* = zlSq and any suffices refer to bearing geometry; see Figure 1) that are not particularly
area fractions. influenced by the laws of Euclidean geometry,
• Svi - this is the valley fluid retention index, which namely defined planes and analytical functions, is of
is the ratio of the void volume of the sampling relatively recent origin. The reasons for this situa-
area within the valley zone (i.e., 80-100%) over tion are not at present particularly clear, but
the Sq parameter. For example, if a larger Svi presumably occur from the operation of process
occurs, this indicates excellent fluid retention in hierarchies that chaotically dissipate energy across
the zone of the valleys. many dimensional scales. Initial work in this field

Svi = -1
A
l
Sq 1 - JZ*0008 p(z*) dz*
z*1
l was prineipally concerned with discovering the
existence of fractal behaviour over a large variety of
distinct eircumstances. Later, fractal research was
influenced by attempting to understand and uncover
where z* = zlSq and any suffices refer to bearing the causes and relevance of this behaviour and,
area fractions. in particular, trying to find functional dependen-
eies and relationships that correlate fractal dimen-
sions of engineered surfaces to speeific in-service
Hybrid characterisation performance.
A considerable body of research-based publica-
tions has been written on the measurement of fractal
The result of combining two or more of the prev-
dimensions, with some techniques being inappro-
iously discussed three-dimensional classification
priate for surface assessment, namely of a general
techniques is termed a "hybrid". By combining such
self-affine nature, as opposed to those being self-
parameters, it is possible to acquire many hybrid
similar, while other work has proved unsuitable for
parameters, with the most significant being:
anisotropic surface characterisation. For most of this
fractal research work the techniques employed have
• SLlq - this is the root mean square slope of the been used to estimate the bounds of the fractal
three-dimensional surface texture, within the dimension with varying degrees of numerical preei-
sampling area: sion. Generally, the lack of speeific physical mean-

SLlq = !1 Jt z'(x,y) dx dy
ings in fractal characterisation has meant that work
in this field has been restricted in the main to acad-
emia.
where

z'(x,y) = (::r (:;r

+
ln-service performance
Depending upon its potential end-use, a surface can
be "engineered" to meet the anticipated in-service
• Ssc - this is the (arithmetic) mean summit curva- demands and now, with three-dimensional charac-
ture of the surface and is defined as the average terisation, quantitative measurements can be made
of the principal curvatures of the summits within to ensure that the desired conditions have been met.
the sampling area. Namely, the sum of curvatures For example, in many sheet metal applications the
of a surface at a point along any two orthogonal retention of a lubricant at speeifically defined sites
directions equals the sum of predominate curva- is considered an important factor and parameters
tures; have been suggested that define lubricant trapping;
• Sdr - this is the developed (interfacial) area Figure 63 indicates just such a site. This figure can
ratio, which is the ratio of interfaeial area of a be defined by intersecting lines at different positions
surface over its sampling area. The parameter along the profile heights to produce the percentage
describes the hybrid nature of surfaces, with a length. The three-dimensional version is an exten-
large numerical value equating to either signifi- sion of its two-dimensional counterpart, although
cant amplitude, spaeing or both surface condi- currently there is not a specification for such
tions. filtering. The technique is generally more suitable
for characterising surface conditions where the
profile has either a flat top or when porosityl
Fractal characterisation grooving is present. In fact, the ability to esti-
mate and determine both isolated valleys and the
The assumption that industrially-developed sur- possibility of valley connectivity have been investi-
faces might "mirror" natural phenomena (fractal gated through a "Maxwellian/Motif landscaping
 Surface texture: three-dimensional

Opcn
an:3
J
10 cd
aren
7
Equi alcllt
arcas

Figure 63. Abbott-Firestone curve with closed areas. (After Sacerdotti et al., 2000.)

technique" or "change tree approach" - but more will

be said on this topic later in the chapter under
2.2 Three-dimensional
Section 2.4 (Fractal techniques). analysis software
In Table 6 are indicated the functional character-
istics for a range of conditions, showing the
three-dimensional surface texture parameters In the last few years three-dimensional measure-
ment technology has been introduced by many of
and their expected correlation on a matrix for a
the larger surface instrument manufacturers, with
series of characterisations. The table cannot be
the specific objective of being able to overcome defi-
an exhaustive list of functional performances but ciencies resulting from two-dimensional assess-
indicates just some of the current applications ment. In general, industry has been somewhat
where they may, or indeed may not, be utilised. Such reluctant to invest in three-dimensional equipment,
tables indicating functional performance are apart from academia (see Table 7). The success of
important, but exact classification cannot be quite any three-dimensional surface texture application is
as clear-cut as indicated in this chart. It has derived through the abilities of its software surface
been argued that three-dimensional measurements performance and characterisation. This section will
are significant because, in practical terms, surfaces briefty review just some of the visual images gen er-
are three-dimensional and not two-dimensional ated from the three-dimensional characterisation of
in nature. In Section 2.2.1 is a discussion on aspects surfaces; it is not meant to be an exhaustive account
of three-dimensional functional performance of of the subject but simply a "visual tour" of 3-D
surfaces. surface potentialities.
 InClustrial metrology , .

Table 6. Functional performance correlations

Function Characterisation

Amplitude Spacing Functional Hybrid

Forming and drawing *** **-

....
••** ..
Painting and plating
Friction
Galling
Wear
-*-
*** "***
****
..**.
..**
*...
.***
..".
***

****
****
Joint stiffness
Slideways
Electro-contracts
Bonding and adhesion
****
****
**it"
**
****
-**
****
****
****
**-
-
**

****
**
Fatigue *** ****
...
Stress and fracture
Reflectivity
Hygiene
Bearings
**
**
*-
****
**

**
.***
**-
**-
****
--
**
*.*-
Seals n_
***" ** ****
Related 3-DParameters Sa, Sq, Sz, Ssk, Sku Sds, Str, Std, Sal Spk, Svk, Sk, Srl, SD.a, 55(, Sdr
Sr2, Svi

Key: **** high correlation; * very little correlation.

After Griffiths (1988).

Table 7. Two- and three-dimensional stylus profilometer usage • colour representation - for example, an axono-
metric surface characterisation - either of the
Industry applications: Surface texture usa ge wired or meshed type (Figure 64);
• colour altitude co ding - indicating relative heights
Two- Three- of peaks and valleys (Figure 64 again);
dimensional dimensional • autocorrelation imaging - analyses directions
(isotrophy) and periodicity of the 3-D relief (not
Aerospace 10% <2% shown);
Automotive 75% 2%
Electrical 10% <1 %
• photo-simulation with lighting effects - which
Paper 10% 0% enables the captured image to be rotated or
Cosmetic 10% <1 % inverted to gain a complete visual impression of
Academic 5% 94% the measured surface's terrain (Figure 65);
• volume studies - defining, say, a hole's volume,
After Griffiths (1999).
perimeter, area, mean depth and maximum depth
Many of the three-dimensional software systems of an irregular cavity in the surface (Figure 66);
employ a Windows( -based software interface, giving • surface filtering - employed to separate out any
the user a highly intuitive visual capability with waviness, roughness and form data into discrete
logical, menu-driven functions having comprehen- entities; this occurs by FFT filtering to separate
sive help facilities for on-line user support. Typical out components of the 3-D relief, frequency by
of such fast operational industrial-based software is frequency and direction by direction (Figures 67
that illustrated in the following self-explanatory and 68);
examples, giving not only clear visual and easily • contour diagram studies - enabling points having
interpretative imagery but also considerable direct identical heights to be joined (Figure 69);
surface-related 3-D information. • pattern recognition - highlighting the points
A representative example of such comprehensive with specific topological properties (Figure 70);
3-D surface software is described below. This soft- • threshold operator utilisation - enables excessive
ware enables the user, once the 3-D profilometry raw peaks to be removed or extreme holes to be filled
data has been captured, to manipulate a measured in (Figure 71);
surface in numerous ways by selecting appropriate • a re-sampling operator - this technique clarifies
colour palettes or individual colours - for clarity - the image after successive zooming operations
together with zoom function ability, giving: have occurred and insufficient data points exist
 Surface texture: three-dimensional

I ...
In

Figure 64. 3-D representation of an axonometrie surfaee produeed from a stylus-based instrument. ((ourtesy ofTaylor Hobson.)

Figure 65. A continuous 3-D axonometrie representation of a surfaee repradueed fram a stylus-based instrument. ((ourtesy ofTaylor Hobson.)
 InClustrial metrology . .

(al A typical distance measurement study of specific

features on the topography

y 0.741 rm1 y 1.08 lTY11

Z 217 IJm z 221 IJm

Hor,zontol dlstance : 0 . 532 mm

Vana tion ,n l eve l : 3.14 \.Im 25. 4 •
Angle
Obi ique d,s ance : 0.532 mm 0.444 rod
(a) ' - - - - -- - - - - - - - - - '(b)

(cl Hole volume study; perimeter, volume, area, mean

depth and maximum depth of an irregular cavity in the
surface

mOx\",un'I dep h : 116 um

Votum : 11 . 3 mm 2 )( vm
(c)
(dl Ahorizontal slices study illustrates which part of the surface is below, or above a given height. NB the area of the "highest slice" is known
as the bearing ratio

143 IJm

88.6 IJm

~ 46.71% l1li 53.29% 27.72% ~ 62.18%

(d)
Figure 66. 3-D software manipulation used to indicate important features on the surface topography. ((ourtesy of Taylor Hobson.l
 (3) (b) (c)

Figure 67. The filter operator can be used to separate out waviness and roughness. (al The unfiltered surface. (bl The waviness ofthe surface. (cl The roughness ofthe
surface. ((ourtesy ofTaylor Hobson.l

I
The original surface .. . can be separated into the form . .. and the surface - allowing form to be removed I
~.
""'
6

+
o

Once the form has been removed, the waviness . . . and the roughness . . .

~. ..... V·
6 2.5
0 .5

9.4
1. 5
0 .3

1+ 9 .2
0 .5 0.1

I) 0 o
Wa v in es~ Parameiers (Cut Off : 0.25mm) Ro u ghll cs~ Parnmetcrs (
~ Wa = 0.332 IJ Ill ~Ra = 0.0536 IJI1l
sW v = 1. 24 IJ m ,R v =0.31 1 IJm
~ Wp = 1.32 IJI11 .. Rp = 0.272 IlI1l
~ Wt = 2.57 IJ Ol ~ Rt = 0.584 IJI1l
sWq = 0.41 3 IJI11 ~ Rq = 0.0698 IlI11
sWsl.. = 0.0173 , R ~ 1.. = -0.0376
sWl..u = 2.9 sR ku = 3. 16

... can be separated (i.e., using the filter operator) and parameters computed

Figure 68. It is possible to remove form from the surface utilising the filter operator. ((ourtesy of Taylor Hobson.)
 Surface texture: three-dimensional

IJm IJm

198
178 173
149
133
124
99.1
88.9
74.3
44 . 4 49.5
24.8

Figure 69. A 3-D contour diagram study joins points which have the same height. (Gourtesy of Taylor Hobson.)

2.2.1 Functional 3-D performance

The surface topography characteristics of an engi-
neered component may have important implications
in the fields of:

• tribology - wear studies;

• lubrication - oil retention in surface valleys;
• adhesion - for paint and plating applications;
• absorbency - measuring paper surface volume to
ascertain the amount of ink needed in the print-
ing process;
• plateau honing - determining honed cylinder
liner characteristics;
• reflectivity - analysis of pattern and direction of
car headlamp glass covers, ensuring optimum
lighting in foggy conditions;
• ageing - comparative analysis of components
before and after accelerated weathering/ageing
processes (typically to predict material life for
The plateaus (Le., in red) and the valleys (Le., in blue) of a human skin painted and canvas surfaces).
surface

Figure 70. Pattern recognition study highlights the points with In many industrial applications speed of data manip-
specific topological properties. ((ourtesy of Taylor Hobson.) ulation is paramount and if identical analytical pro-
cessing is necessary then "macros" allow operations,
calculations, studies and comments added to the
to enhance the image; this can be achieved by image produced - to be efficiently executed in a few
increasing the number of pixels, which includes seconds. In many industries there is a fundamental
optimising the "z resolution" (Figure 72); requirement for the measurement of surface holes
• profile extraction - this enables the 3-D topog- and depressions, this being necessary, when:
raphy to be examined and at particular positions
of interest on the surface the user can pre-select • "surface engineering" is the requirement where
a line - straight, or manipulated around and cavities or holes in the surface topography are
across specific surface features - to be drawn and designed for fluid retention;
a two-dimensional trace at this "cut surface" can • texturised surface features - depressions - need
be illustrated (Figure 73). to be confirmed by 3-D measurement;
• cavities present in the surface need to be
measured that might affect in-service perform-
ance.
 InClustrial metrology . .

,..
",
~m

150

140

130

120

110

100

90

80

70

60

50

~0

30

20

10

Figure 71. By utilising the threshold operator excessive peaks can be removed. This also applies to the filling in of excessive holes. [(ourtesy
ofTaylor Hobson.

Re-sampling enhances the representations when there are no longer enough data points (e.g., after successive zoom operations).

Figure 72. The re-sampling operator allows one to change the data sampling space. ((ourtesy of Taylor Hobson.)
 Surface texture: three-dimensional

100 +----r""""'cc------l
so
0 +----+----.----+----~--~----~--~--~----~~~
o 0.5 1. 5 z
Length
2.53 "'"
rs.t z
x :
155 "'"
0.93 "'"
01Aft
2.42 rm'I

~ .
4 0+---~

20

0 +----4-------lr--~~--~----~--~----+_~~----+__+
o 0.7
length : l'C~ z : 57.9 L.Jf"I 0..-
0 . 97 """ ~ x. : 0 .54 rrm 0 . 76 lmI

"m
40

Z0

o 0.5
L "9l1\ : z : 62.3 ""' r7TZ 'O~
2.58 11m ~ : 1.67 ftWI'I V':* )t : 0 . 04 nnI

NB: the path followed along the surface can be a straight line, the bottom of a valley, a circular line, or any other user-defined contour.

Figure 73. The profile extraction operator allows one to extract one trace out of the surface. ((ourtesy of Taylor Hobson.)

Hole calculation Importance must be given to differentiating

methods for calculating the volume of a single "blind
In Figure 66( c) a "hole volume study" of a cavity was hole", whose periphery can be delineated by either
shown being examined. It might be prudent to ask tracing around it with a mouse or moving the
the question: "Why calculate the hole's volume?" cursors. Such an outlining technique allows for
As has been described above, many industries calculations of either the average volume of a cavity
have specific requirements to measure the volumes or, conversely, for material that might be situated
of holes in high-grade fine surfaces, in order to between two altitudes on a 3-D trace (see Figure 63).
quantify: Some practical examples and solutions to this irreg-
ular hole measurement dilemma are given below:
• abrasive effectiveness - grinding, honing, lapping
and super-finishing cutting efficiencies; • Measurement of a hole on an uneven trace - this
• textural condition of laminating rollers; represents some degree of difficulty if suitable
• wear and corrosion efficiency; software is not available. By way of illustration: if
• predictions of oil reserves on mechanical compo- there is a requirement to measure the healing
nents; capacity on a leg wound, the wound's surface
• porosity of an interior (open aspect) surface, volume is uneven - due to the anatomy of muscle
such as those found on powder metallurgy and bone - and tends to be represented by a
components; cylindrical rather than a plan ar profile. If a silicon
• impact of graining on a surface; replica of the wound is taken this will be three-
• efficiency of a punch; dimensional and follow the wound's contours, but
• effectiveness of cosmetics; when laid Bat much of the cylindrical contouring
• volume of an embossed bump (after height inver- effect will have been lost. However, the con-
sion, for example, solder on a printed circuit straints imposed when taking an imprint -
board). replica - deform a zone around the inspected
hole. Such deformations cannot be ignored, since
they have an amplitude akin to that of the hollow
 to be measured. Suppression of the deformation Form removal
of the replica can be achieved by applying a poly-
nomial form removal to the contour of the hole The topography of a complex relief surface can be
by separating out both the local effects at the measured by profilometry, by separating out its
inside and outside of the hole, this being delin- component parts depending on their respective
eated by using the mouse; wavelengths and amplitudes. The smaller the wave-
• Measuring an irregular hole volume, when closely length, the greater will be the variations in the alti-
surrounded by other holes - in order to calculate tude for identical horizontal lengths. By moving
the main hole's volume, it is necessary to care- from a larger to a smaller wavelength, the four fol-
fully define the reference plane. For example, on lowing components can be distinguished:
a hemispherical (dished) surface its volume
depends upon its relative depth from the top Form - this is a component of topography having
surface. By using one of the software functions, a wavelength almost equating to the object's
typified by a zoom operator, it is an extremely length. Its measured form is an amalgamation
difficult task to isolate the hole, since this cre- of the theoretical form defined at conception
ates a "squared portion" of a surface - thereby (sphere, cylinder, etc) and the form variation,
containing the adjacent dosely packed holes. which is the difference between the manufactured
However, the erase defect operator enables other and theoretical forms. It should be noted that
holes present in this zoomed area to be erased, form cannot be measured using a skid in situ
giving an acceptable hole contour which acts as a with its stylus, as the skid attempts to follow the
re fe ren ce plane for subsequent calculation of the intrinsic shape of the object; therefore the alti-
hole's volume; tude signal will only contain variations in the
• Zone surrounding a hole has a complicated relief relief (see" Figure 38);
- rather than attempting to exacdy define the 2 Waviness - this is a component of the surface that
reference plane and then calculate the hole's gradually varies with respect to its horizontal
volume between the bottom of the hole and this position. For example, waviness might be defined
plane (as in the previous case), one can vertically by a wavelength of between 0.5 and 2.5 mm;
cut the image into elementary profiles. For each of moreover, on a production part such waviness
these separate profiles a calculation can be made could be the result of low-frequency vibrations
for the surface between the hole's bottom and of the component in relation to the machine
a line defined by the edge of the hole. For each tool;
hollow its "top" will differ with every vertical 3 Roughness - this is a component of the surface
cut of the image, since the software recalculates that rapidly varies in relation to the horizontal
every respective hollow. The advantage is that a position. Roughness could be defined by wave-
complex shape to the hollow's contour can be lengths between 20 and 500 fLm; with a manu-
assessed, but the disadvantage is that smalliocal factured part this roughness might be due to a
variations in the contour's relief can introduce number of different but discrete production
significant variations in the measured volume. processes, such as shot-/sand-blasting, grinding
Furthermore, this technique has a tendency to particles (i.e., abrasives) and tool deflectionl
consider possible lips adjacent to the hole, typi- vibration;
cally crater edges and material displaced from the 4 Micro-roughness - this, as its name suggests, is the
hole, whereas the reference plane method ignores finest component on the relief, with wavelengths
the presence of disturbances promoted by lips of, say, less than 20 fLm. Micro-roughness is
when calculating the height of the hole; generally associated with the structure of the
• Average volume, or volume of a single hole - measured material, rather than being the result of
certain studies require the calculation of the a particular production process. Prior to calcu-
average hole volume, while others just need the lating the roughness parameters, micro-rough-
volume of a single hole. An example might be ness can be eliminated by suitable filtering.
when it is necessary to calculate the quantity of
adhesive needed to bond two surfaces together, The analytical study of arelief requires that the
requiring a calculation of the volume of air components described above, which relate to
between these two surfaces. The software best different properties in the overall relief, are normally
suited to this problem is one that can produce a separated out before dose inspection occurs. As
volume study, giving the total cavity volume previously mentioned, the separating out of wavi-
between the highest and lowest points of the ness, roughness and micro-roughness can be
surface, or between altitudes that can be readily achieved by filtering. Generally, two main filters tend
calculated. to be used:
 Surface texture: three-dimensional

• Gaussian filter (convolution of the profile to a

"Gaussian function");
• RC2 filter (mathematical simulation of a "second-
order electric capacitor-resistor filter").

The delineation between waviness and roughness

is termed "cut-off" (see Section 1.6) and can take a
certain number of normalised values. This technique
is utilised for both two-dimensional (profile) and
three-dimensional (surface) measurements.
Form separation is not normally carried out by
filtering and probably the main reason for this is that
filtering, as defined by standards, may relinquish an
important section of the surface when employed.
When utilising surface texture instruments, the aim Figure 74. Portable 3-D surface profiling instrument allowing
is to suppress the form component so that an surface topography measurement and Windows scanning software to
isolated study of waviness and roughness can be create visualisation of the 3-D surface. ((ourtesy ofTaylor Hobson.)
attained. The form can be suppressed by using a
polynomial approximation, while imposing the depth analysis of the topographical features at high
minimum of constraints. The method consists of resolution and accuracy.
seeking the polynomial which best adjusts to the The fast operational stylus pick-up is scanned
surface or profile being measured. The coefficients across the surface under examination using two
of the polynomial are automatically calculated using traverse mechanisms. The mechanism controlling
the least-squares technique. The only constraint the Y-position of the pick-up simply steps the stylus
imposed is the degree "n" of the poiynomial, which sideways in synchronisation with the X-position
defines the complexity of the form to be subtracted. controller. These two systems result in aseries of
Therefore, the high er n becomes, the more the poly- profiles being obtained from the measurement head,
nomial "mimics" the surface relief, hence: which are built-up into a 3-D presentation of the
surface under examination. The measurement
• subtracting the first-order polynomial is the process can be enhanced for small components by a
equivalent to subtracting the least-squares plane; positioning table having: X-, Y- and Z-positional
• subtracting the second-order polynomial allows control over a 10 mm volumetric envelope.
uniform curvature to be eliminated from the The main features of the measurement head's
surface; specification, include traverse area of 1 mm 2,
• the higher the degree of polynomial, the more the measurement time <60 s for a 0.5 mm 2 area, gauge
form described by the polynomial becomes range of 200 f.Lm, gauge resolution 3 nm, gauge
complex and the less relief remains inside the range/resolution 4096:1; traverse speed ranges from
waviness. The choice of degree will therefore 10 mm/s (fast) to 2.5 mm/s (slow), with a head
depend on the limit that one defines between weight of 0.85 kg.
waviness and form. The software mapping capabilities are compre-
hensive for such a relatively inexpensive instrument,
and include the following:

2.3 Portable three- • Visual functions

dimensional measuring • 3-D plots with heights coded by colour (Figure
64);
instruments • 3-D continuous plots (Figure 65);
• surface smoothing abilities;
• a contour diagrammatic representation of the
This instrument for the assessment of three-dimen- surface (Figure 67);
sional surfaces (Figure 74) is used in conjunction • height viewing in pseudo colour for an alter-
with 3-D analysis software. It can be readily cali- native visual interpretation;
brated automatically and features a relatively low • photo simulation.
cost but high-speed measurement facility in
conjunction with a portable scanning head. The • Analysis functions
design enables the measuring instrument to be • 3-D surface parameters - previously men-
placed directly onto the surface and provides an in- tioned in Section 2.1.3;
 • limited 2-Dsurface texture parameters - for Numerous procedures are now available for the
comparison; determination of fractal dimensions. Some of these
• zooming abilities, for specific topographieal procedures are unsuitable for textural surfaces,
feature investigation/analysis; which can be considered as gene rally self-affine, as
• both Gaussian and 2CR filtering capabilities; opposed to self-similar, while other procedures are
• form removal - if this either influences the inappropriate for anisotropie surfaces. Many of the
surface topography or is visually distorting the fractal techniques estimate bounds of the dimension
parameters, such as a scratch (Figure 68); and these methods have varying levels of numerical
• hole volume computation, for regular/irreg- precision. However, a fractal model - via its surface
ular peripheries (Figure 66c); height distribution - can prediet or give a reasonable
• distance measurement between two topo- estimate of the fractal dimension for a rough an-
graphieal features (Figure 66b); isotropie surface and can even simulate such a sur-
• rotation of any angle/mirror, enabling a clearer face. The accuracy and robustness of these current
visual interpretation of the captured image to "direct" and "Monte Carlo" simulation techniques for
be assessed; fractal models still require further research to give
• inversion of heights, employed to measure higher degrees of correlation to actual situations.
surface replicas; The following technique indicates a change of
• Abbott-Firestone curve and amplitude disti- emphasis in determining embossed (convex) or
butions to be generated. concave features on three-dimensional engineering
surfaces, by employing an "areal motif" technique,
• Desk-top publishing functions which has its roots in the discipline of cartographic
• macros for rapid analysis and report genera- surveying.
tion.
The complete system requires connection to a
suitable PC, which ideally requires a screen resolu- 2.4.1 Topological characterisation
tion of at least 1024 X 768 pixels having 65,000 colour
codes, if the system is to per form in an adequate In recent years a new development in attempting to
manner for visual interpretation and assessment. determine the "connectability" of three-dimensional
surface features via "areal motifs" - previously devel-
oped by the French automotive industry - using
cartographic characterisation has been proposed
2.4 Fractal techniques (see Standards in References). This areal motif
method is based on the cartography procedure,
Introduction whose discipline was established for measuring
surfaces over thousands of years, with an extensive
It has been recently shown in published work that array of techniques to characterise features on a
many types of engineered surfaces exhibit a fractal surface, albeit in the macro-environment.
geometry. Under certain conditions modelling the A singular motif consists of that portion of a
surface generation predicts this effect and provides profile between two peaks having two adjacent peaks
correlation between dimensions, historieal develop- together with a dominant valley between them being
ment of the surface and its properties. Observations used to characterise individual motifs; namely, these
have indicated that many surfaces are non -Euclidean stationary points (critical points) are employed to
in nature (are not defined by planes and analytieal characterise the individual motifs. The initial motifs
functions), but apparendy have a fractal geometry- can be defined between neighbouring peaks on the
mirroring the natural world (see Figure 1 back- profile and rules are then applied to combine adja-
ground). The reasons why fractals occur cannot as cent motifs, in pairs, to form even larger new motifs.
yet be readily determined; presumably they arise The concept he re is to eliminate any "insignificant
from the operation of hierarchieal processes that peaks" while retaining "significant peaks". By this
dissipate energy in a chaotic fashion across many method the combined motifs enable aseries of para-
dimensional scales. Some of the initial papers were meters to be calculated which characterise the
primarily concerned with identifying the existence surface texture. In order to achieve satisfactory areal
of fractal behaviour in a wide variety of differing cir- motif analysis by this technique, two basic concepts
cumstances. Later work attempted to investigate and are needed:
understand the causes and importance of this fractal
behaviour and to detect functional dependencies • a definition of the areal motif itself - the areal
and relationships that correlate the surface's fractal equivalent of the profile motif;
dimension with specific his tory and performance. • defining areal combination rules.
 Surface texture: three-dimensional

(a) Contour map illustrating critical points and lines, with peaks "P", pits "V" and saddle points "S".

(b) "Change tree" chart derived from the surface contour lines.

Heighl P4

10

0+ __
va
·2

V3 V2

Figure 75. Topologieal surfaee texture eharaeterisation by fraetal analysis. ((ourtesy of Taylor Hobson.)

Cartographic characterisation between dales are ridge lines (watersheds). Maxwell

was able to demonstrate that ridge and course lines
Many years ago, Maxwell (1870) proposed dividing are respectively maximum uphill and downhill paths
alandscape to be assessed into those regions that emanating from saddle points, which either termi-
consisted of hills and those that consisted of dales. nate in peaks or pits (see Figure 75a).
In principle, a Maxwellian hill is an area by which Using these Maxwellian concepts and rules, Scott
maximum uphill paths lead to a particular peak; (1998) has proposed an extension of Maxwell's defi-
conversely, a Maxwellian dale is an area from which nitions to define a dale as consisting of a single
maximum downhill paths lead to a specific pit. By dominant pit surrounded by a ring of ridge lines
definition, the boundaries between hills are termed connecting both peaks and saddle points; a hill, he
course lines (watercourses) and those boundaries argued, would consist of a single dominant peak
 Industrial metrology , .

which is surrounded by a ring of course lines that • dale change tree - which represents the relation-
connected the pits and saddle points together ships between pits and saddle points;
(Figure 7Sa). Moreover, within the dale or peak in • hill change tree - representing relationships
question, there may be other pits/peaks, but they are between peaks and saddle points;
irrelevant compared to dominant pit/peak features. • Jull change tree - representing relationships
In this example a dale is considered to be the between critical points in hills and dales (Figure
"areal equivalent" of the profile motif, while the hill 7Sb), from which dale and hill change trees can
can be thought of as a useful complementary be calculated.
concept. Similady as for the profile motif method,
several types of surface-specific lines and points can In all of the previous discussion on these change
characterise dales and hills, respectively. These trees it was assumed that the lands cape did not have
features include the critical points (peaks, pits and edges, which unfortunately in practice is not the
saddle points) and critical lines (ridge and course case. Four types of edge critical points can occur:
lines; see Figure 7Sa). these are edge peaks, edge pits, saddle peaks and
saddle pits.
Change tree charts
The topographical "change tree" chart (Figure 7Sb) Areal combination
is ideal for organising the connectability relation-
ships between critical points in hills and dales, while Due to noise and other factors, in practice change
still retaining relevant information. The change trees tend to be dominated by quite short contour
tree represents the relationships between contour lines that impede interpretation. Therefore in order
lines from the surface, with the vertical direction to reduce factors such as noise, a mechanism is
representing height. At a predetermined set height needed to "prune" the change tree while still
all individual contour lines can be represented by retaining relevant information. By utilising the tech-
a point which is part of a line representing that nique of areal combination, this simplifies the
contour line continuously varying with height. change tree and any relevant information is still
Saddle points occur by the merging of two or more kept. For profile motifs, they can combine with adja-
of these lines into one. Peaks and pits are indicated cent features, while in the areal case the problem is
by the termination of a line. somewhat more complicated. Eleven different types
For example, consider a dale that gradually fills of possible combination on a change tree exist for
with water. The point where the water eventually areal combinations (see Table 8); Scott (1998) gives
flows over the edge of the dale is a saddle point. The a full descriptive account of such combinations.
pit in the dale is connected to this saddle point, as The logical steps developed below highlight how
indicated in the change tree. By continuing to fill the the areal combination algorithm has been developed
new lake, the next important occurrence where for a fuH change tree. This algorithm can be modi-
water flows out of the lake is again another saddle fied for hill and dale combinations, although for dale
point. As before, the line on the change tree repre- combinations (see Table 8) six occur, but in practical
sents the contour of the lake on the shore line, being terms types 3 and 11 are not usually applied, leaving
connected to this saddle point in the change tree. only four basic types of dale combinations.
This process can be continued, establishing connec- The following rules apply when attempting to
tions between the pits and saddle points in the develop an outline for the areal combination algo-
change tree. If the landscape is inverted, so that now rithm for a full change tree, but they are not a
the peaks become pits and so on, then a similar comprehensive account of all the potential hill and
process establishes connections between peaks and dale combinations, as they can be modified, as expe-
saddle points and the change tree. In general, at least rience dictates:
three types of change tree can be utilised:

Table 8. The application of areal combination types

Type 2 3 4 5 6 7 8 9 10 11

Dale combination )( t/ t/ t/ )( )( t/ )( )( t/ t/
Hili combination t/ )( t/ )( t/ t/ )( )( t/ )( t/
Full change tree t/ t/ t/ t/ t/ t/ t/ t/ t/ t/ t/

After Scott (1998)

Step 1 - find all Maxwellian hills and dales, then the forming process, enabling adherence and reten-
generate the full change tree. tion of surface conditions such as its painted appear-
Step 2 - classify all peaks, pits, edge peaks and edge ance. Typically, the stochastic surface structures
pits which are significant, or not, according to the resulting from processing by shot-blast texturing do
function of the surface. not always meet the demands of production for the
Step 3 - combine non-significant peaks and pits final sheet product. Both laser and electron beam
using combination types 1,2,3,6 and 7. texturing provide surface topographies of a deter-
Step 4 - combine non-significant edge peaks and ministic nature, which are fundamentally different
edge pits that do not involve a fourth-order from shot-blast rolling production processes. There-
saddle, using combination types 4,5,9, 10 and 1l. fore, the more commonly described two-dimen-
Step 5 - combine possible type 8 combinations, if the sional parameter characterisations will not be
fourth-order saddle has at least one of the pertinent here for textured sheet surface assessment.
following: a non-significant edge peak and edge Adequate characterisation for deterministically
pit. textured metal sheets necessitates suitable 3-D para-
Step 6 - if no further combinations occur, cease, or meters (mentioned above) to describe features of the
otherwise go back to Step 4. sheet topography.
In this model, in the case of mixed lubrication the
The resulting change tree indicates the signifi- load on the surface can be transmitted by three
cant peaks, pits, edge peaks and edge pits, together distinct bearing ratios, these being the solid contact
with the relationships between them. This "pruned" and lubricant contact area, both the static and
change tree has reduced noise, yet it still retains dynamic lubricant of the pockets. In such situations,
relevant topographical information concerning the the surface ratio of the solid contact relates to the
three-dimensional surface. relative quantity of the real contact area, this being
equivalent to the asperity area of contact. Dynamic
lubricant pockets are the zones oflubrication having
a connection to the boundary of the loaded area.
2.5 Textured metal sheets Hence in the process of sheet forming this lubricant
is squeezed out. Under such conditions the trans-
Textured sheet metal is produced in rolling mills mitted load is the result of hydrodynamic press ure,
(using rolls that have previously been suitably which in turn depends upon the rheological para-
textured) which enable a textured surface to be meters of either the lubricant or the surface's
reproduced on the sheet as it passes through the roll structure. By way of co nt rast, the pockets of static
stand. The techniques that have been used to texture lubrication have no connection to the boundary of
rolls with a three-dimensional deterministic struc- the loaded area, thereby trapping the lubricant,
ture are laser beam and electron beam texturing of resulting in hydrostatic build-up of pressure. For
the roll's peripher al surface. As a result of the deter- actual sheet surfaces, these pockets of static lubri-
ministic nature of the resulting topographies, they cant will only transpire when a ftattened surface
have completely differing features, requiring appro- topography condition occurs. Normal stresses in
priate techniques to characterise them. Such suitable the sheet can be transmitted by the three contact
parameters can be derived from a so-called mechanisms, while shear stresses may only result
"mechanical-rheological model" (see Figure 76). from solid contact. From the ratio of solid contact
The basic parameters of this method are: associated with dynamic and static pockets of
lubrication, it is apparent that the bearing ratios
• the material area ratio, a ma ; play a significant part in inftuencing tribological
• the open void area ratio, a op ; properties in forming processes. Thus, from these
• the closed void area ratio, a cl • bearing ratios of the mechanical-rheological model,
the three-dimensional parameters are derived.
Other methods can be employed to texture the From Figure 76, the relative portion of the solid
rolls and hence the sheet, but here the discussion will contact represents the material area ratio. In brief,
centre on the nature of deterministic surface sheet the three-dimensional parameters for either the
structures. One of the functions of a generated sheet dynamic or static lubricant pockets are the open
surface is to inftuence the tribological properties in void and closed void area ratios, respectively.
 L
______M_e_c_h_an_i_ca_
-_r_he_O_
o_~_·c_a_I
model
____~I~,
_ _
___ T_h_r_e_e-_d_im_e_n_Si_o_na_I_S_Url
parameters __ ac_e__~

F F
I I
'L - - - - - - - - - - /J
/:
Tool

/
/
/
F /

~~::-
I
I
I
I

Solid contact u"", Material area ratio um.

Static lubricant Closed void area

pockets U,tat ratio UcJ

Dynamic lubricant Open void area

pockets u dyn ratio (lop

Figure 76. Definition of three-dimensional surface parameters - textured metal sheets. (After Pfestorf et al., 1998.)

formation data streams from a multiple-sensor

2.6 Surface topography source. The advantages of ANNs are that they can
characterisation by neural be employed to integrate and fuse data, then adapt
to instructed environments, coupled to their robust-
networks ness to noise, fault tolerance, simultaneous pro-
cessing and feasibility of on-line realisation, via
hardware implementation. These inherent capabili-
Introduction ties of ANN s are composed of many simple pro-
cessing nodes that operate simultaneously. The
Prior to mentioning in detail the subject of neural functional behaviour of the overall system is
network techniques, these have as yet to be truly primarily determined by the pattern of connectivity
utilised in an industrial context, particularly from a of the no des. The ANN algorithms can be many and
decision-making outcome of the anticipated surface varied, with a typical simple architecture shown in
characterisation. However, such techniques have Figure 77, based on a "feed-forward multi-layered
been included, as they may see further development perceptron" .
in the future, as some of the research literature is Operationally, this ANN has input layer neurons
currently indicating. that send out signals across the hidden layer(s) in the
Artificial neural networks (ANNs) mimic the network to the output layer, with the desired
networking of the functional behaviour ofbiological and actual outputs being compared to evaluate the
neurons, and can be used to integrate and fuse in- system error - usually Euclidean error. This system
 Input
vector

Output

Output
layer of
nodes

layer of
Input nodes
layer of "Feed-forward ANN"
nodes
NB: these "multi-layered perceptrons" (i.e. ANNs) can also utilise "error-back-propagation".

Figure 77. Sehematic diagram of a feed-forward "artifieial neural network" (ANN) arehiteeture.

error is applied to adjust the node strengths of con- analysis by the ANN approach. The learning algo-
nectivity amongst the network's neurons, via a feed- rithm constrains the arrangement of pattern con-
forward propagation in the case of Figure 77. Hidden nectivities, although the concept of these learning
layer nodal capabilities are developed during net- algorithms can be adapted to any architecture.
work training, so that the extracted features are From the training data, the fundamental ANN
optimised for the classification task. Within the learning problem can be resolved in the follow-
ANN's algorithm, it can be configured around ing manner, by determining a set of network weights
the number of input-to-hidden-to-output nodes re- that can operate in the required manner. This assign-
quired for a given application, which may necessitate ment of weights can be achieved by initially assign-
some initial trials in order to optimise the correct ing minute values of randomness to these weights,
architecture. The greater the layers and nodal struc- then progressively adjusting them according to the
ture the better the discrimination, but at the expense following formula:
of longer "training" of the ANN. Hence a "trade-off"
occurs between a highly accurate and robust system, Adjustment of new = Wold
W jk jk
+ .uW jk
A

termed "dimensionality", to one of slightly less dis- modal weights:

criminatory ability and with faster training. where Wjk is the weight from unit k to unit j; and
Many ANNs can perform computational tasks by L1Wjk can be calculated in a manner of ways,
learning from examples, then generalising them according to different algorithms.
to new conditions. PracticaIly, these examples of Two major paradigms of learning can be
training - sometimes referred to as training data, employed:
patterns or vectors - are normally observations
associated with modelling the physical process. • supervised learning - where a "teacher" is neces-
ANNs can still extract the underlying structure from sary to facilitate the network's response during
training data that might be corrupted by noise or are training;
fuzzy in nature, providing a virtual representation of • unsupervised learning - where no "teacher" is
the modelled physical process. As a result of this present and the network must discover for itself
compliance, surface topography data fits weIl to the underlying structure while training.
 InClustrial metrology . .

Surface topography utilising ANNs tion (ACO), which has been the goal for many years
in unmanned manufacturing environments.
In any application of 3-D surface topography assess-
ment utilising ANNs, reduction of dimensionality is
vital, as training becomes lengthy and tedious, being
exacerbated by considerable noise. Dimensionality 2.7 Non-contact measurement
reduction can considerably restrict the number of
degrees of freedom of the ANN, typically from
16,384 (128 2 data points), down to 1024 (32 2 data Non-contact measurement provides a means of
points); the latter might be employed to suitably assessing not only rigid components but, more
train the network from a moderate amount of data importantly, fragile and delicate parts which would
samples. either distort under the measuring pressure -
For example, to perform an intra-surface char- causing induced measurement errors - or, more
acterisation - to discriminate between worn and significantly, degradation to a delicate and unsub-
unworn surfaces for a particular manufacturing stantial surface under test. In fact, for the latter case,
process - it is desirable to identify surface topo- it is the only real means available to inspect soft and
graphy features, whose presence/absence might sometimes pliable part surfaces with any degree of
uniquely characterise the surface. When attempting confidence in a speedy and efficient manner.
to discriminate between, say, the worn and unworn Many of the optical systems previously mentioned
surface conditions that might be apparent from an in Chapter 1, can operate in 3D, especially the inter-
isotropically manufactured surface, such as honed ferometers, confocal microscopes and triagulation
bores in automobile engines, then a demarcation systems.
between them by some geometric or statistical para-
meter provides a degree of objectivity in the system.
An obvious choice for this plateau-honed surface in
the selection of three-dimensional parameters would
be one to measure amplitude, such as that of disper-
sion Sq ("root mean square deviation" - relating to
surface departures within the sampling area). Shot-
blasting is another production process that can
benefit from utilising the Sq parameter. Conversely,
for the more anisotropic a surface topography, such
as that resulting from face turning, the greater would
probably be the discrimination by using another 3-D
parameter, such as the spacing parameter Sds
("density of peaks" in the unit sampling area). The
major strength of the ANN approach to 3-D surface
topography assessment lies in the fact that, once
trained, little human intervention is required - elim-
inating the tedium associated with the conventional
approach - and consistency occurs, provided the
basic data integrity rules have not been corrupted.
Systems can be built that are independent of the
methodology of data capture and a practical system
might link the ANN to an automated data capture
station - such as a vision system - to investigate, for
example, on-line the effects of process variables for
a turned surface finish. Such a system would be able
to highlight any changes within the production
process, e.g. the too1's flank wear, which might
unduly affect the surface finish. The complete system
could then be incorporated into a feedback loop to
modify some or all of the cutting data, such as the
feedrate, cutting speed, depth of cut or coolant, as
necessary to maintain consistent turned products.
This artificial intelligence (AI) technique can be Figure 78. Multi-sensing scanning instrument for contact/non-
considered as a form of adaptive control optimisa- contact 3-D measurement. ((ourtesy ofTaylor Hobson.)
 Surface texture: three-dimensional

Table 9. Technical data far 3-D multi-scanning instrument

Slides

Traverse length 97 mm x 97 mm (X, 11

Motorised column 83mm (l)

Straightness:
Profile (measured in Xdirection) 10 mm traverse 50 mm traverse 95 mm traverse
Peak-to-valley 0.25 fl-m O.6fl-m 0.9fl-m
PHT p (O.l t099.9%)

Surface (measured in X-Ydirection) lOmm 2 50mm 2 95mm 2

Peak-to-valley 0.7 fl-m 1.6fl-m 3.5 fl-m
SHTp(O.l t099.9%)
Minimum spacing X=0.5 fl-m; Y=5 fl-m
Maximum scanning speed 10500 fl-m /s
Maximum sampie load 5 kg

Inductive gauge Laser gauge

Vertical resolution 0.06 fl-m (60 nm) 1 fl-m

Range 2.5mm Vertical accuracy 1 fl-m (on diffusing surfaces
Principle FTSS inductive probe homogeneous optical
properties)
Range 10mm

Principle illumination Laser beam triangulation

Maximum angle of
measurement surface 90 0

Courtesy ofTaylor Hobson.

Non-contact three-dimensional surface texture Technically the instrument, as illustrated in Table

measurement offers considerable flexibility for a wide 9, has a comprehensive and versatile specifica-
range of component assessment. With the instrument tion, which can be employed in a host of industrial
illustrated in Figure 78, scanned surface areas are applications, ranging from soft material analysis
quite large, typically 10 mm 2 , whieh can be completed - cosmetics, biomedical and paper technologies -
bya laser probe in approximately :; minutes. This par- through to precision engineering, advanced mater-
ticular instrument type is very versatile, as it can be ials, semiconductor and automotive/aerospace tech-
used with either an inductive laser or dual-gauge nologies. However, they do have some limitations
probe. The non-contact probe utilises laser triangula- when compared to two-dimensional surface capture
tion technology, so that speed and large surface areas techniques - see Chapter 1.
can be mapped, whereas the contact probe is of the
inductive gauge variety that has been extended to a
range of 2.5 mm. Switching from a non-contact to a Holographie interferometry
contact operation can be completed in 5 s. Both single
and multiple areas on any component shape are easily Holographie interferometry utilising a non-contact
selected, ranging from a minimum size of 50 J.Lm 2 up 3-D measuring instrument is shown in Figure 79, the
to a maximum of 97 mm 2 • instrument being connected to a pe. This particular
The instrument has Windows™ "Talymap 3-D" instrument operates solely in a non-contact mode,
analysis software, offering a range of functions, with the capability of accurately measuring compo-
including view rotation, axis directions, automatic nent features over an area of 300 mm 2 • The system
levelling, resampling, defect rem oval, multiple pro- has been designed with no moving parts, ensuring
files and cylindrical/spherieal/polynomial form reliable and repeatable measurements. The measure-
removal, together with multiple output formats and me nt is undertaken by simply positioning the
custom colour palettes. component to be inspected in the cabinet, then flood
 InClustrial metrology , .

Figure 79. Non-contact holographie imaging instrument for

3-D measurement. ((ourtesy ofTaylor Hobson.)

illumination is applied aeross the eomplete field of delieate/pliable parts, as previously mentioned, is
view, prior to extensive data being obtained during relevant with this teehnique.
inspeetion. This measuring eyde is aehieved in The software features are enhaneed for both
approximately 2 minutes, with a Z-resolution of visual and surfaee measurement output, providing a
0.1 fLm; the system ean also eater for batehes of eomprehensive 3-D reporting proeedure, and this
parts to further inerease throughput. The instru- allows the following advantages:
ment is equipped with three field-of-view options -
50 mm 2, 175 mm 2 and 300 mm 2 - with eaeh eapable • measurement of eomplex and diseontinuous
of aehieving >1,000,000 data points. The measure- surfaees for flatness;
ment is aehieved by a 3-D holographie inter- • automatie levelling of the workpieee measure-
ferometer, utilising a diode laser, with precision ment planes;
opties and a diamond-turned parabolie mirror, • ineorporation of advaneed funetionality, the
eoupled to a high-resolution CCD eamera having system being simple to operate.
proprietary data-proeessing software algorithms.
The depth of foeus is 100 mm (fixed window), with Holographie interferometry is at the pinnade of
a Z-axis display resolution of 100 nm. All the advan- development for three-dimensional eapabilities of
tages relating to precision measurement of hardl today's instrumentation, offering new means of
solid parts are still valid for non-eontaet inspee- visually representing 3-D surfaees, giving unique
tion; in addition, the enhaneed ability to inspeet insights into surfaee inspeetion assessment.
 Morrison, E. The development of a prototype high-speed stylus
References profilometer and its application to rapid 3d surface measure-
ment. Nanotechnology 7, 1996,37-42.
Journal and conference papers Pfestorf, M. Engel U. and Geiger, M. 3D-surface parameters and
their application on deterministie textured metal sheets. Inter-
Ahmose. The Rhind Mathematical Papyrus. Year 33 of the Reign national Journal of Machine Tools Manufacture 38(5-6),1998,
of King Auserre (Apophis): Fifteenth Dynasty, Second Inter- 607-614.
mediate Period (Ancient Egypt), mid-1600s Be. Radhakishnan, V. and von Weingraber, H. Die Analyse Digital-
Bay, N. and Wanheim, T. Contact phenomena under bulk plastic isierter Oberllaechenprofil nach dem E-system. Fachberichte f( r
deformation conditions. Advanced Technology of Plasticity 4, Oberflaechentechnik (11112) 1969,215-223.
1677-169I. Russ, J.R. Fractal dimension measurement of engineering sur-
Blackmore, D. and Zhou, G. A new fractal model for anisotropic faces. International Journal of Machine Tools and Manufacture
surfaces. International Journal of Machine Tools and Manu- 38(5-6),1998,567-57I.
facture 38(5-6),1998,551-557. Sacerdotti, E, Griffiths, B.J., Butler, e. and Benati, E Surface topo-
Burrows, J.M. and Griffiths, B.J. A vector modelling technique for graphy in autobody manufacture: the state of the art. To be
the representation of 3-dimensional surface topography. published.
International Journal of Machine Tools and Manufacture Scott, P.J. Foundations of topological characterization of surface
38(5-6),1998,537-542. texture. International Journal of Machine Tools and Manu-
De Chiffre, 1., Christiansen, S. and Skade, S. Advantages and facture 38(5-6),1998,559-566.
industrial application of three-dimensional surface roughness Sherrington,1. and Howarth, G. W. Approximate numerical models
analysis. Annals of the CIRP 43(1), 1994,473-478. of 3-D surface topography generated using sparse frequency
Dong, W.P., Sullivan, P.J. and Stout, K.J. A comprehensive domain descriptors. International Journal ofMachine Tools and
study of parameters for characterising three-dimensional Manufacture 38(5-6),1998,599-606.
surface topography. III: Parameters for characterising Smith, G.T. Developments in 3D characterisation of surfaces,
amplitude and some functional properties. Wear 178, 1994, Metalworking Production August 1998, 59-62.
29-43. Sobis, T., Engel, U. and Geiger, A. A theoretical study of wear sim-
EOOO.14.015.N. Etats Geometriques de Surface Calcul des Criteres ulation in metal forming processes. Journal of Materials
de Profil, August 1983. CNOMO. Processing Technology 34, 1994,233-240.
Griffiths, B.J. Manufacturing surface design and monitoring for Tholath, J. and Radhakrishnan, V. Three-dimensional filtering of
performance. Surface Topography 1(1), 1988,61-69. engineering surfaces using envelope system. Precision
Hall, e. and Griffiths, B.J. A comparison of two and three-dimen- Engineering 23, 1999,221-228.
sional stylus surface measurement. Quality Today July 1999, Whitehouse, D.J., Parameter rash: is there a cure? Wear 83,1982,
S30-S34. 75-78.
Haesing,J. Nestimmung der Glaettungstiefe rauher Flaechen.PTB- Williams, J. Surface data puts users in the picture. Machinery and
Mitteilingen, 4, 1964,339-340. Production Engineering December 1989,1-15.
ISO 12085. Geometrical Product Specifications - Surface Texture -
Profile Method - Motif Method for Measurement of Surface
Roughness and Waviness. Books, booklets and guides
Kiran, M.B., Ramamoorthy, B. and Radhakrishnan, V. Evaluation
of surface roughness by vision system. International Journal of Bishop, C.M. Neural Networks for Pattern Recognition. Clarendon
Machine Tools and Manufacture 38(5-6),1998,685-690. Press, Oxford, 1995.
Kweon, 1.S. and Kanade, T. Extracting topographie terrain features Blunt, 1. et al. Development of Methods for Characterisation of
from elevation maps. CVGIP: Image Understanding 59(2),1994, Roughness in Three Dimensions. Penton Press, 2000.
171-182. Blunt, 1. and Stout, K. Three Dimensional Surface Roughness.
Li, C., Dong, S. and Zhang, G. Evaluation of the root-mean-square Hermes Penton Science, 2002.
of 3D surface topography. International Journal of Machine Stout,1. and Blunt, 1. Surface topography. Hermes Penton Science,
Tools and Manufacture 40, 2000, 445-454. 2002.
Littlefair, G. et al. On-line tool condition monitoring using artifi- Falconer, K. Techniques in Fractal Geometry. Wiley,New York, 1996.
dal neural networks. Insight 38(5), May 1996,351-354. Hertz, J., Krogh, A and Palmer, R. An Introduction to the Theory of
Littlefair, G., Javed, M.A. and Smith, G.T. Fusion of integrated Neural Computation. Addison-Wesley, Reading, MA, 199I.
multi-sensor data for tool wear. Proc of IEEE Conference, Perth, ISO Technieal Report 16610 Parts 5 and 6. Geometrical Product
Australia, November 1995. Specifications (GPS): Data Extraction Techniques by Sampling,
Long, E, Fractals, engineering surfaces and tribology. Wear 136, Part 5: Concepts of Envelope Filters, Part 6: Envelope Morpho-
1990,141-156. logical Techniques ISO.
Lopez, J. Hansali, G., Le Bosse, J. and Mathia, T. Caracterisation Mehaute, A. le. Fractal Geometries: Theory and Applications.
fractale de la rugosite tridimensionelle d'une surface.]ournal de Penton Press, 199I.
Physique III 4,1994,2501-2519. Robertson, W.S. Lubrication in Practice. Macmillan, New York,
Mainsah, E. and Ndumu, D.T. Neural network applications in 1987.
surface topography. International Journal of Machine Tools and Russ, J.c. Fractal Surfaces. Plenum Press, New York, 1994.
Manufacture 38(5-6), 1998,591-598. Stout, K.J. et al. The Developments of Methods for the Charac-
Mainsah, E., Sullivan, P.J. and Stout, K.J. Effects of quantisation on terisation of Roughness in Three Dimensions. Pub. no. EUR
3-D characters. Measurement Science and Technology 5(2), 1994, 15178EN, Commission of the European Communities, 1993.
172-18I. Stout, K. and Blunt,1. Surface Topography. Hermes Penton Science,
Majumber, A. and Tien, e. Fractal characterization and simulation 2002.
of rough surfaces. Wear 136, 1990, 313-327. Thomas, T.R. Rough Surfaces (2nd edn). Imperial College Press,
Mandelbrot, B. Self-affine fractals and fractal dimension. Physica 1999.
Scrip ta 32, 1985,257-260. Vicsek, T. Fractal Growth Phenomena. World Scientific, Singapore,
Maxwell, J.c. On hills and dales. London, Edinburgh and Dublin 1992.
Philosophical Magazine and Journal of Science 40(4), 1870, Whitehouse, D.J. Handbook of Surface Metrology. Institute of
421-425. Physics, Bristol, 1994.
 Surface microscopy

"With affection beaming out of one eye,

and calculation shining out of the other."
(Martin Chuzzlewit, Charles Dickens, 1812-1870)

101

G. T. Smith, Industrial Metrology

© Springer-Verlag London 2002
Summary of surface microscopy 3.1 Introduction
What was said ab out the advantages to be gained in Surface microscopy covers a wide range of metallo-
visual recognition of surfaces for three as opposed graphicallmorphological and instrumental tech-
to two dimensions is equally the case for surface niques used either directly, in the case of an optical
microscopy. Here, the optical definition is signifi- microscope, and indirectly, via scanning electron,
cantly greater and so will be the depth of field in transmission electron and atomic force microscopes
focus, particularly for imaging with scanning elec- as a means of visually depicting surfaces. These and
tron microscopes. Measurements taken on the other techniques for viewing the surface topography
surface and particularly those for two-dimensional are complemented by spectrographical analysis
surface topographical characterisation are quantita- instrumentation, typically Auger electron and X-ray
tive in nature. In the case of three-dimensional photoelectron microscopic methods. This chapter
imaging, if this is compared to the surface geometry will be principally concerned with electron micro-
information obtained from two-dimensional sur- scopic techniques, most notably:
faces, then the surface characterisation data in the
main will tend to be possibly more qualitative. • scanning electron microscopes (SEM);
Therefore, any attempt at comparisons between the • transmission electron microscopes (TEM);
different systems becomes somewhat superfluous • atomic force microscopes (AFM);
and, at best, would be misleading. This takes nothing • X-ray photoelectron spectroscopy (XPS).
away from the fact that surface microscopy has
superb three-dimensional surface visualisation and In the past and indeed currently, many inspec-
as an analytical tool offers considerable and valid tion, quality assurance and research and develop-
topographical details that can be extracted from the ment facilities employ conventional optical micro-
surface or subsurface, with certain configurations scopy as the main means of surface assessment.
and types of equipment. However, such instruments are not without their
Unlike the previous surface characterisation limitations, with the best optics giving magnifica-
instruments described in Chapters 1 and 2, much of tions of around x2000, which can be somewhat
the equipment to be discussed in this present limiting for minute surface detail inspection, or
chapter can interrogate the surface for a wide range when a larger "depth of field" (i.e., the vertical height
of elemental and topographical surface detail analy- which remains in focus) is necessary. Even at x10
sis. Moreover, in many instances, the specimen magnification, the depth of field is quite small,
under test can be manipulated about aseries of rota- requiring sampie preparation as thin sections,
tional axes, to obtain the best possible angle for "tapered sections", or polished and etched surfaces
surface assessment. Magnifications can be increased to yield meaningful information at the instrument's
or decreased quite readily, enabling minute surface higher magnifications. Yet another frustrating
features to be visualised, then quantified by an occurrence is that although it is possible to see what
incredible array of analytical techniques. This en- might be causing the problem, it is not possible to
ables the operator to perform sophisticated analysis determine by visual investigation what it is, there-
in situ, without having to change operational set-ups fore no explanation or estimate for a eure to the
to any significant degree. problem can be given and it remains one of pure
This chapter cannot attempt to cover in anything speculation. By utilising electron microscopic tech-
but superficial detail the information appertaining niques, these instruments overcome the drawbacks
to a component's surface, nor the complexity and found with purely optical microscopes, as they can
range of equipment currently available to the user. exploit spectrometric techniques, among others, for
Each surface microscopy instrument covered here in detailed analysis of surface chemistry.
this narrow selection could easily be expanded to fill NB Tapered sections at around 11 increase the
0

a complete book in their own right. However, it was apparent field of view, with minimal feature distor-
thought reasonable to include abrief res urne of tion. Etched surfaces are normally necessary on
some of the current equipment available and to metallic surfaces to preferentially etch the surface
attempt to show where, when and how such instru- and to reveal otherwise hidden surface detail.
ments complement our understanding of surfaces, In Figure 80(a) can be seen the application areas
which would otherwise be lacking if only two- and for a range of complementary electron microscopic
three-dimensional surface characterisation had instrumentation, indicating their vertical and lateral
been undertaken. resolutions. Such equipment can be exploited at a
range from the picometre (i.e., 10- 12 m) area for an
AFM, to an area weIl over 1 mm 2 using an SEM. In
 (a) Range of lateral and vertical resolutions of instruments

Vertical resolution
SEM

Imm

Inm

0.01Ä.
Ipm l Lateral resolution
1 1nm l~m Imm
lÄ.

STM/AFM
\ \ AES

(b) Resolution, excitation method and observations of instruments

STM/AFM: atomic force microscope;

SEM: scanning electron microscope;
TEM: transmission electron microscope;
AES : Auger electron spectroscopy;
XPS : X-ray photoelectron spectroscopy
OM : Optical microscope

Analytical Resolution Excitation Object of Observation

instrument Vertical Horizontal method observation environment

Tunnelling Atmosphere, gas,

STM/AFM 0.001 nm 0.1 nm electrons/atomic vacuum, liquid
force
Secondary
SEM 8nm 5nm Electrons vacuum
electrons
Transmitted
TEM 0.08 nm Electrons vacuum
electrons
AES 3nm 10 nm Electrons Auger electrons Ultrahigh vacuum
Characteristic
XPS 2nm 100 m Photoelectrons Ultrahigh vacuum
X-rays

Figure 80. Range and resolution of a selection of surface microscopy instruments. ((ourtesy of Jeol (UK) Ltd.)
 u ace mlcroscopy

Figure 80(b) a tabulated chart of the resolutions, to become proficient. Instrument size has been
excitations and observations available from this significantly reduced by utilising PC-based tech-
group of analytical instruments is shown. nology and the complete equipment is not much
larger than an office desk in terms of its working
area (see Figure 81).
Many of the latest instruments available can
3.2 Scanning electron examine objects up to 200 mm in diameter, weighing
microscope up to 3 kg, while still retaining micrometre precision
movement of the sampie (10- 6 m positional resolu-
tion). The SEM process of examination is non-
SEMs can overcome the limitations of optical destructive in nature, thereby retaining the value of
microscopy by offering depths of field of several an expensive sampie being investigated. As a quality
millimetres at low magnifications (x20-200) and control tool the SEM can quickly examine either
their resolution is similar up to x2000, whereupon random or sequentially produced sampies, which if
the SEM significantly extends its range, at x100,000, necessary can then be reintroduced into the produc-
or greater, over conventional optical techniques. tion line, if they prove to be acceptable. After many
Instruments of the SEM variety are not a new devel- years of development, the SEM has become a "work-
opment: were first designed in Britain about 50 years place-hardened instrument" that can be utilised in
ago, but then they were big, complicated and some- reasonably controlled production environments,
what temperamental instruments requiring great necessitating minimal component preparation with-
skill, patience (and some luck!) to produce generally out damaging the parts in any way.
mediocre results. The modern SEM is a totally
different instrument, controlled by PC technology
with Windows™-based software, having a minimum Basic SEM operation
of knobs and buttons to manipulate the specimen
and necessitating minimal training for an operator The principle of operation of the SEM is quite
simple. At the top of the column (see Figure 82) an
electron gun is situated, consisting of a tungsten fila-
ment in a strong electrical field. As a result of the
electrical field the electron gun emits electrons
(negatively charged atomic particles), which are
then accelerated to high speeds. These high-speed
electrons travel down the column, being influenced
by lenses lower in the column which squeeze them
together, forming an electron beam of very small
diameter. This electron beam is then focussed so that
it collides with the specimen in the microscope spec-
imen chamber as a diminutive spot. This minute
spot is scanned from left to right and up and down
over the surface, in an identical manner to that of a
normal TV as it scans the screen. By way of an
analogy for its visual operation, the SEM can thought
of as a TV manipulated so that its screen faces the
floor. Under these conditions, an SEM "tube" is rede-
fined and can be thought of as a column, whereas
the screen becomes the specimen being examined.
Although in the case of an SEM a considerably
smaller "screen" is scanned with an appropriately
smaller spot. In the same manner that anormal TV
tube is operated under a vacuum, the SEM column
and sampie chamber are also evacuated.
Once the electron beam strikes the specimen
surface many different processes occur, but for a
Figure 81. A typical "basic" scanning electron microscope, based on basic understanding of the mechanism it is only
proven PC technology and Windows TM-based software, which can be necessary to state that secondary electrons are gen-
equipped with a range of analytical tools for diagnostic investigation. erated. These secondary electrons can be thought of
[Courtesy of Jeol (UK) Ltd.) as electrons on the specimen's surface which will
 Sccondary
clcctrons Elcctron beam
X-rays /
..... /
/'
Baclm:attered
/'
..- eleclrons
Cathodo-
lumincscence __ Auger
electrons

b orbed
eleclrons

Backscattercd eleelrons
(in Ihe sevcral 10's of nm
range to 100 nm)
Electromotive ,/ ~LL.<..L.o<~,\

force
Characteristic
Transmitted _ X-rays (in the lJ.ßl
range)
electrons
Continuous
X-rays

Condenscr Icns
Where:
I =sizl! of clcctron bcam
scanning on the specimen
Objcctive lens aperture
L =size of image on the CRT
Magnilication M =/lL

Signal detector

idco signal amplifier

Cathode ray tube Dcnection coil

Figure 82. Underlying principles of the scanning electron microseope. [(ourtesy of Jeol (UK) Ltd.]
have been displaced by the high-speed electrons that the X-ray analysis indicates what causes them; given
originated from the electron gun. Any secondary this information it is possible to deduce why the
electrons that are emitted from the specimen surface problem arose. Once this level of understanding of
as the beam strikes are collected and counted. The the problem has been gained, preventive measures
total number of electrons counted for a particular can be defined and implemented.
point on the specimen is converted to a spot of light Apart from telling the investigator what a spec-
on the SEM display screen. The intensity of the spot imen's chemical constituent elements are, micro-
depends on the total secondary electron count. An probe analysis can determine respective quantities
identical operation is performed for each point on of each element. Such instruments can provide
the specimen being scanned and, an image is built- analytical accuracy to better than several parts per
up which represents the secondary electron yield million on a routine basis. X-ray detectors can also
over the surface. Fortunately, one of the main factors be "tuned" to detect only specific substances and
influencing secondary electron yield is the shape of then monitor their concentration while the electron
the specimen's surface, hence if the SEM display beam scans, thereby"mapping"the distribution over
(viewing screen - eRT) is scanned at the same rate the sampie surface. Using the SEM in conjunction
as the electron beam over the surface, an identical with X-ray analysis and mapping, provides all the
image occurs of the specimen's surface. To ensure information needed to characterise the problem,
ease of use, the SEM display and electron beam are thus:
synchronised at the TV frame speed, enabling a TV
image of the specimen surface to be viewed on the • imaging locates potential problem areas;
SEM screen. • normal X-rays determine what causes them;
The spot on the eRT is shown in real time, with • X-ray mapping deduces where else problems
the information emitted from the specimen surface occur in the specimen.
being displayed as an image. The magnification of
the displayed image is defined as the ratio of the size Yet another vital facility available to the SEM user
of the image on the eRT to that of the electron beam is backscatter electron (BE) detection. Here, some of
scanning on the specimen surface. The type of infor- the electrons that strike the specimen do not
mation obtained can be changed by switching the "interact" with it (i.e., producing secondary elec-
signal. In this way, the specific desired characteris- trons or X-rays), but simply collide with atoms in the
tics of the specimen surface can be viewed on the specimen and bounce immediately back - termed
eRT in a magnified scale. backscatter. The closer together the atoms are within
An SEM can be regarded as a miniature labora- the specimen, the more potential collisions, and
tory, with the specimen chamber being used for hence the greater the certainty ofbackscattered elec-
dynamic experiments such as heating, cooling, trons. This backscattering of electrons can be
tensile, compression and hardness testing, which detected, giving an atomic number contrast. Denser
can be examined in real time and at an appropriate materials exhibit tightly packed atoms, resulting in
magnification. Not only are secondary electrons increased collisions and causing more backscatter
collected for imaging, but also many other emissions than their lighter material counterparts, wh ich have
occur and can be employed to yield additional spec- looser packing densities. The resultant image in the
imen information. üf these additional emissions, eRT will indicate the well-known saying: "Dense is
perhaps the most useful one is the generation of bright and light is dark!" The technique is often
characteristic X-rays. When an electron beam strikes utilised to determine irregularities, since they
a specimen X-rays are emitted (see Figure 82), the usually appear as bright or dark areas on an other-
energies of which are directly related to the chem- wise uniform (normal) background. In many
ical constituents of the specimen at the focal point circumstances, the backscatter electron detector is
of the beam. This is the principle of microprobe utilised in conjunction with X-ray analysis, because
analysis, where a finely focused beam in the SEM is the backscattered electron image is normally less
aimed at a particular surface feature that has been complicated than the secondary electron image, and
identified by normal imaging, the object here therefore it is simpler to select areas for analysis.
being to determine its local chemical composition. X-ray analysis and backscattered electron detec-
Unlike analytical techniques based on "wet chem- tion are essentially the most widely used supplemen-
ical" methods, X-ray analysis is very fast, typically tary techniques, although many other characteristics
100 s being the standard acquisition time. As may be determined from the results of the inter ac-
opposed to normal optical microscopy where only tion of an electron beam with the specimen. Other
the image can be seen, using an SEM the chemical supplementary techniques include crystallographic
composition can be determined. Hence, an SEM and magnetic information, surface topography and
enables one to establish where problems occur and composition, semi-conductive or electrical charac-
 InClustrial metrology , .

teristics; however, it should be stressed that some of now that image-processing procedures have been
these methods require quite complex analytical developed, to further expand the visual impact to
stages to obtain valid information. the observer - more will be mentioned about this
shortly. The number of secondary electrons emitted
from a specimen surface greatly depends on the
Principle of magnification incident angle of the electron beam to that surface.
In other words, the secondary electron signal
If a specimen surface is scanned with a finely focused depends upon undulations of the specimen's surface.
electron beam of only a few nanometres (see Figure Furthermore, since the energy of secondary elec-
82), then information will be emitted from each point trons is very low, they are only emitted from a
of the scanning operation. This emitted information thin layer on the specimen surface. Thus, secondary
is converted into an electrical signal, amplified, then electron signals are considered to be the most
fed into the eRT for subsequent observation. On the appropriate signal for observing a specimen's topo-
eRT, the information is used to control the bright- graphy.
ness of the corresponding spot. The spot on the eRT The following describes just some of the fields of
is shown in real time with the electron beam scan- science and engineering that might exploit SEM
ning on the specimen surface. Hence, information observation techniques:
emitted from the specimen surface is displayed on
the eRT as an image. In Figure 82, the magnification • materials science - utilised in most advanced
of the displayed image is defined as the ratio of the materials science fields, typically for supercon-
size of the image on the eRT (L) to that of the size ducting materials, or for advanced composites,
of the electron beam scanning the surface (I). The etc.;
type of information that could be obtained can be • biology - particularly biotechnology applications
changed by switching the signal; in this manner a or for the examination of entomological taxon-
specific desired characteristic of the specimen can be omy (insects, bacteria, viruses and animal
visually displayed on the eRT in a magnified scale. tissues);
• electronics - utilises the power of such instru-
mentation in the R&D and quality control fields,
SEM applications typified by the semiconductor/electronics appli-
cations for failure analysis techniques, etc.;
As previously described in this section, various • mechanicallindustrial engineering - for investi-
kinds of signals can be obtained by an SEM. These gation of process technologies, machinability
signals carry distinct types of information and, as research, component failure modes and metal-
such, are employed for different purposes, as listed lurgicallmetallographical investigation.
in Table 10.

SEM: images
Topographical observation: secondary
electron imaging In Figures 83-86 are depicted aseries of photo-
micrographs produced across a diverse range of
This secondary electron image (SEI) technique is metallic - ferrous and non-ferrous - non-metallic
probably the most utilised method of general and biomechanical structures. It is only not only at
applications for an SEM by industry, particularly so the high-resolution range of magnification that

Table 10. Typical scanning electron microscope applications

Signal Mode of operation Pur pose of SEM - information carried

Secondary electrons SEI Topographical observation of surface

Backscattered electrons BEI Compositional observation of surface
X-rays X-ray Elemental analysis of specimen
Transmitted electrons TEl Internal structure observation
Cathodoluminescence CL Internal characteristics observation
Electromotive force EBIC Internal characteristics observation
Secondary electrons or ECP Crystalline structure
Backscattered electrons MOl Magnetic domain observation

Courtesy of Jeol (UKj Ltd.

 u ace mlcroscopy

(a) Cleavage planes on natural dia mond surface: 15 kV x200. tially worked to achieve the inclined geometrie
facets so admired in cosmetic jewellery. Conversely,
Figure 83(b) shows a more mundane, but neverthe-
less important surface - that of paper - which must
absorb the ink or colour pigments from printed
media. With such surfaces good adherence on the
surface is essential; therefore the quality and texture
of this paper surface are vital if it is to respond to,
say, the ink and not blot/spread, thereby losing
line delineation and definition. Figure 83c illustrates
a typical semiconductor memory surface, where
surface structure and separation of discrete memory
elements are vital for efficient operational perfor-
mance of these IC chips.
The two SEM photomicrographs displayed in
Figure 84(a and b) are notable not from the images
(b) Paper surface: 15 kV xl 00. themselves, which are markedly different in texture
and composition, but because of the information
provided overlaying the actual images. Of particular
relevance is the "error bar" or linear scaled bar which
automatically changes its dimensional magnitude as
magnifications are either increased or decreased.
Such an auto-scaling facility can be employed as a
form of "indirect measurement" of notable dimen-
sional/surface features, weIl beyond the range of
most conventional dimensional techniques, certain-
ly up to the nanotechnology level (10- 9 m) and even
approaching that of picotechnological values (10- 12
m) in size. Each SEM image can be labelIed with an
identifying number for subsequent later reference
and the mode of operation can also be displayed
(for example, Figure 84a: SEI = secondaryelectron
image).
(c) Semiconductor memory surface: 15 kV x1500.
In Figure 85 is shown an SEM "soft scanned" image
of an "ultra-high-speed" face milled austenitic-grade
stainless steel surface machined at 3500 m/min. The
SEM processed data in this instance was magnifica-
tion xlOO, accelerated voltage 20 kV, working dis-
tance 20 mm, pressure Pa 200 /-Lm, with secondary
electron imaging. üf particular note was that the
original photomicrograph was in fact a left-hand
stereo image of the surface. Both a left- and right-
hand stereo image can be superimposed/merged into
one image. If the original of such a "merged pair of
images"had been viewed through 3-D glasses (green
and red - as worn in the early days of 3-D cinematog-
raphy), then this combined image would take on a
three-dimensional aspect. This visual enhancement
Figure 83. Low-vacuum scanning electron micrographs. (Courtesy is often useful when attempting to interpret whether
of Jeol (UK) Ltd.) peaks or valleys are being viewed, this being a partic-
ular optical distortion that is visually difficult to
interpret in plan view, as shown in Figure 85. More-
useful analysis can be undertaken, but also at lower over, this face-milled image has a topographie al
magnifications, where these instruments' large height profile superimposed onto the surface, show-
depths of field are exploited. In Figure 83(a) can be ing peak and valley features, giving greater dimen-
seen a fractured natural diamond surface, illus- sional and visual interpretation of the machined
trating the cleavage planes which can be preferen- surface topography.
.. InClustrial metrology

(a) A processed photo-cell surface.

. .

MAG:xSOO

Acc.V:1SkV

Signal:SEI

WD:22mm

SS:20

Pressure:Pa

40.0pm

(b) High-speed face-milled surface: 316 grade stainless steel, peripheral speed 2500 m/min.

MAG:x2S

Acc.V:30kV

Signal:BES

WD:36mm

SS:40

Pressure:Pa

800pm

Figure 84. Scanning electron microscope images. [(ourtesy of Jeol (UK) Ltd.]
 u ace mlcroscopy Im
Height Profile
Heighl (pm)

Figure 85. 5canning electron microscope photomicrograph with "soft imaging" and topographical height profiles. [(ourtesy of Jeol (UK)
Ltd.]
SEM: image processing ports. Angled detectors are preferred for practical
reasons, because they are less likely to be damaged
Digital images can be subjected to a variety of post- by movement of the specimen within the working
processing functions, with processed images being chamber of the instrument. The X-rays are discov-
separately stored from the original images. ered by employing either a thick- or thin-frame
Therefore, when writing reports the image data can window (typically an Si (Li) detector), for heavier or
be converted into a suitable form for the content of lighter elements, respectively. This semiconductor
the report. Windows™-based software enables a host detector is positioned at the tip of the detector (see
of image processing techniques and enhancements Figure 87). The height of current pulses that are
to be carried out; some these are as follows: generated by the X-ray illumination is proportional
to the energy of incident X-rays. By calibrating
• On-screen measurement - this is a cursor the. multi:channel analyser prior to performing
measurement function allowing measurements test mg usmg a standard specimen, characteristic
between two cursors at 90° to each other (Figure X-rays from a previously unknown specimen can be
86i), or a multi-point measurement function measured for element identification. The sophist i-
which measures many points in any direction cation of these EDS systems enables the investigator
(Figure 86ii); to determine by an analysis of the specimen, its
• Look-up table (LUT) and pseudo-colour display - percentage weight of elements in the vicinity of the
when a stored image is to be downloaded, the sampling area, spectra for specific elements that
LUT allows the image to be optimised for visual might be located in the test zone, or a general spec-
impact, through adjustment of the contrast and trographical search for a totally unknown specimen,
brightness of the original image (Figure 86iii). when attempting to identify its overall elemental
Furthermore, the image can be visually enhanced composition.
by the application a pseudo-colour display (Fig- When an electron beam irradiates a specimen,
ure 86iv); characteristic X-rays are emitted (Figure 87). By
• Dual and quad displays - either two or four detecting and analysing these X-rays, identification
images can be combined into an overall new of elements contained within them can be under-
image (see Figure 86v and vi, respectively); taken (qualitative analysis). More specifically, it is
• Dual magnification and digital zoom displays - possible to determine weight concentrations of the
the digital zoom function provides an enlarged contained elements (quantitative analysis). Electron
portion of the stored image, which allows the beams are finely focused, so by utilising the "spot
images to be positioned adjacent to each other mode" on the instrument an elemental analysis of a
(Figure 86vii). The dual image shown in the very small area on the specimen can be investi-
previous Figure 86(vii) can be increased in size gated, or the more general averaged element concen-
to any desired magnification to observe notable tration, mentioned above. Furthermore, both a "line
features, for further investigation, or analysis analysis" and "areal analysis" can be performed as
(Figure 86viii). well as X-ray image observation, the procedure
being:
The following two sections relating to SEM
enhancement attempts to describe and illustrate observing the specimen surface with a secondary
several types of the large array of instrumental electron image and/or a backscattered electron
equipment that can be fitted to an SEM. Such inves- image;
tigative items considerably increase the versatility 2 designating the point to be analysed with the
and the scope of the SEM as an analytical too1. cross-point of the cursor lines displayed on the
image;
3 positioning the electron beam at the designated
point on the specimen, then performing a quali-
3.2.1 Energy-dispersive X-ray tative analysis;
spectrometer 4 observing the distribution of an element, by
electing a specified characteristic X-ray.
The characteristic X-rays emitted from the specimen
under test have energies that represent individual Such procedures as those just described for EDS
elements. These energies can be detected with an analysis enable considerable elemental data to be
energy-dispersive spectrometer (EDS) fitted at a derived, then interpreted for appropriate future
specific location port on the SEM. Its detector orien- actions, based on sound and factual judgements.
tation and angular relationship to the axis of the X-
ray beam is conveniently located in one of the SEM's
 u ace mlcroscopy Im
(i) Cursor measurement (ii) Multi-point measurement

(iii) Look-up table (iv) Pseudo-colour display

(v) Dual display (vi) Quad display

(vii) Dual magnification display (viii) Digital zoom display.

Figure 86. SEM image-processing techniques. [Courtesy of Jeol (UK) Ltd.l

l1li InClustrial metrology . .

Elcctron bcmn Operation/display unil

X-ray dctcctor

Liquid nitrogen DC\I'ar

l\lulti-channcl
analyser

Si ( Li) detcctor

Figure 87. The energy-dispersive X-ray spectrometer (EDS) - block diagram. [(ourtesy of Jeol (UK) Ltd.]

3.2.2 Transmitted electron image 3.3 Transmission electron

Prior to abrief discussion of the operational proce- microscope (TEM)
dure of this instrument that can be fitted to the SEM,
it is should be mentioned that the operation can be
performed in its own right on a purpose-built With the recent developments in the fields of high-
instrument - see the following section. If a specimen temperature superconductors and the introduction
is thin enough for the irradiated beam to penetrate of new materials, more sophisticated instrumenta-
it, electrons will be scattered during its subsequent tion has been created. Research at the atomic level
penetration (see Figure 88). has become quite commonplace, being essential for
The degree of electron scattering depends on the a realistic and an interpretative understanding of the
product of the density of the specimen's surface (p) behaviour and properties of such materials. The age
and its thickness (t). Namely, any of the electrons of research into subatomic operational activities on
penetrating through a less dense area (having a both nano-structures and nano-fabrication can be
small "p x t" value) will not be as widely scattered said to be in existence today. The transmission elec-
as those from a more dense area (large "p x t" value). tron microscope (TEM) is a field-emission, high-
Thus, a larger signal is detected through the less resolution analytical instrument, which has many
dense area. When a thin section is cut from a bulk operational applications and functions, but most
specimen and observed with transmitted electrons, notably it has been widely employed for the investi-
information concerning the inner structure of the gation into new material developments. A typical
thin section can be obtained, this being the basis for instrument is illustrated in Figure 89 and can be
the transmitted electron image. configured with, say, a 200 kV thermal field emission
gun having approximately 100 times brighter image
 u ace mlcroscopy

canninl: coil

ScanninJ.l clc('lron Scanninj:

üb ervation scrccn
beam circuit

/ J_
Dclcclor

/~\I
/: "

\ ,.,

" --_/
\ / SIJ.:nal IIltcnslty
/

Figure 88. The principle underlying the SEM transmitted eleetron image observation, [(ourtesy of Jeol (UK) Ltd.]

than that found on many earlier instruments, basic diagrammatic representation of aTEM is
without sacrificing long-term emitter stability. One shown in Figure 90. Recent progress in analytical
of the advantages of high-grade versions of this microscopy has enabled improved theoretical reso-
instrument is its ability to instantaneously select the lution, this being calculated from the spherical aber-
analysis point, while observing a high-resolution ration coefficient, with yet further enhancements in
image, with certainty that both the observed and analytical capability through improvements in the
analysis points coincide. Other applications for illumination system. Such improvements include
TEMs might include electron holography and micro-area diffraction, convergent-beam electron
coherent convergent beam electron diffraction. diffraction and high-sensitivity analysis employing
finely focused probes.
The illumination lens system might consist of a
Operational characteristics of the TEM series of lenses (four), in conjunction with an
imaging lens system (i.e., six lenses) - see Figure 90
A typical field emission gun might use either a (many of the lenses are not shown, for simplicity).
ZrO/W (100) emitter, or a W (100) emitter. A very This imaging system provides various illumination
 InClustrial metrology . .

Condenscr lens

Objecti\'e lens aperture

Figure 89. Transmission electron microscope, equipped with:

• an energy dispersive X-ray speetrometer;
• a parallel deteetion eleetron beam loss speetrometer (PEELS);
• A seanning image observation deviee (ASID); and
• TV units.
[(ourtesy of Jeol (UK) Ltd.]

modes to accomplish high resolution and diffrac-

tion, together with analytical requirements for the Figure 90. Diagrammatic representation of the TEM. [(ourtesy of
microscopist, while the imaging lenses provide high Jeol (UK) Ltd.]
magnification, rotation-free imaging and orienta-
tion correspondence between the diffraction pattern
and its associated image. Typical values for the accel- Additionally, the TFE has a high probe current,
eration tube evacuation are of a high order, being ensuring that it is an ideal electron source for analyt-
around 3 x 10~8 Pa. The specimen chamber vacuum ical microscopy.
pressure range is normally up to 3 X 1O~5 Pa. The advantages of the field emission gun can be
In general, a field emission gun is available as two summarised as follows:
distinct types:
1 the field emission has approximately two orders
• cold field emission (CFE) variety - which uses the of magnitude higher brightness than a LaB 6 fila-
emitter at ambient temperatures; ment and provides a considerably greater probe
• thermal field emission (TFE) variety - which current - in a probe of less than several nanome-
heats the emitter to approximately 1600K. tres;
2 the specimen illumination angle can be config-
The CFE normally features a small electron ured to be exceedingly small, due to the gun
energy spread, but has the disadvantage of both brightness; this, together with a small electron
long- and short-term emission fiuctuation, due to energy spread, enables the information limit to be
residual gas contamination of the emitter surface. improved;
Conversely, the TFE, being constantly heated, has no 3 this small energy spread of the emitted electrons
gas adsorbed onto the emitter surface, resulting in improves the energy resolution of the electron
extremely high stability of the emission current. energy loss spectrometer (EELS);
 (i) Ultra-high resolution. 1
d = - Cs a3
j 2
\
\ (minimum disc due to spherical aberration),
~ 1 \

! \
'\
'-::
~ 0.5 d3" " d /d 1
"--- /

°0~~~5~=d~1=0===215~~
a?~- ~ ==_ ___ Minimum disc due to chromatic aberration,
a(mrad) A
d3 = 1.22-
a
(ii) Specimen high-tilt configuration.
Blur due to diffraction (Airy disc).

Here, B, Ip, Cs and Ce are, respectively, the electron

~ 1 gun brightness, probe current, spherical aberration
! coefficient (Cs of objective lens pre-field), and the
~ 0.5 chromatic aberration coefficient. With the field emis-
sion electron gun, the probe diameter is determined
by d o' d j and d 2, owing to the realisation of a suffi-
5 10 15 ciently small and stable light source.
a(mrad)

Figure 91. Relationship between illumination angle and probe

diameter. [(ourtesy of Jeol (UK) ltd.] Operating functions and TEM ray
illumination lens system configurations
4 the minute effective source size and the high The TEM can resolve down to the atomic spacing of
coherence of the beam are effective for magnetic materials, as depicted in Figure 92(a), where the
domain observation and are essential for holog- probe has been focused on an Au particle, its profile
raphy. being taken through the line of measurement as
indicated in Figure 92(b). As mentioned above, the
TEM can utilise a number of operating modes, some
Functions and ray diagrams of of which are as follows:
illumination lens system
• Nanometre beam diffraction (NBD) - an NBD
In Figure 91 are shown the calculated values of the pattern can occur with high angular resolution
specimen illumination angle (a) versus the probe when illuminating the specimen with a
diameter (d), in both the ultra-high-resolution nanometre-sized probe having a small illumina-
configuration (i) and the specimen high-tilt config- tion angle;
uration (ii) for a typical TEM. From these graphs it • Energy-dispersive X-ray speetrometry (EDS) - a
can be seen that a probe diameter of 0.5 nm is wide range of elements can be investigated, from
obtained at a = 10 mrad with the ultra-high-resolu- light to heavy, which can be found by analysing
tion configuration (i) and 0.5 nm at a = 7 mrad with the energies of characteristic X-rays generated
the high-tilt configuration (ii). from a specimen. The optimum arrangement of
The probe diameter on the specimen is given by the detector and collimator allows for a good P/B
the following equations: ratio and high detection efficiency;
• Electron energy-loss speetrometry (EELS) - the
electrons that have passed through the specimen
are dispersed by a sector magnet to perform
energy analysis. This technique features high-
where energy resolution and the acquisition of infor-
mation on inner shell excitation.
d = _1 {iP
o
7TaV B A variety of ray diagrams can be produced using
aTEM, most notably:
(Gaussian image with no aberrations),
(a) An electron probe focused on a specimen and an (b) Profile on the line in the image at left.
Au particle, which were photographed on the
imaging plate (IP).

(c) An image of carbon graphite, observed with a scanning image observation device. A lattice image
of 0.34 nm is c1early observed, proving that the probe was this order of diameter.

Figure 92. Measured probe diameter and image of carbon graphite taken on aTEM. [(ourtesy of Jeol (UK) Ltd.]

atomic lattice images - with probes configured to where X-rays of known wavelength (A) and
approximately 1 nm diameter. Such diagrams are measuring angle «(}) can determine the atomic
visual representations of the physical associated spacing (d) for various planes/orientations in a
and geometrical positioning of atomic lattice crystal;
structures within materials; 3 convergent-beam diffraction - simplistically
2 diffraction patterns - diffraction can occur speaking, the atomic structure here closely resem-
whenever the Bragg law is satisfied, varying for bles that observed in elementary "bubble-raft
each material and its orientation, this being patterns", which are used to simulate simple
related in the following manner: grain boundaries and dislocations in crystalline
materials.
A = 2d . sin () (Bragg law)
3.3.1 Transmission electron surface reflection conditions are governed by the
"Bragg law", so they also must vary. The inter fe ren ce
microscopy: general application between both the transmitted and diffracted beams
will produce intensity variations of the brightfield
In X-ray diffraction, X-rays incident on a crystal are
image - termed extinction contours. Such an effect
diffracted by parallel sets of atom planes, in a similar
appears as dark bands across the image, which indi-
manner to that of light reflection from a mirror. The
cates that a set of planes near these bands has a
position of the diffracted beam will be in accordance
significant diffracting orientation. Therefore, when a
with the "Bragg reflection conditions", described in
foil bend occurs in the TEM during its observation,
the previous seetion, and they are similar to the way
"extinction contours" seem to appear to move across
an electron beam is scattered by a crystal. Any dislo-
the test specimen. From a simplified analysis tech-
cations and stacking faults in the surface region can
nique, this approach can be employed to account for
be revealed by a technique which examines the inter-
the contrasting effect resulting from surface regional
ference between transmitted and diffracted beams,
dislocations. Moreover, "stacking faults" in the spec-
resulting from a beam of electrons incident on a
imen will also produce a characteristic diffraction
crystal foil of between 100 and 500 nm in thickness.
image. Hence, the study of specimen defects by these
The technique for forming images of these diffracted
various diffraction techniques is applicable to all
elements has previously been mentioned. In oper-
materials that can be produced as adequately thin
ation, a parallel beam of electrons - typically accel-
sections.
erated by a potential of 100 kV - is transmitted
In order to show the potential of the TEM's abil-
through a thin foil and is diffracted in a variety of
ities as an analytical tool, Figure 93 illustrates a
directions by the crystal under test. The diffracted
TiAl/Ti3Al two-phase alloy in lamellar form. This
beams are subsequently focused forming a diffrac-
particular material has shown promise as a new
tion pattern that can be magnified for investigation,
dass of lightweight high-temperature structural
as necessary. A diffraction pattern (see Figure 93)
material. Three different types of TiAl/Ti3Al inter-
normally consists of a two-dimensional array of
variant lamellar boundaries, in such a TiAI-based
spots, with each spot representing a specific series of
alloy containing Sn are observed by high-resolution
reflecting planes. From this diffraction pattern, the
electron microscopy. Of the three types of bound-
foil orientation can be determined.
aries shown in Figure 93, the ternary element Sn,
Individual diffraction pattern spots are the
which is added to improve the mechanical proper-
product of the incident beam's diffraction from a
ties, is observed to preferentially segregate onto two
reasonably large vicinity on the test specimen (typi-
types of high-energy boundaries. In Figure 93 the
cally 104 atom spacings in diameter). A single elec-
visual image and diffraction patterns, together with
tron beam can produce a corresponding image by
graphical displays, show just some of the investiga-
inserting an objective aperture into its path.
tive/analytical abilities of the TEM.
Two main optical conditions can arise:
• brightfield image - resulting from the aperture
being centred on the main transmitted beam;
• darkfield image - produced by situating the aper- 3.4 Atomic force microscope
ture over a diffracted beam.
Generally, the predominant amount of analysis The atomic force microscope is just one variant of a
occurs in the brightfield mode of illumination, wide variety of scanning probe microscopes and
although some critical applications required the accompanying techniques that were developed in
darkfield illumination technique. the early 1980s and indude the following:
Hypothetically, if a test specimen was completely
flat and without defects, under these conditions a • atomic force microscope (AFM) - contact; AC; also
completely homogeneous image would result when visco-elasticity operational modes;
the objective lens was centred over the main trans- • scanning tunnelling microscope (STM);
mitted beam. Subsequent changes in image bright- • scanning tunnelling spectroscopy (STS);
ness are the result of any effect that changes the • friction force microscope (FFM) - static and
diffracted beam's path, so that it enters the objective lateral operational modes;
aperture and interferes with the transmitted beam. • magnetic force microscope (MFM);
If very thin test specimen foils are buckled, the • scanning near-field optics microscope (SNOM).
orientation of the specimen's surface with respect to
the electron beam will be slightly varied for different This section on scanning probe microscopes, will
zones on the specimen. These variations for local almost exdusively discuss an overview of the AFM
 InClustrial metrology , .

(a) Ordered twin boundary

....,
AI, Ti (al. %)
42 44 46 48 50 52 54 56 58
20
15

10

e......c 5
0
c'"
'S -5
Q"

-10

-15
-20
0,8 1.2 IA 1.6 1.8 2
Sn (al. %)
--1
(b) Pseudo-twin boundary

AI, Ti (at. %)
J5 40 45 50 5S 60 6S
20+---~~~--~--~--~--r

15
10

ec 5

~
AI
!Jc 0 -o- Ti
-0- D
'E -5
Q"

-10
-15
-20
0.5 1.5 2 2.5 3 3.5
Sn (at. %)
I
(c) 120°, rotational boundary

6~
AI, Ti (at. %)
35 40 45 50 55 60
20

15

10

e......
c
5

0
!l
c
'ö -5
~

-10

-15

-20
0.8 1.2 1.4 1.6 1.8 2 2.2 2.4
n (at. %)
Figure 93. TEM sub-nanometric elemental analysis (Le., point analysis), ((ourtesy of Jeol (UK) Ltd/Dr H. Inui and Prof. M. Yamaguchi, Kyoto
University, Japan.)
 u ace mlcroscopy

Table 11. Information obtained from various analytical techniques

Analysis Morphological Elemental Element Crystal Bonding

instrument observation analysis mapping structure state

STM/AFM
AES *
EPMA * * *
nlR * * *
SEM * *
SIMS *
TEM *
XPS * *
* *
AES: Auger electron spectroscopy; EPMA: electron probe micro-analysis; FTIR: Fourier transform infra-red absorption spectroscopy; SEM:
scanning electron microscope; SIMS: secondary ion mass spectroscopy; TEM: transmission electron microscope; XPS: X-ray photoelectron
spectroscopy.
Courtesy of Jeol (UK) Ltd.

instrument and just some of its potential applica- • specimens having no conductivity require an AFM
tions. Today, with the diverse range of scanning used in either contact or AC mode, depending on
probe microscopes and observation methods that the sample's surface roughness and its softness;
exist, they can offer a wide range of analysis tech- • specimens with conductivity - both the AFM and
niques, as indicated in Table 1l. STM may be utilised.
Recently, scanning probe microscopes have been Any scanning probe microscope allows specimen
utilised across a diverse variety of disciplines, in- observation in various environments, from normal
cluding fundamental surface science and routine atmospheric conditions, to liquids, also with an
surface roughness analysis, together with three- ultra-high vacuum. Conversely, some electron
dimensional imaging. The scanning probe micro- microscopic applications necessitate situating the
scope is an imaging tool with a large dynamic range, specimen in a vacuum and in certain instances the
encompassing the realms of optical and electron sampies need to be either stained or coated with a
microscope applications. These instruments can be conductive film. Certain types of specimens in their
employed as a three-dimensional profiling instru- original form may be destroyed by such sampie
ment - with superb image resolution abilities - that
in certain cases can quantify physical properties
such as surface conductivity, static charge distribu-
tion, localised friction, magnetic fields and elastic
moduli. The instrumental applications for scanning
probe microscopes can be shown on achart relating
to their lateral and vertical resolutions, as previously
indicated in Figure 80.

3.4.1 Criteria tor using scanning

probe microscopes
As can be seen in Table 11, the AFM can be utilised
for morphological analysis, with the visual display
having high-resolution capability for extremely Hat
sampies, whose topographical features cannot be
readily distinguished even with an SEM. The AFM
(see Figure 94) is particulady applicable when

• topographical features lie within an observation

area of <1 fLm 2 ;
• non-conductive specimens use either an STM, or
more notably an AFM - in accordance with the Figure 94. Scanning probe/atomic force microscope. [Courtesy of
sample's conductivity; Jeol (UK) Ltd.]
 Tip and sampIe \
interaction for
anSTM 1= GVe- kd

Sensing means for

vertical position of tip

Coarse
positioning
system
bringing
tip into
sampIe
vicinity
Feedback system
controlling vertical
tip s position
SampIe
-==:::l==
Piezoelectric scanner moving
sampIe under tip, or tip over sampIe,
in a raster pattern

~c=J
Computer system to
drive scanner, for
L----=.J
measured data converting
it into an image

Figure 95. Schematic representation ofthe generalised operation of a scanning probe microscopes (Le., STM/AFM). [(ourtesy of Nikon/Park
Scientific Instruments.]

treatments. If observation occurs in an atmosphere, important surface-related factors, but care is needed
gas or liquid, the scanning probe microscope can be when quantitative measurements are required.
effectively used, but careful interpretation of images The scanning tunnelling microscope (STM) can
is necessary, due to the potential of adsorbed impu- be considered to be the forerunner of all derivative
rities. These AFM/STM microscopes can be used as types of scanning probe microscopes. It was
an effective means for measuring specific physical invented by Gerd Binning and Heinrich Rohrer at
properties, such as surface magnetism, friction IBM (Zurich) in 1981, who later received a Nobel
force, viscosity, elasticity, surface potential and other prize (Physics, 1986) for this invention. This STM
instrument was the first of its type to generate real- 3.4.2 Atomic force microscope:
space surface images at atomic resolution. STMs use
a sharp conducting tip (see Figure 95, detail), with a operating principle
bias voltage being applied between the tip and
sampie. When the tip approaches the sampie to The AFM's tip will probe the sampie surface with a
within 10 A ("d") electrons from the sampie will sharp tip having dimensions of several micrometres
"tunnel" through this gap into the tip or vice versa, ~ong and approximately 100 A in diameter. The tip
depending on the bias voltage sign. The resulting IS located at the free end of a cantilever which is

tunnelling current is variable, this being related to between 100 and 200 j-Lm long (see Figure 96).
the tip-to-sample spacing which, in turn, is used in Applied forces between the tip and the sample's
the creation of the STM image. In the case of the surface cause the cantilever to either bend or deftect.
STM, both the tip and the sampie must ideally be Cantilever deftections are measured as tip scanning
conductors, or at least semiconductors, which is not occurs over the sampie or, conversely, with the
the case for an AFM that can image insulating mate- sampie being scanned under the tip. These deftec-
rials. tions of the cantilever allow the production of a
From Figure 95 it can be seen that the tunnelling computer-generated surface topography map.
current is an exponential function of the tip-to- The AFM cantilever can be deftected by the con-
sampie distance relationship. If this separation tribution from several microscopic surface-related
distance changes by 10% (i.e., equating to 1 A), then forces. The principal one of these minute forces is the
the tunnelling current will change by an order of inter-atomic force termed van der Waals force. The
magnitude. Such an exponential dependence van der Waals theory states that atomic attraction
enables an STM to have a notable sensitivity, increases as atoms are progressively brought together,
allowing surface imaging to sub-Angström vertical until their electron clouds begin to electrostatically
precision, with lateral atomic resolution. STMs are repel each other. However, the electrostatic repulsion
designed to surface scan in either of two mo des: progressively weakens this attractive force, as the
inter-atomic separation continuously decreases.
constant height mode - where the tip's travel is in Eventually the van der Waals force reaches zero,
a plane horizontal to and above that of sampie when the separation distance between the atoms
and variation of the tunnelling current is related reaches several Angströms - which is approximately
to the local surface topography and the electronic the length of a chemical bond. When the atoms are in
properties of the sampie. This tunnelling current contact, then the total van der Waals force becomes
is measured at each sampie surface feature, which positive (i.e., repulsive). As a result of this repulsive
constitutes a data set for the topographic image; van der Waals force, any attempt to force the atoms
2 constant current mode - the instrument uses into a doser atomic bond are negated.
feedback to maintain the tunnelling current, by In practice for the AFM, this results in the
constantly adjusting the height of the tip at each cantilever pushing the probe's tip against the
measured topographical feature. If the system sampie, and under this condition the cantilever
detects an increase in tunnelling current, the bends rather than forcing the tip and sampie atoms
voltage is adjusted to the piezoelectric scanner to doser.
increase the tip-to-sample distance. The potential energy that acts between two
neutral atoms can be expressed from the formula
In constant current mode the scanner's motion derived by J.E. Lennard-Jones, as follows:
constitutes the topographic data set, such that the
system maintains a constant tunnelling current to
within several per cent. This control equates to a tip-
to-sample distance constancy of less than several
hundredths of an Angström. where, by calculating the potential energy of say, Xe
The main consideration in the following text is atoms, using e = 0.02 eV, a = 0.4 nm; it is possible
principally reserved for a discussion related to the to derive a "distance-potential energy graph" (not
AFM type of scanning probe microscope, its design, shown). Moreover, using the relationship for the
operating principle and potential surface-related total potential energy of two adjacent atoms this can
applications. be obtained and the force acting on these two atoms
may be found, after differentiation, to be in the
region of 0.2 nN.
Additionally to the repulsive van der Waals force,
two other forces often occur during the AFM -contact
operation:
InClustrial metrology , .

(a) SEM image of an atomic force

microscope (AFM) probe, indicating
the depth of field of an SEM (specimen
tilt 0°).

(b) SEM image ofAFM probe

geometry
(specimen tilt 60°).

(c) Indentations in soft material and profile height distribution, as indicated.

N .--'- ur

... ..,. ..... ~".I...,.-

- Figure 96. Atomic force microscope probe geometry. [(ourtesy of Jeol (UK) Ltd.]
 cantilever force - exertion applied by this cantilever bends as contact with the surface occurs,
mechanical device; resulting in the laser beam's position shifting in the
2 capillary force - exerted by a thin layer more often detector. The PSPD is capable of discriminating light
inspected in an ambient environment. measurement displacement down to lOA, with the
ratio of optical path length between the cantilever
The cantilever force tends to act in a compressive and its detector to that of the actual cantilever giving
spring-type manner, with the magnitude and its a mechanical amplification. This further mechanical
associated sign being either attractive or repulsive amplification enables the overall system to detect
in nature, depending upon the cantilever deflection sub-Angström vertical motion of the cantilever tip.
and its spring constant. The spring constant k can be There are other methods of detecting cantilever
found by the following relationship: deflection, such as optical interference, or simply by
manufacturing the complete cantilever arm from a
Bad 3 piezoresistive material and detecting its deflection
k=4[3 electrically - due to the fact that any mechanical
deformation causes changes in the arm's material
where B = Young's modulus for cantilever material, resistivity. Once the AFM instrument's electronics
a = cantilever width, d = cantilever thickness, I = have detected the deflection of the cantilever, it can
cantilever length. compute and then generate a topographical data set,
To obtain the atomic resolution with an AFM, the by operating in two distinct modes - as previously
spring constant needs to be 0.01 nm and, to detect mentioned, these are the constant height and
10- 9 to 10- 12 N between the atoms the constant constant force mo des.
needs to be

1XlO- 12 N
k = 1X 10- 9 - ---:-c--
11
1 X 10- m
3.4.3 Atomic force/scanning probe
microscope: applications
Therefore
The diagrams schematically illustrated in Figure 97
k = 0.1 - 10 [n/m] are just some of the typical applications for either
an AFM or some form of STM, most notably:
The capillary force arises when water "wicks" its
way around the tip, enveloping it and applying a • Static AFM (contact AFM, Figure 97i) - this mode
strong force attraction (10- 8 N) holding the tip in obtains sampie surface structure by scanning the
contact with the sample's surface, with the magni- surface, while keeping a constant repulsive force
tude of the capillary force being dependent upon the acting between the tip and sampie. The tip is
separation distance of the tip-to-sample. So long as normally in contact with the sampie during this
the probe's tip maintains contact with the sampie mode of imaging. Historically, the contact mode
surface (in contact mode of operation), then the was the first probe microscope mode of operation
capillary force would normally be constant due to that enabled the imaging of non-conductive
the distance between tip and surface being virtually sampies.
incompressible. An assumption is made that the • STM (scanning tunnelling microscope, Figure 97ii)
retained water layer is relatively homogeneous, with - this mode of operation applies a bias voltage
the variable force in the contact mode being the between a sharp conductive probe and sampie
force exerted by the cantilever. Hence, the total force exhibiting conductivity. When the probe and
exerted by the probe's tip on the surface of the sampie are in elose proximity, some of the elec-
sampie, being the sum of the capillary and cantilever trons can "tunnel" between the tip and sampie
forces, must be balanced by the repulsive van der without ohmic contact. Scanning the surface
Waals force for the contact AFM. The total force while using a feedback loop to maintain a con-
magnitude exerted on the sampie will vary around stant tunnelling current reveals the structure of
10- 8 N - with the cantilever being repulsed as hard the sampie surface.
as the water provides attraction to the tip - to the • LFM (lateral force microscope, Figure 97iii) - this
general operating range of typically 10- 6 to 10- 7 N. technique is sometimes termed friction force
The cantilever arm's measurement positioning microscopy. It measures the friction between the
must be accurately known and optical techniques tip and sampie by measuring the torsion al
are the most common form of sensing media. bending of the cantilever. LFM is both a static and
Typically, the cantilever bounces back a laser beam contact mode technique which simultaneously
onto a position-sensitive photodiode (PSPD). The acquires an AFM topography image.
• Force modulation/phase detection (Figure 97iv) - signal. In the case of "discrete contact" mode of
the technique adds a modulation signal in the operation, qualitative visco-elasticity images of a
Z-direction, while observing the sampie with sampie can be obtained by measuring the phase
static AFM. Qualitative elasticity of the sampie is lead or lag of the cantilever relative to the modu-
obtained by measuring the displacement of the lation signal.
cantilever with respect to the modulation signal. • Dynamic AFM (non-contact and discrete contact
The "phase detection" technique measures the AFM, Figure 97v) - dynamic AFM is accom-
phase lag of the cantilever relative to the driving plished by oscillating the cantilever near its reso-

(i) Static AFM (i.e., contact AFM). (v) Dynamic AFM (i.e., non-contact, or
discrete-contact AFM).
Photo Laser beam

(vi) STS (i.e., scanning tunnelIing spectroscopy), or CITS

(i.e. current imaging tunnelIing spectroscopy).

(ii) STM (i.e., scanning tunnelling microscope). Principle I

diagram
ofCITS

128

(Hi) LTF (i.e., lateral force microscope). (vii) LTF (i.e., lateral modulation).

Friction min.

(iv) Force modulation/phase detection. (viii) MFM (i.e. magnetic force microscope).

Figure 97. Atomic force/scanning probe microscope applications. [(ourtesy of Jeol (UK) Ltd.]
 u ace mlcroscopy

nant frequency and using forces at or near the method, this technique removes most of the
sample surface to damp the motion of the "artefacts" associated with the scanning direction
cantilever. Non-contact mode utilises attractive and sample morphology.
van der Waals forces to damp the cantilever's • MFM (magnetic force microscope, Figure 97viii) -
motion, allowing imaging to be undertaken this is a variation on the NC-AFM that uses a can-
without the tip contacting the sample surface. tilever coated with a magnetic material. The mag-
Discrete mode uses tip-surface contact to damp netic coating makes the tip sensitive to magnetic
the motion of the cantilever. Discrete contact domains on the sample surface. Further, the mag-
mode of operation is sometimes referred to as netic domain distribution can be simultaneously
"tapping". Additionally, it is possible to use an obtained with a topographical image.
"FM detection" method .
• STS (scanning tunnelling spectroscopy) and CITS Figure 98 illustrates an AFM image resulting from
(current imaging tunnelling spectroscopy, Figure the face milling of an austenitic (316-grade) stain-
97vi) - STS is a measurement of current versus less steel specimen being machined in the ultra-
voltage (1- V curves). These 1-V curves can pro- high speed region - 3000 rn/rn in. The image area is
vide an insight into the characteristics of a sample; 80 /-Lm 2 , which shows the passage of the milling
typically these 1- V curves are taken at a point. inserts and the periodicity of the feed motion at a
However, the CITS technique acquires an 1- V rate of approximately 30 /-Lm/cutter revolution. Such
curve at each point in the topographical image. three-dimensional images can impart valuable
• Lateral modulation LFM (Figure, 97vii) - in this information on the surface topography, the shearing
technique a lateral modulation signal is applied mechanism and the influence of cutting geometry
in the scanning direction while taking an LFM across the surface, which might otherwise remain
image. Compared to the non-modulated LFM unseen.

Figure 98. An atomic force microscope (AFM) 3-D image of an ultra-high-speed face-milled austenitic stainless steel surface. [(ourtesy of
Jeol (UK) Ltd.l
3.4.4 Ultrasonic force microscope: tion of the UFM to local elastieity (i.e., finer than 4
nm), with sensitivity differences in the elastic
developments moduli to better than 0.5%.
UFMs have the ability to detect cracks and delam-
The relatively re cent development of an ultrasonic inations, based on the prineiple that cracks interrupt
force microscope (UFM) amalgamates acoustic the propagation of acoustic waves - a widely used
microscopy, with its sensitivity to elastic properties, technique in ultrasonic non-destructive testing
with atomic force microscopy, having nanoscale (NDT) and acoustic microscopy. The UFM has the
resolution imaging for materials. In the UFM instru- capability of inspecting either surface or subsurface
mental technique, a high-frequency ultrasonic cracks to nanoscale resolution. Although the UFM
vibration with a frequency of several megahertz is instrument offers considerable scope in assessment
applied to a sampie that forces it to elastically of surface and subsurface conditions, its physics
"indent" itself against a static AFM tip. This tip is have yet to be fully understood. By way of example,
positioned at the end of a force-sensitive cantilever, the thin layers of water that cover virtually every
which has very high stiffness at megahertz frequen- surface in an ambient environment act at ultrasonic
eies (see Figure 99a). The contact between the AFM frequeneies as a tough sticky film, modifying the
tip and the sampie, being only a few nanometres in UFM signal. Fortunately for the aspect of UFM
size, serves as a mechanical diode, detecting the imaging, the images can be acquired at different
amplitude-modulated ultrasonic vibrations in the "normal" forces and this enables differentiation of
kilohertz frequency range. The mann er in which the the elastic and adhesive contributions, resulting
system operates is analogous to that of a crystal from the influence of minute levels of water. This
radio (see Figure 99b). The UFM instrumental tech- approach to discrimination of contributing effects
nique takes advantage of the intrinsic properties of suggests a way towards quantitative measurements
the AFM cantilever, being of high stiffness at the of nanoscale elastic and surface adhesive properties,
indentation frequency (i.e., 10 2 to 104 times greater while another manner of controlling the surface
stiffness than at the lower modulation frequency) adhesion problem is to use the UFM in a liquid envi-
and has significantly better force sensitivity at this ronment. This development of an "underwater
modulation frequency. UFM" not only reduces potential damage to delicate
Owing to the advantages found when using the material, but also enables a direct study of nanoscale
UFM, the elastic properties of many materials can elastic properties of materials such as biopolymers
be readily determined and this indicates the remark- and medical materials in their natural environment.
able co nt rast between them, from soft polymers to One of the latest discoveries is that ultrasonic
very hard ceramics. For example, the AFM topo- waves can be direcdy propagated from the base of
graphic image shown in Figure 100(a) indicates an the minute AFM cantilever, travelling down to its
interface on a surface as a jagged trench that divides end, in a similar manner to light propagating along
a silicon carbide fibre (i.e., positions land 2) and a an optical fibre. These ultrasonic waves can vibrate
mullite matrix (Al z0 3 -SiO x ) at position 4. By util- the AFM tip at megahertz and even sub-gigahertz
ising the "elastieity image" in Figure 100(b), much frequeneies. The cantilever simultaneously acts as a
greater definition - and hence detail - of the inter- stiff and high-quality conductor of high-frequency
face structure can be seen. In Figure lOO(b) a rela- ultrasonic vibrations, while also acting as a gentle,
tively soft intermediate concentric carbon-rich layer sensitive detector of non-linear ultrasonic force at
(position 2) and a softer reaction layer (position 4) the sampie frequency of several kilohertz. This dual
separate the higher-stiffness regions of both the Sie ability is of significant benefit, enabling bulk mate-
and mullite. Other images (Figure 100c and d) are rial sampies of materials to be studied by the UFM,
also depicted, comparing topographical images that as access is only necessary from one side of the
were simultaneously acquired to the FMM image sampie.
(different region; Figure 100c) with an FMM image An intriguing further development in the UFM is
of the area shown in "c" (Figure 100d). When applied the combination of ultrasonic vibrations from both
to semiconductor low-dimensional structures such the sampie and cantilever. When such vibrations
as that found in quantum wells and dots, this mate- occur at adjacent frequeneies, the AFM tip detects
rial-dependent elastic contrast enables one to differ- the oscillating force at the difference frequency, in a
entiate between areas of disparate material similar manner to that of a heterodyne radio
compositions - for example, image the percentage of receiver, prompting the name for this new instru-
Ge in the SiXGe 1_Xcompound - together with high- ment: heterodyne force microscope (HFM). Just as
lighting the growth defects, or to detect residual modern heterodyne radios are far superior to crystal
polishing damage. Defect-free areas such as nanos- detectors, the HFM has superior sensitivity to small
tructures can be employed to test the lateral resolu- vibrations than that of a UFM. Even more significant
 u ace mlcroscopy

(a)

Additional pie:wtransducer allow thc launch of

an ultrasonic wave down the canti lever in
hetrodyne force microscope (HFM)

Microfabricalcd canlilever

~
La er

Position-sensitive photodeteclor monitors

average force on the tip

Studied sam pie is attached to the

Piezotransducer-vibrated sam pie is attacbed to the piezotransducer and
piezotran ducer and placed on
placed on thc AFM scanning stage
the AFM sc,a nning stage

(b)

\ j_
Insta nt force acting on thc tip, " recUfied" by the tip-surface
F , force acting belween non-linear contact
AFM tip and surface
r.= ""«Iod 'y'fM

Ö, tip-surface distance

_ _ _ _ Modulation of tip-surface dislancc at

ultrasonic frcquency

Figure 99. (a) Design of the ultrasonic force microscope. (b) Principle of non-linear detection of ultrasonic vibration in UFM. [(ourtesy of Dr
O. Kolosov, Isis Innovation and University of Oxford.]
 InClustrial metrology , .

o 4 8 12 16 jJ.m 0 4 8 12

Figure 100. Topography (al and UFM (b), topography and FMM (dl images of a sam pie made of coated SiC fibre embedded in a mullite matrix
(A 120l -Si0 2), image width 10 f.lm. [Courtesy of Dr 0, Kolosov, Isis Innovation and University of Oxford.]

is the fact that the HFM picks up information about

the phase of high-frequency vibrations, which are
3.5 X-ray photoelectron
directly related to the time-dependent processes of spectroscopy (XPS)
tip-surface force interaction. This will enable obser-
vations of the mechanical structure of materials
with both nanometre lateral resolution and nano- For most technologies today, crucially important
second time sensitivity. information is required on surfaces and associated
chemical composition. This surface detail might
include corrosion, lubrication, adhesion, micro-elec-
tronics features, catalysis and surface coatings. For
 u ace mlcroscopy

(a) Photoelectron spectrum of lead superimposed (b) Sigma Probe is an XPS instrument and is optimised for
on a schematic diagram of the electronic analysis of small areas.
structure of the element
4f

4p 4p4s '

Binding
cncrgy cncrgy

(c) The important features of a typical XPS instrument.

CCDcamera

Analyscr

Figure 101. The important elements of an X-ray photoelectron spectroscopic (XPS) instrument. [(ourtesy of VG Scientific.]

example, even factors such as the sub-monolayers of (for chemical analysis) and has the following char-
material can affect bonding properties. Hence, acteristics:
gaining an understanding of these surfaces requires
sophisticated and powerful analysis techniques, • surface sensitivity - typically, the information
typified by that of X-ray photoelectron spectroscopy depth is less than 10 nm; by comparison, an EDX
(XPS). information depth might be several micrometres;
The XPS instrumental technique and associated • elemental detection - all elements with the excep-
equipment are shown in Figure 101; the technique is tion of Hand He can be detected;
sometimes referred to as "electron spectroscopy" • chemical state identification - chemical shifts in
 InClustrial metrology , .

(a) Spectrum of barium acquired in 10 seconds using a (b) An overlay of two silicon images.
monochromated X-ray source

Bn 3d

2.0
-( I)
CI.
u
S OKLL
C 1.0
.... Ba N

.-
~
s Ba ~p
BIl4d

Ba ~ s I
Cl

0~10
~0~
0---8~0~
0---6~0~
0---4
~O=
O--~~~~0
Binding Enel'gy (e V)

_60 "iurfnce Bulk

"::!e ))\

--
=
----. -
(c) ARXPS measurements from a GaAs
surface, showing the presence of an
oxidised region at the surface. Right: a
montage of
As 3-D spectra taken at aseries of angles,
showing the change in relative intensity of
the As in the oxide and GaAs states

20 40 60 80
Angle (Degrees)

(d) Images from a 5 fLm field of view using

2.7 nA ofbeam current at 5 eY. The spatial
resolution in these images is 100 nm. (Top
left: SEM; top right: cop per; bottom left;
titanium; bottom right: copper/titanium
overlay)

Figure 102. Just some of the X-ray photoelectron spectroscopic (XPS) graphs and images available from quantitative surface assessment.
[(ourtesy of VG Scientific.]
 the spectrum indicate the chemical state of the surface region. For many technological disciplines,
element; it is essential to be able to analyse very small surface
• quantitative analysis - the sensitivity of XPS to features. These surface features may be such prob-
an element is independent of the matrix in which lems as conductive tracks on semiconductor mate-
it is found. Sensitivity factors can be used to rials, areas of impurity on ceramics, or likewise for
quantify the spectra produced. polymers impurities. Apart from taking spectra
from small areas, some XPS instruments produce
The XPS method relies upon the photoelectric chemical images or maps from the surface of the
effect, where the material to be analysed absorbs the specimen. The illustration shown in Figure 102(b)
X-rays and electrons are rejected; these are then depicts an overlay of two XPS images from a semi-
detected and their kinetic energy analysed. The conductor device. Both images use the Si 2p peak,
kinetic energy of a photoelectron depends upon but one (red) is of elemental silicon, while the other
(green) is from silicon in the form of its oxide. These
1 the element from which it was ejected; are representations of chemical state images and
2 the orbital in the atom from which the electron have a spatial resolution of approximately 1 fLm.
originated; X-rays generate photoelectrons from deep within
3 the chemical state of that element. the matrix, with those in the top several nano-
metres being able to escape from the solid without
As an example of this XPS spectrum, that of lead undergoing inelastic collisions. If the electrons are
is shown in Figure 101(a), superimposed onto a collected along the surface normal, then consider-
diagram showing its electronic structure. This able information can come from this region, near the
diagram indicates where the electrons that form surface. However, if electrons are collected at a
each of the peaks originate. grazing emission, then the surface sensitivity is
All XPS instruments have three major compo- further increased. The escape depth varies with the
nents: the X-ray source, the energy analyser and the sine of the take-off angle, relative to the surface.
detector (see Figure 1OIc). The X-ray source can be In angle-resolved XPS (ARXPS), this phenomenon is
either monochromated, or non-monochromated. It exploited to study the variation of composition of
is accepted that monochromators provide an the specimen with depth. An ARXPS assessment
improved energy resolution over their non-mono- is usually performed by setting the angular accept-
chromated counterparts, providing a focused spot ance of the spectrometer to a small value (typically
size of the X-ray beam to smaller than 15 fLm - 1-10°) and measuring the XPS spectra at a range
which can be the case. The energy analyser is usually of sampie angles. Figure 102(c) shows ARXPS data
of the spherical sector type. A lens is normally fitted from GaAs, which has a thin oxide layer at its
to the energy analyser to maximise the number of surface. Measurements were made of the Ga, As and
photoelectrons entering the analyser and to define o peaks as a function of angle and for the Ga and
the area from which the electrons are collected. The As both the elemental and oxide peaks were meas-
detector is usually an array of channel electron ured. The data observed at small angles is dominated
multipliers or a channel plate. The complete instru- by the surface composition, while the data at larger
ment must be housed in an uhra-high-vacuum angles has a greater inftuence from the bulk mate-
(UHV) system, so that surfaces remain uncontami- rial. The decrease in the oxygen and oxide peaks will
nated and the photoelectrons have a sufficiently therefore indicate the presence of a thin surface
large mean free path to reach the detector. A typical oxide.
variant of this type of instrument is shown in Figure Many more analytical activities can be under-
101(b), where it has a spatial resolution that is less taken with an XPS instrument, such as surface
than 15 fLm. topography (Figure 102d), depth profiles or sputter
An XPS experimental procedure would usually profiling, to identify impurities at interfaces and so
commence with the acquisition of a survey spec- on, as it is a quantitative chemically specific analysis
trum, to identify the elements present in the spec- instrument for both conducting and non-conduct-
imen. This acquisition tends to be rapidly taken ing surfaces. A wide range of XPS-based instruments
over a wide range of bin ding energies at low resolu- are available to suit specific and general analysis
tion (see Figure 102a). After acquiring the survey needs, often being the principal surface analytical
spectrum, aseries of high-resolution spectra are instrumentation purchased for surface science labo-
normally acquired for each element of interest, ratory and industrial studies.
providing quantitative information about the
 Uchida, H. et al. Single-atom manipulation on the Si (111) 7 x 7
References surface by scanning tunneling microscopy (STM). Surface
Science 287/288,1993,1056-1061.
Warren, P.D. et al. Characterisation of surface damage via contact
Journal and conference papers probes. Nanotechnology 7, 1996,288-294.
Wolstenholme, J. XPS reveals all about the state of the surface.
Dinelli, F. et al. Elastic mapping of heterogeneous nano-structures Materials World 7(7), 1999,412-414.
with ultrasonic force microscope. Proceedings of Surface and Yoshimura, M. et al. Low-coverage, low-temperature phase of Al
Interface Analysis, Basel, Switzerland, 1999. overlays on the Si (111) (-7 ( 7 structure by scanning tunneling
Griffith, J.E. et al. Characterization of scanning probe microscope microscopy. Physical Review B47, 1993, 13930-13932.
tips for linewidth measurement.Journal ofVacuum Science and Yuan, C.W. et al. Low temperature magnetic force microscope uti-
Technology B9 (6), 1991,3586-3589. lizing a piezoresistive cantilever. Applied Physics Letters 65,
Hill-King, M. Measurement in the detail. Quality Today March 1994,1308-1310.
1999,22-26.
Hibino, H. et al. High-temperature scanning-tunneling-micro-
scopy observation of phase transitions and reconstruction of a
vicinal Si (111) surface. Physical Review B47, 1993, 13027-13030. Books, booklets and guides
Hutter, J.1. and Bechhoefer, J. Calibration of atomic force micro-
scope tips. Reviews of Scientific Instruments 64, 1993, 1868- Barrett, C. and Massaiski, T.B. Structure of Metals (3rd edn).
1873. Pergamon Press, Oxford, 1980.
Ichimiya, T. et al. Formation of (21 ( (2l structure by gold deposi- Briggs, D. and Seah, M.P. Practical Surface Analysis, Vol. 1. Wiley,
tion on Si (111) (3 ( (3 Ag surface and the wavering behavior. NewYork.
Surface Review and Letters 1, 1994. Cullity, B.D. Elements of X-ray Diffraction. Addison-Wesley,
Kolosov, O. and Yamanaka, K. Nonlinear detection of ultrasonic Reading, MA.
vibrations in an atomic force microscope. Japanese Journal of Gonzales, R.C. and Woods, R.E. Digital Image Processing. Addison-
Applied Physics 32(Part 2, No. 8a), 1993, Ll095-Ll098. Wesley, Reading, MA, 1992.
Kolosov, O. Imaging the elastic nanostructure of Ge islands by Hull, D. Introduction to Dislocations (2nd edn). Pergamon Press,
ultrasonic force microscopy. Physical Review Letters 81 (5), 1998, Oxford,1979.
1046-1049. Jeol (UK) Ltd. Invitation to the SEM World.
Kolosov, O. UFM shakes out the details at the nanoscopic scale. Howland, R. and Benatar, 1. A Practical Guide to Scanning Probe
Materials World December 1998,753-754. Microscopy. Nikon Corp.!Park Scientific Instruments, 1996.
Kuk, Y. and Silverman, P.J. Scanning tunneling microscope instru- Kuzmany, H. Solid-State Spectroscopy. Springer, Berlin, 1998.
mentation. Review of Scientific Instrumentation 60(2), 1989, Vickerman, J.c. Surface Analysis: The Principal Techniques. Wiley,
165-181. New York, 2000.
Markoff, J. A novel microscope probes the ultra small. New York Walls, J.M. Methods of Surface Analysis. Cambridge University
Times 23 February 1993, Ci, C8. Press.
Nechay, B.A. et al. Applications of an atomic force microscope Watts, J.F. An Introduction to Surface Analysis by Electron
voltage probe with ultrafast time resolution.]ournal of Vacuum Spectroscopy. Oxford Science.
Science and Technology May/June 1995,1369-1374. Wiesendanger, R. Scanning Probe Microscopy. Springer, Berlin,
Sato, T. et al. Dynamic observation of Ag deposition process on Si 1998.
(111) surface by high-temperature scanning tunneling micro- Yates Jr, J.T.ExperimentalInnovations in Surface Science. Springer,
scopy.Japanese Journal of Applied Physics 32,1993,2923-2928. Berlin, 1997.
Schöfer, J. and Santer, E. Quantitative wear analysis using atomic Desjonqueres, M.-C. and Spanjaard, D. Concepts in Surface Physics.
force microscopy. Wear 222(2),1998,74-83. Springer, Berlin, 1996.
Stahl, U. et al. Atomic force microscope using piezoresistive can-
tilevers with a scanning electron microscope. Applied Physics
Letters 65, 1994,2878-2880.
 Roundness and
cylindricity

"Round and round the rugged rock

The ragged rascal ran."
(Nursery rhyme)

135

G. T. Smith, Industrial Metrology

© Springer-Verlag London 2002
 isties of the part will have a significant effect on the
Roundness - setting the scene in -service performance of the component in any pre-
In the past, the simplistic technique for the assess- cis ion work, etc, where either the maleIfemale
ment of roundness was measuring two or three fitment is vital and/or its rotational assembly must be
diameters on a component to determine diametral strietly controlled. The "pressure tightness" in com-
variations, with symmetrical part features being ponent sealing applications will also be heavily inftu-
assessed for variations in radius about an axis of enced by the manufactured accuracy and precision
rotation. These simple roundness measurement of assemblies, due to departures from roundness
techniques give information about geometric and factors. Ball and roller bearings must be round and
harmonie errors ab out an axis of rotation, therefore produced to tight tolerances and have good surface
could only be employed as a very superficial guide texture. These bearings would otherwise potentially
to a part's potential in-service performance. If there have a reduced bearing life and reliability. Further-
exists from within a selected cross-section a point more, such bearings will have increased noise and
from whieh all subsequent points on the periphery vibration tendencies, while providing some variation
of a component are equidistant, then geometrically in torque-related assembles. To give a practical
at this plane - position where measurement occurs example of just one type of potential operational
- the component can be said to be round. In reality, condition where roundness is important, the follow-
the radius of, say, a nominally round component ing is worth consideration. In a rotating ball bearing
tends to deviate - from a true cirele - around the application, if aminute roundness error occurred in
periphery of the part, making these variations the balls of not greater than 0.25 /-Lm, this error
subject to subjective interpretation of results. would result in an increase in operational noise by of
Although it should be emphasised that variations in approximately 15 dBA, which greatly increases the
a component's radius are not always detrimental to sound level that would otherwise be caused by
its potential performance, certain parts whieh can either track wobble or lack of shoulder squareness.
exhibit periodic radius ftuctuation may operate as Therefore it can be seen that a part's roundness can
though they have a constant diameter - that is may have far-reaching effects weIl beyond that of the
be lobed in nature (see Figures 107 and 109). There simple expedient of its precision manufacture.
might be a whole host of reasons for a component
to vary in its radius for astated cross-sectional
plane, some of which might be:
4.1 Introduction
• the production process inducing rotational im-
perfections - from the machine/tool/workpiece
system; Many devices and instruments critically depend
• the release of strain, or that induced in the mate- upon rotation, coupled in many cases to controlled
rial - the former case may be the result of linear motion. Therefore, an automobile and
releasing the part from elamping press ure, while machine tool could not operate without such kine-
the latter may be the result of plastie deformation matics, nor would the smallest watch or the largest
that prornotes localised surface residual hoop power station. Many factories engaged in engi-
stresses; neering activities such as precision manufacture
• induced radial vibration - potentially resulting have equipment that produces components: bear-
from tool displacement; ings, shafts, bushes - the list continues - but all have
• circumferential surface texture - residual effect one significant feature: they have radial symmetry
resulting from the production process. and are round. The question is: "How round?"
Departure from roundness assessment, or a more
Roundness is a condition of a surface of revolu- convenient term, roundness, has been historieally
tion, which can take the form of a cylinder, cone or practised by inspectors or by craft-based engineers.
sphere, where all the peripheral data points intersect. Practitioners without access to sophisticated metro-
The question that can be asked is: "Why is it so logy equipment assess the roundness of a compo-
important to be able to establish elose control over nent by carefully rotating it in a vee-block, or
departures from ideal roundness?" Firstly, a circular between bench centres, then measuring the run-
cross-section is the most fundamental shape utilised out with a dial gauge (Figure 103). This technique,
by industry and is generated or formed by the man- \ (BS 3730: Part 3) although useful for elementary
ufacturing process. Secondly, due to the symmetry of roundness inspection, has its limitations and may
a round component, it may exhibit uniform me chan - result in misleading results - somewhat negating the
ieal strength in all directions that are symmetrieal to simplicity of the operation, but this topic will be
the axis. The dimensional and geometrie character- discussed further shortly.
 (a) (b)

Dialgauge

Maximum
measured
difference

_____ J
Vertical
movement
of
workpiece

(e) In-phase (d) Out-of-phase

Figure 103. Assessment of roundness in a Vee-block. ((ourtesy ofTaylor Hobson.)

 With industrial demands being firmly focused on present. The questions relating to the bearing
precision manufacture, necessitating the require- fitment that might be asked are:
ment for even doser tolerances and greater accuracy • Is this fit sufficient?
in roundness measurement, this has meant that an • Will a combination of shaft and bearing continue
even greater emphasis is necessary on the control of to offer reasonable in-service performance when
geometrie form. During the last sixty years instru- heavily loaded and after many years of operation?
ments have been developed with greater levels of • Does the lubricant have ideal conditions to work
sophistication, for measuring a component's round- effectively?
ness and any associated parameters, many of which
are still in common use today. It is important that The answer to these questions must depend
modern-day engineers, designers, machine tool upon what we consider the word round to mean.
operators and inspectors have a basic understanding Superficially, the bearing and shaft fitment illustrated
of the working principles of these instruments, in Figure 104 may appear to be round when visually
together with their capabilities and limitations and assessed; it might also have a constant diameter when
how to interpret subsequent roundness results. measured with a micrometer. However, if this same
Roundness results may be used to either accept or bearing/ shaft relationship is considerably magnified,
reject round components. Moreover, results from then a different picture arises, as shown by the mag-
roundness inspection can be employed to monitor a nified inset image in Figure 104. From this image it is
range of process-related component manufacturing apparent that departures from the "true" circular
capabilities, typically the performance of a machine profile exist and that these lobes at points "1>:' will
tool, tool wear detection, being the resulting effects carry most of the load of the shaft as it runs in the
of poor operating procedures or similar in-service plain bearing. The lubricating film thickness must
production problems. ideally be kept within specific limits if the bearing
assembly is to perform as the designer intended,
although in this current situation the thickness of the
Why is roundness important? oil film at positions "B" will be considerably greater
than at points "1>:'. In a similar manner, the bearing's
A plain bearing and its mating shaft are depicted in bore may not be circular: it could be either slightly
Figure 104. Provided that both the bearing and shaft oval or lobed (having a constant diameter, but not
are round and the dearance fit is neither too tight geometrically round - see Figure 109), creating vari-
nor too loose, then the shaft will run smoothly, able lubricating effects to the complete assembly
particulady if full-film hydrodynamic lubrication is which were not as the designer intended.

Bearing

Lubricating film

Figure 104. Roundness of a bearing will affect its performance.

Further aspects of the measurement of (e) the bore is straight;
(d) the horizontal faee is square to that of the axis;
cylindrical components (e) bore is tapered, straight and eoaxial with the
outside, also that the diameters are coneentrie
The diseussion so far has only been related to that and the reeessed undereut is square to the axis.
of shaft and bearing assemblies, with roundness
only deseribed from a single eross-seetional view- In these examples (Figure 105), the eomponents
point. However, other information needs to be ean be eonsidered to have eomparatively simple
known whieh is conneeted to cireular features in geometrie features; eonversely, this is not the ease for
engineering eomponents. For example, in the a erankshaft for an auto motive applieation (Figure
eomponent errors depicted in Figure 105, it might 106). This erankshaft diagram indicates that the rela-
be neeessary to investigate whether tive geometry of bearing surfaees beeomes eritieal
to its desired operating efficieney. Inspeeting this
(a) two different diameter portions are eoneentrie type of eomponent, having aseries of eomplex and
with one another; interrelated geometrie elemental features, is very
(b) the bore of the thick-walled tube is eoaxial with mueh within the seope of roundness measuring
the outside diameter; instruments.

(a) Eccentric boss

(b) Offset hole

Eccentric undercut

(d) Angled face (e) Tapered & bowed

hole
(c) Bowed hole

Figure 105. Typical component errors of form that can be checked.

 Roundness and eylindridty

~_
I
_ Axis of ehe (WO end bearings • an elliptical shape - having a major and minor
axis of different lengths (Figure 107 "B");
• a trochoidal or three-Iobed shape - exhibiting a
difference in distance points on the periphery to
its centre (Figure 107 "C").

From the lobed shape in Figure 107 "C", if "Rl"

is the maximum distance of the periphery from the
cent re and "R2" the minimum distance, then the
roundness will be "Rl - R2". When the profile is
,.... .- - - - - A.i. 01 cr.nkpin oul 01 parallel symmetrical this value is simple to establish;
however, specifying the departures from roundness
for an irregular profile like the example shown in
Figure 108 is only possible if the "centre" of the part
can be found, from where the measurements are
determined. The question is: "where is the centre?"
Axis 01 mein bearings If an arbitrary point ''1\' is established where we es ti-
out 01 alignment
mate this point to be somewhere near the true
centre, the maximum departures from roundness
might be "Rl - R2", but if we were to choose a
different centre "B", then the maximum roundness
could be "R3 - R4". So, determining the centre from
where the variation of the profile is an important
OUI 01 round
. . . . .-
factor of roundness assessment and more will be
said on this topic later.
To measure roundness, it is often the practice to
include a rotational factor to this measurement;
conversely, diametral measurement is normally
undertaken in a static manner, perhaps utilising either
a micrometer or co ordinate measuring machine
(CMM). Although a component's roundness and dia-
meter are two distinct parameters, its roundness, or
more particularly its departure from roundness, does

c
B

Not squale (0 Clxis

Figure 106. The possible roundness-related errors in atypical crank-

shaft's geometry. ((ourtesy ofTaylor Hobson.) A A

Roundness measurement: how is it

achieved?
In general terms, apart can be said to be round in c c
a specific cross-section if there exists within the
sectional plane a point - namely its centre - from
which all other points on the periphery are equidis- B
tant. If these geometric conditions are met, then the
cross-section will be a perfect circle (see Figure 107
''1\'). If the cross-sectional geometry does not equate Figure 107. Typieal roundness geometrie shapes: A, "true" round-
to a perfect circle, then two conditions that might ness; B, elliptical shape; C. troehoidal (Le., lobed) shape. NB they all
occur are: have the same diameter when measured at specifie points.
 Causes of departures from roundness
The determination of a component's roundness
necessitates the measurement of irregularities. Such
irregularities are normally referred to as undula-
tions, peaks and valleys, and lobes - this latter term
describes regularly spaced undulations (which are
usually few in number). It is important to realise that
undulations and lobes do actually exist on nomi-
nally round components and are not simply theo-
retical concepts, but are attributable to all machining
and process-related operations.
The reasons for possible departures from ideal
roundness of mechanical parts might be due to a
number of process-related factors, such as some
poorly maintained bearings in a machine tool, or by
deflections of the workpiece resulting from the tool
experiencing interrupted cuts or differential material
thicknesses. Roundness departures mayaiso occur
due to shafts being either ground, or turned which
Figure 108. Determination of the (entre of a roundness profile is an are inadequately supported, causing deflections, or
important factor in predicting its roundness measurement. centres that are not correctly aligned.
A particularly difficult problem to avoid in many
machining processes is that of lobing, partieularly
have a practieal effect on the measurement of its diam- when the component is manufactured from either
eter - possibly making these diametral measurements round bar-stock or ring-type parts that are securely
very misleading. held in either a three- or five-jaw self-centring
chuck (for turning and grinding, respectively). The
part's periphery will be compressed at the points of
contact for individual jaws, giving rise to compres-
Differences in measured and effective size sive stresses in the material. After a perfectly circular
diameter has been bored, upon its release from
The representation of the three-lobed shape illus- the chuck a lobing effect will be seen, this lobing
trated in Figure l09(a, left), has a measured size of being related to the number of chuck jaws.
exactly 25 mm; therefore it would be expected to fit Centreless grinding operations are yet another
into an appropriately toleranced hole of <1>25 mm. source of lobing error, particularly where the orig-
Here, the term measured size refers to the diameter inal bar stock has a poor and irregular shape, whieh
as measured between a pair of parallel faces by is exacerbated by the machine's work-rest being
either a micrometer or limit gauge (gap gauge). In incorrectly positioned beneath the centreless
reality, the hole must be increased to <1>28.9 mm ground part. Tool marks on turned parts are not
before a mating fit can be established (shown in readily described as roundness features, whereas a
Figure 109a, middle). Similarly, a three-lobed hole worn tool, or one which is set incorrectly, could
(Figure l09a, right) that has a measured size of increase the tendency for surface chatter marks on
<1>25 mm will not allow a truly round <1>25 mm shaft the component, showing up in subsequent round-
to fit. In fact, the largest shaft that can be located ness measurements. These production variables and
in this lobed hole has a considerably smaller size, their associated process, tool and machine-related
namely of <l>2l.1 mm. Hence, the effective size of factors will be discussed in more depth in this and
these two mating components will be <1>28.9 mm the following chapter.
for the shaft and <1>21.1 mm for the hole. This lobed Components that are either drawn or extruded
hole and shaft geometrie relationship is still further take their final shape from the drawing/extrusion
reinforced in Figure 109(b), where similar differ- dies. Roundness checks on such parts will not only
ences are exhibited by parts having odd-numbered reveal changes of shape of the die as it wears, but can
lobes, so the shape - roundness - of a component also highlight imperfections that are the result of
can affect its size when measured in the conven- scoring along the surface, as drawing/extrusion
tional way. occurs.
 Roundness and cylindricity

(a) Geometrie relationships of lobed-shaped, holes and shafts.

Hole enlarged to 1128.9 mm

p rior to shaft, which has a
measured 1125 mm, can be
Largest shaft t hat will enter

r
25 111m inserted.
this hole is 1125.1 Olm.

1125 111m hole

LCJhcd s harc

(b) As lobing increases, holes become larger and diameters smaller.

ROUNO: 3-LOBEO: 5-LOBEO: 7-LOBEO:

ft25m m jl28.9 ß1111 1126.3 111m 1125.6mm

ElTccU\'C c Icrnal sizc

ElTccti\'c intcrnal sizc

Figure 109. Differenees between measured and effeetive sizes. (Courtesy ofTaylor Hobson.)

4.1.1 Roundness measurement: basic in a vee-block (BS3731 : 1987) and carefully rotate it
with a dial gauge or dial test indicator in contact with
approach the surface (Figure 103a, b). A truly round part will
show no movement on the dial gauge pointer;
In the previous section it was mentioned that however, if the component is not round then irregu-
departure from roundness cannot always be fuHy larities as they contact the vee-block's sides will
detected, or indeed measured by dial guage static cause the part to move up or down. These component
techniques (BS3730: Part 3). If access to sophisticated irregularities (Figure 103a and b) will dis pi ace the
roundness instrumentation is not available, then dial gauge's plunger as the part is slowly rotated; "cor-
some degree of elemental roundness measurement rection factors" should be applied. The amount by
by the rotation technique can be made. The simplest which the gauge pointer is displaced depends on
method that can be adopted is to place the part several factors:
 InClustrial metrology , .

• the height of the irregularities - peaks and

valleys;
• the angular spacing of the irregularities;
• the vee-block included angle.

The pointer movement will be greatest when the

peaks or valleys simultaneously contact the plunger
and faces of the vee (Figure 103a). The movement
will be smallest when a valley is under the plunger
and peaks are contacting the vee-block (Figure
103b) and vice versa. If the included face angle of the
jaws utilised to inspect the part differ, then the
plunger motion relative to component rotation can
cause two distinct conditions to occur:

• in-phase - with a wide vee-block face angle Figure 110. Multidimensional modular receiver gauge and concen-
(Figure 103c); tricity inspection with data capture and manipulation. [(ourtesy of
• out-of-phase - with a narrow vee-block face angle Marposs Ltd.)
(Figure 103d).
NB: See BS3730 Part 3 (Figure 103 "Summit and rider
tion of the measurement status, even from a distance,
method").
where for example:
Vee-block roundness measurement is essentially
• green illumination indicates the part is within the
a three-point technique, with several variations in
tolerance limit;
the mann er in which it can be applied, such as when
• yellow illumination indicates the part feature is at
inspecting a large external diameter or a similar
the tolerance limit;
internal bore.
• red illumination indicates measurement is out-
A simple artefact can be manufactured that in-
side the pre-set limits.
corporates the dial gauge and datum feet, its body
shape designed to suit the particular metrological
Multiple diameter assessment with such
application. Bench centres for longer parts, with
modular-constructed receiver gauge elements allows
centre-drilled ends, are a useful inspection apparatus
the part's functional performance to be assessed.
for assessing either roundness or concentricity of a
By arranging the transducer indicators in suitable
stepped-diameter shaft, in conjunction with a dial
banks (see Figure 110) a host of the component's
gauge. This technique is further refined when bench
centres are employed, as they allow all of the above geometric features can be simultaneously assessed.
roundness parameters to be assessed, together with
tapered diametrical features, although care must be
taken to prevent component gravitational sag for
long slender parts. The dial gauge previously men- 4.2 Roundness: measuring
tioned can be replaced by an electronic gauging head
and its associated meter unit, the main advantage
instruments
being that both range switching and data-Iogging
can be undertaken. These latter refinements to radial
The design of a roundness instrument (see Figure
measurement are necessary when many parts are
111 for a schematic diagram of a rotating worktable
required to be inspected and for statistical process
type) incorporates a number of extremely accurate
contro1!quality control applications. This aspect of
parts that are configured to assess specific elements
multi-gauging can be quite sophisticated, as shown
of a part's roundness. The roundness-related
in Figure 110 for a typical modular component
features are:
feature set-up of the myriad of industrial applica-
tions, where a multiple receiver gauge station can • rotation - rotary motion providing the reference
simultaneously inspect critical elements of stepped axis;
diameters and geometric features. Such electronic • rotational axis - this must be independent of the
indicators equipped with a digital display provide part being measured;
both numeric and analogue visualisation of mea- • measuring device - some form of stylus, and a
surement. A three-colour analogue display (e.g. transducer (pick-up), which converts mechani-
green, yellow and red), provides clear visual defini- cal stylus motion into a proportional electrical
 Roundness and eylindridty

signal, that is then subsequently electronically ment, as it allows measurements to be taken at

amplified; various heights on a component, to establish the
• indicator - these roundness instruments gen er- part's roundness characteristics for specific geo-
ally employ a meter in association with the elec- metrie features, such as concentricity, eccentricity
trical pick-up for part set-up. The final displayed and runout or simply roundness at a number of
profile would normally be displayed in either a heights;
polar or linear form on the display screen. The • additional feature - coarse and fine adjustments
polar plot from an integral printer in a dedicated for part positioning (centring and levelling),
roundness computer produces a permanent hard which have the advantage of reducing the set-up
copy for future reference; time (using automatie set-up procedures on some
NB One of the main reasons for not using a sophisticated instruments).
meter as an indication of roundness is that the
fluctuations of the meter's needle will not only
indicate the component error magnitude, but also
the eccentricity and ovality caused by an incor- 4.2.1 Types of instrument
reet set-up. This is also true of a polar plot, unless
a reference circle is used. Roundness-measuring instruments tend to be of
• reference datum - this is required for referenced two basic configurations as illustrated in Figure 112.
and repeatable roundness measurements; For components of small size and weight, the most
• vertical straightness column - this is not essential suitable configuration is the version based on a
but is a desirable feature on a roundness instru- rotating table, or turntable (Figure 1I2a). The work-

cl·tical straightness column

Horizontllll1djustmcnt

Rotat ionl1ll1xis of lable ~

Levclling adjustmcnl

Stylus

Componcnt
under
te t

Ce ntring adjuslmcnt

Table
rotation

Precision
lable

Base

Figure 111. Schematic diagram of the main features of just one configuration for a roundness-testing machine.
piece rotates while the pick-up remains in a set instrument's design; also long shafts, crankshafts
position. For large and notably heavier parts, the and many more large parts can be simply accom-
rotating pick-up type, which rotates the piek-up modated on the table;
(Figure 112b), is preferred. Each type of instrument 3 in order to speed up the measurement task and
has advantages and is more suitable to certain types minimise uncertainty of measurement, com-
of measurement, the choiee of instrument being puter-aided centring and levelling is available as
largely dependent on the antieipated measurements an option on the more basic versions, or a stan-
to be undertaken and the shape and size of the dard feature with the sophisticated versions. This
parts. latter feature is true for both instrumenntal types.
The first equipment available commercially,
"Talyrond I" was of the rotating pick-up type. • Turntable type (Figure 114).
The rotating piek-up design became the model
from which all other manufacturers' roundness- Operational characteristies of this instrument's
measuring instruments were developed. The success roundness configuration include the following:
of the roundness instrument's design was such that
over the years only minor modifications were neces- due to the pick-up not being directly associated
sary to enhance the equipment's capabilities. Most with the spindie, this type of the roundness
of these modifications were necessary to change the
machine is more easily adapted to other measure-
electronics from valve amplification to transistor
ments in addition to roundness, notably concen-
amplification, resulting in the "Talyrond 2" design.
tricity and alignment. Repositioning the pick-up
Later, integrated circuit electronies replaced the
transistors in the design of the "Talyrond 4". The on a component - transferring it from external
current instrument is illustrated in Figure 113; this diametral measurement to an internal surface -
instrument has the ability to accommodate an has no effect on the reference axis. Further, pick-
extremely large workpiece size and weight, featuring up positioning is simple; enabling it to reach into
automatie component cent ring and levelling - an slots or to the undersides of shoulders becomes a
important feature with parts that are very large and straight forward operation, without having to
less moveable. resort to using long, or cranked stylus arms;
2 by incorporating a straight vertical movement of
the pick-up, any straightness measurements can
Comparison of instrument types be made without modification to the spin dIe, or
its mounting;
This seetion will discuss the reasons why two types 3 the turntable weight, together with the part
of roundness instruments (Figure 112) were devel- being measured, has to be supported by its
oped, explaining design differences and, more spindie bearings, which places a restrietion on
importantly, the situations why one type is more the weight of the part requiring measurement;
suitable than the other for certain kinds of round- moreover, it also limits the amount of offset that
ness measurement: can be permitted;
4 computer-aided centring and levelling are stan-
• Rotating pick-up type (Figure 113). dard on the more sophisticated instruments. If the
centre of gravity of the workpiece falls into the tri-
Operational characteristies of this roundness instru- angular demarcation marks engraved into the top
ment's configuration include the following: of the table's surface (see Figure 114), the manual
and somewhat time-constraining centring and
the precision spindie need only carry the com- levelling activity becomes an automatie adjust-
paratively light and constant load of the piek-up ment function.
- its high accuracy is attainable without excessive
cost;
2 as the worktable is not part of the measuring
system it can be of substantial construction, 4.2.2 Spindie and bearings
so the weight of the part is not a limitation of
measuring capacity. Moreover, large parts - like In any roundness-measuring instrument, the instru-
the four-cylinder engine block shown being ment's spindie is the most vitally important com-
inspected in Figure 113 - can be of asymmetri- ponent in its assembly. Here, the instantaeneous
cal shape, with the bore or surface being error of rotation is the key feature of the spindie;
measured offset from the centre of gravity of the since it is from the axis of rotation that all measure-
part. This offset load is not a limitation with this ments are taken, it must allow only minute deviation
 (a) Thrntable type of instrument

Radial adjustment
Component

Stylus

Table

(b) Rotating pick-np type ofinstrument

Radial adjustment

Component

Figure 112. Types of roundness instrument. ((ourtesy ofTaylor Hobson.)

 InClustrial metrology . .

Figure 114. Highly accurate and sophisticated roundness instru-

ments, based on modular design principles. ((ourtesy of Taylor
Figure 113. Large-capacity roundness instrument "Talyrond 4", fully Hobson.)
automated for roundness geometry measurements. ((ourtesy of
Taylor Hobson.)

from its fixed centre as the spindie rotates. To without loss of accuracy. Moreover, it allows the
achieve this minimal spindie error, not only must the spindle's kinematics to be quite sophisticated,
spindle's bearings be of the highest quality, but also as it can be traversed axially for straightness
they should approach almost perfect roundness. measurement and, as the same bearings are
Several types of bearing design can be employed employed for either rotary and linear motions,
in the construction of precision roundness spin dies; the axis of rotation is also the axis of linear
these include: movement;
• hydrodynamic bearing - oil (Figure 116) - from
the original roundness instrument concept, this
• dry bearings (Figure 115) - these are normally type of bearing has predominantly been used
dry, but can be of the type that are lightly lubri- for the rotating pick-up instruments. Metal-to-
cated, using a steel ball rotating in a plastic cup metal contact occurs between the two spherical
for the thrust bearing and pads of the same mate- bearing members when the spindie is stationary,
rial for controlling radial movement. This type of but as the spindie rotation reaches or exceeds 6
bearing requires minimal maintenance, having rev/min a thin and even film of oil is maintained
constant accuracy under varying conditions of between the surfaces - due to the hydrodynamic
speed and load, accepting li mi ted radialloading. "wedging effect" - resulting from the action of
Its main limitation is that the load-carrying the rotating surfaces. This film of oil is of constant
capacity is restricted; thickness at a steady speed. At either slow speed
• ball bearings (not illustrated) - these are of a type conditions or heavy loads the oil film may cease
in which the balls support the spindie both cir- to exist. These latter operational conditions
cumferentially and longitudinally. This axial and means that such hydrodynamic bearings are
radial support enables the spindie to be utilised mainly used where the load is relatively small and
in either a vertical or longitudinal orientation, rotational speed is constant. One factor that needs
 Roundness and eylindridty

to be considered is that the spindle has to be oiled

every day (when in use); Bcaring pads
• hydrostatic bearing - air (not illustrated) - this is
more commonly known as an "air bearing". This
type of bearing has been used with considerable
success notably on co ordinate measuring
machines (CMMs) and certain machine tools and
instruments, where accuracy of movement must
be combined with load-carrying capability,
making these bearings ideal for specific types of
turntable-type instruments. With these types
of hydrostatic bearings the air supply must be
clean and dry;
• hydrostatic bearing - oil (Figure 117) - in this
version of hydrostatic bearing the oil is forced
under pressure through a gap or clearance be-
tween the spindle housing and its mating spindle.
Thus a film of oil is present whether the spindle
is stationary or rotating. An oil hydrostatic
bearing has much in common with its air-driven
counterpart, but due to oil being considerably
more viscous in nature to that of air it has several
important advantages. These benefits include the
following:
• clearances between the spindle and housing Figure 115. Spindie employing a dry bearing. ((ourtesy of Taylor
can be greater; Hobson.)

Spherical bearing (top)

Spherical bearing (bol10m)

Pick-up slide rail

Figure 116. Hydrodynamic bearing for rotating pick-up roundness-testing machine. ((ourtesy ofTaylor Hobson.)
 InClustrial metrology . .

• the bearing stiffness is increased (meaning are the preferred selection for highly sophisticated
that sideways force applied to the bearing precision instruments having heavy load-carrying
has comparatively litde effect on the axis of capacities.
rotation).

In Figure 118 four types of bearings are illus-

trated that can be fitted to roundness instrument
4.3 Methods of measurement
spindIes. Table 12 indicates the "order of merit"
(ranking) performance of these spindIes and Table As the stylus traces the periphery of the component
13 highlights their respective accuracies and load- being inspected, the stylus displacement corre-
carrying capacities. sponds to movements relative to the reference axis of
From these tables the evidence suggests that spindle rotation, hence this motion will also relate to
optimum bearing performance occurs utilising the undulations of the trace on the displayed profile as
oil hydrostatic bearing, with respect to: shown in Figure 119. This displayed profile is the
same for both the rotating pick-up and turntable
1 bearing accuracy; type of instruments. On roundness equipment it is
2 vibration damping capacity - instrument vibra- necessary to position the part on the instrument so
tion is a major factor limiting the level of mag ni- that its centre nominally coincides with the rota-
fication which can be used; tional axis of the device. This centring operation will
3 power consumption - not important from the vary from one instrument to another, depending on
viewpoint of running cost, but crucial because the centring adjustments provided and whether an
low consumption equates to both less vibration automatie facility is available. If a perfecdy circular
and less heat generation. component is to be measured, there is little difficulty
in achieving successful centring, particularly with
Furthermore, this oil hydrostatic bearing exhibits the aid of the mechanical device (self-centring iris)
the penultimate performance with respect to low shown in Figure 120. Centring is achieved when there
bearing friction and high stiffness. As a result of are no cydic deflections of the stylus, as either the
these characteristics, such oil hydrostatic bearings part or pick-up rotates. However, if the component

____ 1- _ _ _

Oilout on out

Oilin Oil in

Oil out Oilout

Figure 117. 5chematic illustration of the principle of hydrostatic bearings used for turntable-type roundness machines. ((ourtesy of Taylor
Hobson.)
 (a) Dry (b) Hydrodynamie

Bearing housing

Instrument's
spindle

Oil

Spindle

Air pressurised

(e) Hydrostatic (air) (d) Hydrostatie (oil)

Figure 118. Types of roundness spindie bearings.

• InClustrial metrology

Table 12. Comparison of the bearing characteristics (Courtesy ofTaylor Hobson)

, .

Ranking Accuracy Vibration Power Friction Stiffness Load bearing

damping consumption capacity

Best C D D C D D
i D B B D C C
J, B C C B B B
Least good A A A A A A

Table 13. Accuracy and load-carrying capacity of various roundness

spindie bearings (Courtesy ofTaylor Hobson)

Bwing Maximum Accuracy (I-Lm)

load (kg)

Air 0.025 12
Hydrodynamic 0.100 n/a
Dry 0.030 20
Air 0.040 50
Hydrostatic 0.050 225

has an irregular shape optimum centring is attained

when the range of the stylus deflection cannot be
further reduced.
Once the roundness profile has been produced at
a specific cross-section of the part, then this will
allow the cross-sectional plane to be displayed and
Figure 119. Atypical displayed profile indicating the "least squares"
assessed. What these various polar plot displays
fit. (Courtesy ofTaylor Hobson.)
show, and how you can obtain information from
them regarding its roundness value, will be
described in the following section.
initial set-up, the first actlvlty is to centre and
level the component and to align the part to the
reference axis of rotation. Of late, with most round-
4.3.1 Assessment of part geometry ness software, it is possible to remove any eccen-
tricity and ovality due to initial centring and
Often a single roundness trace can be rather limiting levelling by the operator. From this single set-up, a
in its scope of information, as it is only partially series of roundness, straightness/flatness traces can
informative and gives no information about the be obtained, allowing the following analyses to be
overall functioning of the part. However, if several calculated:
roundness and/or straightness measurements are
made they can produce a considerably broader • Activity A (traces 1 and 2) - these are round-
understanding of the component's subsequent func- ness profiles that can be analysed in cylind-
tional performance. In other words, the whole geom- ricity mode with a datum set between the two
etry of the part must be considered, assuming polar trace centres. This datum will allow a
this can be assessed on a single instrument and, similar exercise to be carried out between traces
preferably, at one setting. The cross-hatched part 3 and 4, with the cylindricity of these two
shown in Figure 121 is used by way of an illustrative displayed in addition to the coaxiality of the
example of just some of the measurements that two cylinders namely, traces 1 and 2 together with
can be obtained on roundness instruments. In the 3 and 4.
 Roundness and eylindridty

In the automotive industry, one of the internal

combustion engine's most highly stressed com-
ponents is its crankshaft. This crankshaft is subjected
to continual bending, torsion and shearing loads, in
combination with variable rotational speeds, tem-
peratures and oil viscosities. For smooth running and
lifetime reliability, it is essential that no additional
stresses, strains or vibrations are imposed by the
geometry of the bearings. Normally, a main bearing
is situated between each crankpin and, upon inspec-
tion, a vital check is to ensure that these crankpins are
in line, round and straight along their entire length
(Figure 106). In a similar fashion, the crankpins must
be round and straight, with their axes parallel to that
of the main bearings. All these vital measurements
can be assessed using a suitable roundness-measur-
ing instrument (see Figure 122).

Horizontal surfaces
In most instruments the pick-up can be positioned
to detect vertical displacements of a horizontal
Figure 120. "Self-centring iris" accessory for speedy set-up and effi- surface, such as an end face (Figure 121, trace 9), or a
cient processing of batch production requirements. ((ourtesy of shoulder from a plane that is square to the axis. The
Taylor Hobson.)
whole geometry can include measurements similar
to those shown in Activities D and E illustrated in
Figure 121. By now it should be appreciated that the
• Activity B (traces 1,2,3 and 4) - can be utilised trace is taken only around a circular track and, there-
to establish a new datum running through the fore, does not indicate true flatness of the surface.
whole component. A component feature that has previously been
• Activity C (traces 7,8 and 9) - having completed centred which has a spherical geometrical form
Activity B, a flatness profile of trace 9 will not only would give a trace that superficially appears to be
show the peak-to-valley of the circular trace on flat. However, it is normally only when aseries of
the plane, but can also be displayed with its traces taken at different radii are assessed in rela-
squareness relationship to the defined axis. Flat- tionship to each other that the true value of flatness
ness traces 7 and 8, can once again be compared can be appreciated. This assessment can only be
for squareness to the established cylinder axis, or undertaken if the instrument has a known relation-
for parallelism of their flatness planes. ship between the horizontal and vertical datum.
• Activity D (traces 5 and 6) - the vertical traces When assessing the ftatness of horizontal surfaces,
(5 and 6) can be displayed in relation to a "least the probe must not be moved to the surface between
squares" line or "minimum zone straightness". By traces.
assessing one of the traces as the datum and
comparing it with the trace obtained at 180°, it is Straightness
possible to display and print out the parallelism
ofthe bore. Straightness can be measured using a pick-up
• Activity E (trace 10) - using the datum previously traversing vertically with the worktable stationary
derived from traces 1,2,3 and 4 anyadditional (Figure 121, trace 5). Measurements can be made
roundness profiles can be assessed for eccen- relative to the line of traverse, which is normally
tricity, concentricity and runout (trace 10). parallel to the roundness axis of rotation; this allows
for both the roundness and straightness measure-
One of the main features of the modern analysis ments to have the same datum and can therefore be
packages is the flexibility to store data in raw form, related to each other without difficulty. Such combi-
which can then be manipulated for additional nations of component roundness and straightness
roundness analysis to give a more precise picture of employing the same initial set-up are another
component accuracy. feature of the whole geometry measurement.
 InClustrial metrology , .

A: Rcfcrencc Bearing Surfaces B: Reference Datum

Runout Runout Runout
90·
3 1
2

180· f-Ir----f---I+-- O·

2 4
270· , ...... . .. 1
3
4
Roundness Cylindricity Coaxiality
Datum

9
Measurement Points on Part:
C: Flatness/Squarenessl
Parallelism
1"7""--,.'" ••••••••••••• ;. •••••••••••••• "17"-+"--'"'1

O· 90· 180· 210· 360·

O·
t+1±1
90· 180· 270·
i
360·

...............
4~ ~-~
~10

D: Vertieal Surfaces
E: ConcentricitylRunout 5 6 5
90·
1
I
11
I
I

..
10 270·
Straightness Parallelism

Figure 121. Typical roundness, straightness and functional part measurements that can be undertaken. ((ourtesy ofTaylor Hobson.)
 Roundness and eylindridty

Styli {or roundness instruments

Ihe only contact between the measuring system and
the part is via the stylus. Ihus stylus shape is impor-
tant and influences the information that the pick-up
obtains from the surface. By way of illustration,
many parts are produced on machine tools
and have representative tool feed marks around the
circumference, which are not part of the surface
geometry (shape) but contribute to the surface
texture - more will be said on this topic in Chapter
5. It follows that a sharp stylus (Figure 124) would
sink into every one of the machine cavities (cusps),
or scratches present and the resulting irregularities
displayed could mask the more widely spaced undu-
lations and lobes which contribute to out-of-round-
ness. As a consequence of the problem associated
with using small styli, it is standard practice to
employ a relatively large stylus (Figure 124) known
as a "Hatchet stylus". Ihis stylus acts as a filter and
bridges the gap in the closely spaced surface texture
marks (cusps), ensuring that the pick-up is insensi-
tive to these localised surface irregularities. Ihe
stylus ball tip diameter selected is dependent on
the inspected part's diameter and surface. Iypically,
Figure 122. "Talyrond 400" instrument for automated cylindrical various styli are available with diameters ranging
form measurement - bearing and alignment assessment of a crank- from 1 to 4 mm. Such styli are manufactured from
shah. ((ourtesy of Taylor Hobson.) either hardened alloy steel or sapphire tips, mounted
on an aluminium or carbon fibre shaft. Where a
particular standard stylus is unable to interrogate
the part successfully, because it may be too short
Interrupted surfaces in length, or of the wrong stylus geometry, then
special-purpose styli can be produced by the round-
Occasionally, some components have a peripheral ness instrument manufacturer.
surface that is broken by slots, splines, keyways or
cross-drilled holes whose sides - being inter-
rupted - may catch on the stylus and dislodge or
jolt the stylus arm, thereby making it impossible
to obtain accurate measurements. These interrup-
tions in the periphery can be overcome by using an
.
attachment to the pick-up, in which a stop is used to
limit the depth to which the stylus is allowed to drop
into the recess. If this mechanical stylus stop is used, .
the resulting trace will appear similar to the example
illustrated in Figure 123. It can be seen in this case
that although the trace does not show any recess
features, it is a true representation of the outer Profile
profile of the toothed pulley. If computer software is
used the unwanted data can then be removed. . .
Sly lus ll1ovCll1cnl
sloppcd olT
•
Figure 123. Interrupted surface roundness trace of a toothed pulley.
((ourtesy ofTaylor Hobson.)
 I
I
I
I
I
-----1---- -
I
I

Figure 124. Comparison of various sizes of ball-tipped and hatchet-shaped styli. (Courtesy of Taylor Hobson.)

4 the profile is shown relative to a reference circle.

4.4 Display and interpretation
Abrief discussion on these points now follows.
The majority of roundness instruments can present
The trace is not simply a magnification of the cross-
a displayed profile of the part profile, which in
section of the component.
its most elementary form is a graphical trace of
polar plot, but can also be displayed on the screen.
Due to the variations in roundness being minute,
It is important to recognise what is being displayed
typically 2 IJ-m on the periphery for apart having a
and how measurements can be obtained from such
diameter of 200 mm, a magnification of 2000 would
displayed profiles. The production of a roundness
traced profile from the instrument provides an accu- be required to ensure that the displayed profile vari-
rate and assured method of display of profile ations are visually large enough and can be
features, with an expedient means for interpretation measured. For example using a magnification of
of these measured results. Additionally, the displayed 2000, an undulation of 2 IJ-m on the component
profile can be kept as a permanent record of the surface would impart a recorded gauge deftection of
measured profile. Alternatively, some instruments 4 mm. On a similar scale, if the whole cross-section
have the facility to provide a linear form of the of 200 mm diameter were to be proportionally
roundness displayed profile, which can be construc- enlarged then the diameter of the displayed profile
tive in interpreting the waveform's regularity. The would be 400 m. To press the point still further, this
polar plot (Figure 119) will minimise conjecture in represents a polar plot paper size of 125,664 m 2 -
its interpretation because it is a visual re cord. It is mostly blank - would be necessary on which to
not simply an enlarged representation of the surface re cord the plot (Figure 125)! Moreover, the circum-
and requires the information depicted to be inter- ference of the profile would be 1.25 km, despite the
preted by the operator, before it has any meaning. fact that interest on the polar plot is only for an
Analysis of the polar plot is quite straightforward undulation of 4 mm. This highlights the fact that
on ce the underlying graphical principles have been direct enlargement of the whole profile is obviously
comprehended. The quest ion could be asked: "In impracticable and unnecessary, as the only require-
what mann er is the polar trace's display representa- me nt here is to measure the variations on the
tive of the real roundness profile?" Prior to a discus- surface, not its diameter. Hence the obvious strategy
sion on this point, it is important to fully understand is to retain the magnified integral surface variations,
just what the polar plot illustrates and, just as vital, but reject the enlarged radius, ensuring that what-
what it does not show. ever diameter is selected is of a convenient size to
Four general assumptions can be made relating display the roundness information given that data is
to such polar plots, which need to be addressed radially suppressed.
before the correct interpretation can be made:
Measurements of peak and valley heights are correct,
the displayed profile is not simply a magnification but their shape is visually distorted.
of the cross-section of the component;
2 measurements of peak and valley heights are For many scientific, commercial and statistical appli-
correct, but their shape is visually distorted; cations, the shape of any trace is often a direct visual
3 there is a direct angular relationship between the interpretation of what is portrayed; for example, the
displayed profile and the component; increasing slope of a graph might be used to indicate
 Roundness and eylindridty

Peak shape can also be modified at certain

magnifications. By way of a descriptive example of
this magnification behaviour, if slight undulations
occur at low magnification these undulating features
could be recorded as a needle-sharp peaks at
different magnifications.
The question that this discussion raises is: "Does
this mean that the trace is of no value?" Certainly
not, as the first impression of the trace is not always
valid and judgement is necessary in the mann er of
trace interpretation. This lack of visual correspon-
dence between the physical part and its associated
trace is probably the reason whya few less knowl-
edgeable engineers avoid using a roundness instru-
ment for assessment, but instead favour the far
cruder and considerably less informative dial gauge
Figure 125. Exaggerated diagram illustrating why one eannot and vee-block method (Figure 103).
produee a "polar plot" of the simple enlargement of the cross-seetion
of the eomponent. [(ourtesy of Taylor Hobson.] There is a direct angular relationship between the
trace and the component.

factory output on a week-by-week basis. Conversely, In many cases, if the roundness instrument has a
in the case of roundness assessment, it is often diffi- built-in recorder this allows the traced profile to
cult at first to comprehend that the visual image of have a direct angular relationship to the component,
the displayed profile does not give a clue to the shape because the recorder rotates at the same speed as the
of the undulations on the part. The reason for any spindie. The PC-based roundness instruments have
confusion in displayed profile interpretation was just analytical capabilities, with the display showing the
mentioned, this being the result of the radial peak data at the 0° angular position of the spin dIe on the
and valley heights being retained at their correct screen at the (conventional) 3 o'clock position. This
magnification. While the radial enlargement is great, rotational and angular datuming facility enables the
the accompanying circumferential magnification is user to relate any roundness angular feature abnor-
often less than xl, this being a reduction resulting malities to a known spin dIe position.
from the trace diameter, which must of necessity be
smaller than that of the vast majority of parts that are The displayed profile is shown relative to a reference
measured. The effect on the polar plot of this dis- circle.
parate magnification can be visually shown by the
group of displayed profiles of a four-Iobed part, Current roundness instruments can display the
which in these examples has a 3 f.1m variation in measured profile with one of the four reference
radius and is illustrated in Figure 126. Using a low circles superimposed on it - more will be said on
magnification (Figure 126a) the displayed profile will this topic in Section 4.5. This facility allows the oper-
visually differ only slightly from a perfect circle. At ator to obtain, say, a peak-to-valley value based on a
higher magnifications, the radial distance between pre-selected reference circle centre, by the simple
the outermost and innermost portions of the trace expedient of assessing the largest peak and the
will be proportionally increased (Figure 126b), until deepest valley from the reference. Most software can
at an optimum magnification the displayed profile output the peak-to-valleyvalue, together with eccen-
appears exaggerated, as shown in Figure 126(c). tricity and the runout of the profile - the latter being
These visual effects emphasise the fact that the dis- related to a selectable datum such as the spindle
played profile's shape is not simply a visually aXlS.
enlarged view of the cross-section: the displayed
profile does not contain concave portions. This visu-
ally based fact can be demonstrated by adjusting the
magnification still further to produce a square dis-
4.5 Roundness measurement
played profile (Figure 126d). In all four displayed from the display
profiles the measured maximum departure from
roundness (negating the inftuence of different scales) Previously it was explained that the roundness of a
is identical, namely 3 f.1m. component was expressed as the difference between
 '" '.'" ~

....
'A A' 'A
A'

A. ..... A. .....
~.
. ...

..'.
(a) • ~-++----:----+----;--++---: ... • :..-..+---;.----t----:--t-+---:.... (b)

"'.
",. .
..
",
~.

"' .
.... . ...

....
.... ....

'" '"
'. ~
'"
... ' . ~
....

.... A .... •
'A
A' 'A A' ,
, . .... , .....
A. I
, A.

~.
I \
.... ....
(c) • :....-_-'--+---f---t----:'---+-- ... • :..--+-----t----+----+-+--- ... (d)

"'.
",. ....
.. "'.
....
.
. ...
.
... ...
.... ....

Figure 126. Effects of magnification on the visual shape of the displayed profile. (Courtesy ofTaylor Hobson.)
The scale indicates that each division represents (a) 5 fLm, (b) 2 fLm, (c) 1 fLm and (d) 0.5 fLm on the trace.

the greatest and least distance of the profile from manual technique allows the highest peak and the
an origin. To enable a measurement of this distance, deepest valley to be found and, using the pre-set
or more correctly to assist in identifying the distance magnification utilised in this roundness assessment,
between two features, it is normally necessary to the departures from roundness can be determined.
utilise a reference circle, or a pair of circles, which The difference between the two features on the
can be drawn, recorded or otherwise superimposed displayed profile can be numerically assessed. This
onto the graph. Positioning the reference circle(s) operator-dependent procedure was somewhat sub-
with respect to the displayed profile is not an jective in interpretation and therefore accuracy and
arbitrary activity, as its chosen to fulfil specific repeatability were open to conjecture. The actual
conditions. Only in this manner can an unambig- positioning of the reference circle cannot readily be
uous and repeatable value for the departure from determined by manual methods, but this problem
roundness of any particular roundness profile be can be overcome by software that can display and
obtained. print out the reference circle(s) on the profile. These
The roundness value has historically been calcu- software facilities will not only display the peak-to-
lated by situating a transparent template of radial valley information, but can also indicate its com-
rings over a graphed profile and with the aid of the puted centre and any other relevant data to a
radial rings, centralising the profile visually. This selected datum.
4.5.1 Roundness reference circles for more exacting preClslOn, instrumentally.
Hence, the departures from roundness is the
distance of the lowest valley (V) from the circle,
Standardisation once again being equal to P + V, as now the value
of P in this case is zero (ZJ;
The measurement of roundness, like the majority of 3 Minimum zone circles (MZC) - these are two
other types of measurement, is subject to national concentric circles enclosing the traced or
and to international standards. These standards measured profile such that their radial departure
specify the manner in which measurements should is a minimum (Figure 129). The roundness value
be made and how the results are to be expressed - is then given as the radial separation Zz;
for example, which reference circles are preferred. 4 Maximum inscribed circle (MIC) - this is the circle
Conformity to these standards will ensure that any of maximum radius which will be enclosed by the
measurements undertaken on different instruments profile data (Figure 130). The departures from
are compatible. This conformity requirement is roundness is then given as the departure (peak) of
essential when components are manufactured in one the profile from the circle and is occasionally
country, or by a subcontractor, that have to be
matched to parts made elsewhere.
Manufacturers of roundness-measuring instru-
ments try to ensure that their products closely
conform to the requirements of a standard. Thus,
when a designer comprehends that a certain round-
ness tolerance is specified this component can be
inspected, measured and either accepted or rejected,
according to whether it conforms to tolerances spec-
ified unambiguously not only within the company
but in fact anywhere in the world.
The standards in Britain, Germany, Japan,
America and elsewhere have accepted four reference
circles, fulfilling different roundness requirements:

Least squares circle (LSC) - this circle can be re-

garded as the most commonly used of the refer-
ences (Figure 127). Referring to Figure 127, a circle
is fitted to the data such that the sum of the squares
of the departure of the traced, or modified profile
from that circle is at aminimum. Another way of
considering LSC is that the area bounded by the
profile on one side of the circle is equal to the area
bounded on the other side. Still yet another way of Highest
taking into consideration the mathematical peak
derivation of LSC is: "The sum of the squares of a to
sufficient number of equally spaced radial ordi- lowest
nates, measured from the circle to the profile, has valley
minimum value". The departures from roundness =PtoV
is depicted in the lower diagram in Figure 127,
illustrated as the radial distance of the maximum
peak (P) from the circle, plus the distance of the
maximum valley (V) from this circle(Zq);
2 Minimum circumscribed circle (MCC) - this is a
circle of minimum radius which will enclose the
profile data (Figure 128). The out-of-roundness is
then given as the maximum departure (valley) of
the profile from this circle and is occasionally
referred to as the "ring gauge reference circle". Its
centre and radius can be found manually, by trial
and error using either a compass, template or, Figure 127. Derivation of the "least squares circle" (LSC).
 Out-of-roundness = v Out-of-roundness =p

Figure 130. Maximum inscribed circle (MIC).

Figure 128. Minimum circumscribed circle (MCC).

As mentioned above, both the circumscribed

(Figure 128) and inscribed (Figure l30) circles are
sometimes known as the "ring" and "plug" gauge
circles, respectively, because they simulate these limit
gauges in checking a shaft or bore. It has been
observed that these terms can be criticised, because
when checking roundness features with limit
gauging equipment this is principally a three-dimen-
sional check, whereas a roundness profile can only be
considered as a two-dimensional cross-section.
The choice of referenc circle will be influenced by
the manufacturing process of the part. For example,
the LSC is less influenced by scratches than the other
techniques and is important in concentricity assess-
ment, etc.

4.5.2 Numerical value of roundness

The maximum peak-to-valley height (Z) gives the
component's numerical value of roundness. If this
Figure 129. Minimum zone reference circles (MZC). value is obtained from a displayed profile it will range
from the highest peak to the deepest valley. For exam-
pIe, if the least squares circle method of roundness
assessment is considered (Figure 127), then it is sim-
referred to as the "plug gauge reference circle". ple to obtain the distance from the circle to the
Both manual and instrumental analysis tech- extreme points (largest peak and deepest valley) on
niques can be utilised to establish the MIC. the displayed profile. However, when the minimum
Therefore, the departures from roundness is spec- zone reference circles are employed (Figure 129),
ified as the height of the largest peak (P) above the then the departures from roundness can be numeri-
circle, a value that can still be regarded as P + V, cally specified as the radial width of the zone
because now V equates to zero (as no valley exists measured from the chart between these two circles,
inside the circle) -(ZJ divided by the magnification.
 111
It should be noted that several factors need to be • inclusion of an exceptionally large peak or valley
addressed when considering any form of roundness (possibly due to a scratch or particle of dirt on
interpretation, including the following: the surface) can cause a large alteration in the
departures from roundness value (Figure 13lb).
• roundness value will convey no information as to Today, software can automatically remove these
the general shape of the profile or to the number asperities, if required.
of irregularities that occur around the compo-
nent's periphery; Straightness can be measured by the numerical
• minimum zone is not equal to the maximum value by the departure from a linear trace, for either
peak-to-valley height (compare the displayed the least squares or minimum zone circles (see Figure
profiles in Figure l31a), although in many cases 121, D5). Reference lines are superimposed onto this
the two will be virtually the same; linear trace and value calculations are undertaken in

(a) Examples showing how the relative positions of peaks and valleys can influence the
"minimum zone width"
MZC and component MZC not coincident
axis are coincident with component axis

Note the ditTerence in width

Trace of the minimum zone

(b) An example of how a single large peak can increase the apparent departures from roundness
Exceptional peak

"Minimum zone" includes "Minimum zone" neglecting

exceptional peak exceptional peak

Figure 131. Ca ses where a single dominant peak, or its relative position, (an increase the "minimum zone" width, hence the apparent
departures fram raundness. (Courtesy ofTaylor Hobson.)
 )

1
Minimum zone circles

Circumferentiallay

Maximum inscribed circle

Trochoidal, or three-Iobed
geometrie shape

Figure 132. Schematie representation of an unwound geometrie roundness profile.

the same manner as a polar plot roundness calcula- its surface roughness, which would otherwise give
tion. The linear plot of a roundness trace can visually misleading information concerning the relative
highlight any harmonie departures from roundness roundness of the part's profile. Mathematical
by the trace's sinusoidal deviation from a straight filtering can be applied to the roundness profile, or
line, as indieated in Figure 132, where a three-Iobed modified profile, to separate out the roughness from
(trochoidal) component geometry is depicted on a the general form. As an alternative action, the depar-
traced profile and its associated linear plot. tures from roundness can be suppressed, enabling
examination of the periphery for surface irregulari-
ties at higher roundness magnifications (see Figure
4.5.3 Filtering and harmonics 133). Internationally accepted filters are:

Filtering • 1-15 undulations per revolution (upr)j

• 1-50 uprj
If a component has a rough surface texture the geo- • 1-150 uprj
metrie form of this surface may be concealed by • 1-500 upr.
 Displayed profile A: Typical Harmonie analysis
example of a four-Iobed part
with ascale of 5 !lm. Harmonie analysis of a component is an important
aspect when attempting to control either the manu-
facture of the part or its subsequent in-service

..
.. ..... performance. The component's harmonics are a
product of various factors, such as component
geometry, part material and its method of manufac-
ture. In general terms, the harmonics of a profile can
be grouped together in the manner indicated in
• ~~--~------+-------~--~~ Table 14.

.. ."
"
On a circular profile the harmonie can be thought
of as a uniform waveform (sine wave) that is super-
imposed onto the part's surface (see Figure 134) .
". . Unless a "true cireIe" occurs during roundness
measurement, it can be stated that any roundness
profile will consist of aseries of sine waves that are
combined to form the overall roundness shape and
A its resulting trace. This harmonie feature can be seen
in the series of minimum zone circle (MZC) polar
Displayed profile B: The same traces on an identieal part illustrated in Figure 135,
profile with the lobes filtered where different filters have been utilised to filter out
out to allow the display to be
the relative harmonics (a-d) at aseries of undula-
shown at 1 !lm, enabling the tion cut -off values.
smaller irregularities to be
The calculation of harmonies in a representative
more c1early seen.
polar trace for the harmonie behaviour exhibited in
Figure 134 is comparatively simple to achieve, in
essence by counting the number of harmonics, with
their respective amplitudes being relatively straight-

..•
forward to measure. When a more complex har-

....
monie profile occurs, consisting of either one or two
. harmonics (though typically a component often has
between 10 and 12), then it is essential to use soft-
• ware to perform the harmonie analysis. These dis-
·..
~

crete harmonie features of the part's traced profile

. ~
are normally calculated by using what are termed
...,
Fast Fourier Transfarms (FFT). The application of
FFT algorithms, in effect, breaks down the profile
into its constituent waveforms and calculates both
the amplitude and phase angle of each harmonie.
The results of the FFT study can be visually displayed
as either a histogram or a tabulated display on the
screen (Figure 136). Figure 136(a) is an example of an
Figure 133. Effeet of filters in roundness assessment. ((ourtesy of FFT harmonie output of the roundness assessment
Taylor Hobson.)
for a component having eight equi-spaced splines on
the periphery. The harmonie histogram shows that
the relieved areas of the eight-figured spline have a
significant effect on the remainder of the compo-
Such filters will eliminate any undulations above nent's harmonics; this is evidenced by the large
the pre-selected number. Furthermore, it is also eighth harmonie and that the subsequent harmonics
possible to select a bandpass filter which removes are all multiples of eight. This minor harmonie
the form, typically 15-500 upr, which will then behaviour is the result of the part's relieved sections
display the surface irregularities with the basie form inducing "bounce" on the grinding wheel and hence
removed. workpiece as the periphery is cylindrically ground.
The first harmonie is the result of the setting-up
Table 14. The harmonic behaviour related to either the component manufacturing process or its measurement

Harmonk Cause
1st (1 upr) Function of measurement - only caused by the setting-up error on the instrument being utilised to measure the
departures from roundness. The amplitude of this harmonie is equal to the eccentricity of the part, relative to the
spindie axis of the roundness instrument
2nd (2 upr) Function of measurement or manufacture - this aspect of harmonics is generally termed ovalityand can be caused
either by a setting-up error of the roundness instrument, or the part being machined out-of-square to its axis of rotation
3rd- 7th Function ofmanufacture - these harmonics are normally introduced by the work-holding technique during
manufacture. By way of illustration, if a three-jaw chuck were used to hold the part and excessive clamping force was
employed, then upon machining and removal of the clamping forces a three-Iobed part would have been produced
lSth-upwards Function of material and manufacture - this aspect of harmonie behaviour is usually introduced to the part by either
machine instability (self-exciting vibration - chatter) or by the reaction of the materials used in the component, cutting
tool and lubricant - if used

NB: Higher harmonics could be the result of instrument noise/vibration

upr: undulations per revolution .
((ourtesy ofTaylor Hobson).

5th harmonie

l:,r-___72_o_______ 14_4_~_i_rc_u_m_~_:_r:_:_~_e _____2_8_80________':::

Figure 134. Typical harmonics occurring on the roundness displayed profiles. ((ourtesy ofTaylor Hobson.)

error (eeeentricity of the part to that of the spindie to the influenee of the eomponent's geometry on the
axis), whereas the seeond harmonie mayaiso be due stiffness of the grinding maehine. The harmonie
to setting-up error, in this ease probably due to the histogram given in this example illustrates the diag-
part not being level, that is, the axis not parallel with nostie eapabilities of the software, for interrogation
spin die axis. Conversely, the small group ofharmon- of the departures from roundness eondition of the
ies in the region of the 64th upr level is probably due surfaee profile.
 (i)

Damping faetor
12 dB/oet

Cut-off value: 15 upr 50 upr 150 upr 500 upr

Cut-off frequeney: 1.5 Hz 5 Hz 15 Hz 50 Hz
(Rotation speed: 6 rev/min)
(ii) Changes in the "assessed profile" due to different cut-off values

(a) 15 upr (1.5 Hz) (b) 50 upr (5 Hz)

(c) 150 upr (15 Hz) (d) 500 upr (50 Hz)
(upr: undulations per revolution)

Figure 135. Effect of filtering on an identical profile assessed using the "minimum zone centre" (MZC) at a range of cut-off values.
 InClustrial metrology . .

(a) Typical harmonie histogram (FFT) for a part's Mean Square (MS)
roundness profile
Magnitude Compared to the previously discussed RMS tech-
nique, the MS value tends to be more sensitive
to slight variations in the component geometry
and, under these established conditions, it is used
in preference to RMS, where a greater degree of
control is required. It should be obvious to the reader
that

MS = (RMS)2

Harmonie analysis techniques are not only

" employed in the field of surface metrology (surface
16 24 32 64
127 texture and roundness) but also in a wide variety of
inspection applications. In a11 of these harmonie
(b) An actual FFT screen display of the harmonics of a
techniques the fundamental principle is identical,
part but the methods utilised will vary considerably. In
a11 cases, the techniques used attempt to determine
the frequencies or harmonics at whieh the compo-
nent will vibrate. Yet another important aspect in the
determination of harmonie behaviour is the effect a
component will play inducing sympathetie vibration
or resonance in adjacent components in an assembly
by virtue of its discrete harmonics.

Reasons for filters

The general concepts relating to filtering were previ-
ously mentioned at the beginning of this seetion.
However, abasie understanding of profile harmonics
is necessary, in order to comprehend the reasons that
Figure 136. FFT histograms of the harmonics of a part's profile. filters are required. The operation of a filter for what-
((ourtesy ofTaylor Hobson.) ever purpose, be it for filtering oil, water or even
coffee, is simply a means of separating the unwanted
from the wanted. Roundness filters are no different.
Roundness filters are sensitive to the number of
Root Mean Square (RMS) undulations per revolution on a component; with
the application of filtering it will progressively elim-
The RMS value is exceedingly useful as an analytical inate the contribution of specific harmonics to a
tool, as it exhibits to the user the "power" of the profile.
harmonics between any two associated harmonie In Figure 135(i), the graph depicts a typieal set
values. This relative RMS value is helpful when of roundness filter characteristics. As an example of
attempting to control the part's machining process their filtering behaviour, if the 500 upr curve is
and the prospective in-service performance of the considered, this indieates that undulations occur-
component. It has demonstrably been shown that if ring 500 times per revolution can be reduced by
a designer decides on a specific component oper- 25% in amplitude. The accepted manner of denoting
ating speed, then for its stated band of harmonies the standard undulation range of filters is 1-15,
the RMS level will indicate whether or not the part 1-50,1-150 and 1-500 upr (as previously described).
is likely to vibrate in service. Therefore, by condition These filtering ranges are genera11y termed low-
monitoring, say, certain bands in the component's pass filters, because they a110w lower numbered
in-cyde rotational speed range, it is possible to stop harmonics (upr) than the maximum number speci-
and reset potential production processing, or in- fied in the filter to be retained in the profile. By
service performance before any subsequent parts utilising a 500 upr filter, as one moves to the right
are scrapped. along this curve to high er numbers of undulations
(not shown, but typieally including 1-1500-150 Hz; (a) Profile showing the 8th
1-5000-500 Hz), these signals get progressively and 60th harmonics
reduced. Conversely, if one considers values in the
opposite direction toward the lower number of
undulations their amplitudes are progressively
increased, until 100% of the amplitude is allowed
through the filter. It should be noted that a relation-
ship exists between undulations per revolution and
Hz (rotational speed dependency).
Yet another type of filter is known as a bandpass
filter; as its name suggests, it has a band of undula-
tions which it allows through. The standard value of
such a filter is 15-500 upr. Basically, at 15 undulations
per revolution the signals are reduced by 25%, with
lower numbers being successively reduced. As the
number of undulations is increased, the filter allows
more of the amplitude through until at around 150 (b) Profile with a filter
upr the amplitudes begin to reduce again, until one ofl-15 upr
reaches the 25% reduction point at 500 upr.
An example of the application of filters to a
component's roundness profile is given in Figure
137. Here (Figure 137a), a profile is depieted with the
two distinct harmonie features, namely with the
eighth and 60th harmonies displayed. In Figure
137(b) the 1-15 upr filter has been applied, whieh has
removed the 60th harmonie from the polar trace.
Figure 137(c), on the other hand, illustrates the trace
with only the 60th harmonie present, because a 15-
500 upr filter has been utilised, which removes the
eighth harmonie.
Filters have many applications and as a result can
be used for a variety of reasons, but fundamentally
they are present to "decode/unscramble" the com-
plex harmonie components represented in a round- (c) Profile with a filter
ness profile. The application of filtering enables a of 15-500 upr
detailed examination of individual effects to be
assessed; typieally the filtering operation can deter-
mine machining defects, or a part's in-service func-
tion such as predicting the effects of wear.

Gaussian filters
Gaussian filters and, indeed, double Gaussian filters
are now available for many roundness and surface
textureIform instruments; however, they differ from
some current and previous filter types such as the
2CR, RC ISO second order and RC phase-corrected
filters in two distinct ways:
Figure 137. Filtering a eomponent's profile for differing harmonie
• Gaussian filtering does not simulate a specific effects. ((ourtesy ofTaylor Hobson.)
electronic filter, but is purely a mathematieal
function whieh is applied to the profile data;
• the results obtained from a Gaussian filter have
no phase shift and therefore offer more realistie
results.
 (a) Establishing a "weighting factor"

Weighting factor

Points of inflection

Cut-off value

~-- Surface profile

(b) "driving analogy" for different filtering types

Conventional jiltering - "driver" unable to detect the contour of the road ahead until
the car senses the upward or down ward turn of the road and possibly overshoots at
the top of a steep rise.

Gaussian type - "driver" has the ability to look, as in daylight and get a better
reaction to the profile of the road ahead. In this case, the "driver" can see the general
contour of his route.

Figure 138. Weighting and filtering of roundness and surface texture instruments. ((ourtesy ofTaylor Hobson.)

A Gaussian filter has a weighting factor which has such curve inflection points mark the linear distance
a Gaussian shape (Figure 138a); this gives the ability from the arithmetic me an to each of these points,
to take account of data before and after the effective denoting its standard deviation. In Figure 138(b) a
stylus position when calculating the me an line. In simple but effective pictorial representation is given
weighting, one set of numbers are assigned as multi- illustrating the effect of employing a Gaussian filter
pliers to quantities to be averaged, indicating the compared with conventional types.
relative importance of each quantity's contribution
to the average. The cut-off can be established by the
width of the Gaussian curve and is defined as the stope and windows
width between the points of inflection (Figure 138a),
namely the two discrete points on the curve where The term slope refers to the rate of change of radius
the direction changes. In statistical terminology, with respect to angle (dr/dO). In some roundness
instrumental software, the average and maximum the slope can be calculated in terms of a rate of
slopes are given with respect to a window. The main change per degree, in J-Lm/degree. Once the slope at
reason for the measurement of slope would norm- each data point in the window has been calculated,
ally be to control component performance. For then an average value is derived for the window's
roundness assessment, the slope values have similar slope, which is known as window slope. In this
functions to those used in surface texture. The in- instance, the sign (positive ar negative) of the slope
fluence of slope in roundness analysis can be appre- is taken into account, as follows:
ciated if one considers a large slope value; this
will indicate a greater frictional force when utilised Window slope = slope at each data point in window/
in, say, a bearing application. Normally, but not number of data points in window
true for every case, if the visual finish is degraded
this denotes poor wear and hence reduced in-service Since the sign of the slope at each data point is
characteristics for the part. Displaying the average considered in this calculation, it is possible for the
and the maximum values of slope can also allow slopes at each data point to cancel each other out. By
the determination of whether the manufacturing way of illustration, if far example (Figure 139b) a
process was or is consistent. If an increase in the profile window of 5° was selected this would repre-
relative difference between the maximum and aver- sent a window slope of zero J-Lm/degree, whereas a
age slopes occur, then it may indicate that the window of 2.5° would give a much larger value,
production process has changed for one reason or hence introducing a filtering effect. As the 5° window
another. is indexed around the component's profile, the
window slope will change and it will not remain at
zero J-Lm/degree; similarly the 2.5° window would
Windowassessment eventually yield a zero J-Lm/degree. However, the 5°
window would give a lower average slope value when
The window variable operator is usually set to 5° compared to the 2.5° window.
making it, in effect, a filter. The window is the aper- Once the preliminary window slope has been
ture from which the slope values are calculated, the calculated, then this window is moved around the
calculation method being by the number of data profile by 1 data point, allowing a new window slope
points in this window, which are based on the to be automatically calculated (Figure 139c). This
following relationship: procedure is repeated - 1 data point each time - until
a window slope has been calculated for all 3600 data
Number of data points = n/360 x 3600 points. The average slope represents the mean value
far all of the individual window slopes, that is, 3600
where n = window selected (degrees) and 3600 = window slopes are averaged. The maximum slope
number of data points in the profile (one revolution can be considered to be the largest of the window
of the part). slopes.
For example, if a window of 5° were selected, then The two diagrams given in Figure 139(d) illus-
the number of data points used for each assessment trate the "filtering effect" of the window. For a round-
will be ness profile which consists of 6 upr a window of 60°
would not be sensitive enough; it would therefore
5/360 x 3600 = 50 data points yield a result of zero J-Lm/degree. However, if a
window of 30° were selected, this would be most
Pressing this point further, if a window of 15° were sensitive since it would always be in at least one posi-
chosen, then the number of data points used far each tion where it was at the true maximum slope on
calculation would represent the component. This is a "general rule" that should
be used in slope analysis, or more specifically: "The
15/360 x 3600 = 150 points most sensitive window is half the wavelength of the
undulations under consideration, and the most
As has been mentioned, each profile typically insensitive window is equal to the wavelength of the
consists of 3600 data points and, the slope or rate of undulations under consideration". If, for example,
change at each point is calculated. The slope at any just a small difference occurs between the maximum
given data point is derived by taking three adjacent slope and the average slope, it may suggest that the
data points on either side of the respective data point size of the window needs to be reviewed to deter-
- a total of seven data points (which overcomes mine whether it is sufficiently sensitive.
aliasing) - and then fitting a line through all seven Therefore, by selecting an appropriate window as
points. From this newly created line (Figure 139a), a filter and monitoring both the maximum and
 InClustrial metrology , .

(a) Calculating slope

Slope at this data point

10 3~ 50 70
---~-------
2° 4 6°

(b) The effect of window size, related to angle

\ / I

-
I /

\ / ~
I/
~ ~5°window
\ / I !'-- 2.5" window
V V

(c) Determination of window slope

Window is rotated through 1 data point.

Window slope is
calculated for the
-r------,-..:~::.-.. first window +-===:::!:~z Window slope is
position. calculated for the
second window
position.

(d) The effect of window

60 dcgrec windows 30 dcgrcc windows

odcgrecs 360 degrecs

Regardless of the position of the window, Zero slopc aximum slopc

the "window slope" is always zero.
On this profile a 30° window will go from
a zero slope to a maximum slope.

Figure 139. Effect of "window slope" and the "window filtering effect" on slope and harmonie analysis. ((ourtesy ofTaylor Hobson,)

average slope values, a high degree of control of meter is so that it acts as an indicator as to whether
component performance can be achieved. For the window is set at a suitable angle. If the average
example, if a bearing has high slope values this will and maximum values are similar, then it is likely that
tend to be "noisy" and, as such, will be more likely the window is at an angle that is not sufficiently
to suffer from premature failure in service. The main sensitive to the component features.
purpose in monitoring the maximum slope para-
4.6 Geometrie roundness
parameters (ISO 1101: 1983/
o =Conc

Ext 1: 1983)

Eccentricity
180° 0°
The parameter eccentricity (Figure 140) is a round-
ness term used to describe the position of the centre
of a profile relative to so me datum point. Eccentricity
is a vector quantity, in that it has both magnitude Datum point
and direction. Moreover, the magnitude of the eccen-
tricity can be simply expressed as the distance
between the profile centre and the datum point,
whereas the direction is expressed as an angle from
this datum point.
Figure 141. Roundness parameter: concentricity.

Concentricity
The concentricity parameter (Figure 141) is similar Runout
to that described for the term eccentricity (above),
but only has magnitude and no direction. Con- The term runout (Figure 142) is occasionally refer-
centricity can be described as the diameter of the red to as total indicated reading (TIR). This para-
circle described by the profile centre when rotated meter attempts to predict the behaviour of a profile.
about the datum point. It may be appreciated that the The definition of runout is: the radial difference
value of concentricity is twice the magnitude of between two concentric circles centred on the datum
eccentricity. point and drawn such that one coincides with the
nearest and the other coincides with the farthest
point on the profile. As a practical parameter runout
is useful, as it combines the effect of both form error
and concentricity to indicate the predicted perfor-

E =Ecc
Datum point
mance when rotated about a datum.

I = Runout
Profile centre ~ ------ - ---- ~ Nearest point
Runout ~ to datum
/ ~

E \
180° ----------~~--------
0°
\
!
/
/
~ //
270° Furthest POi~ '------- ~
from datum ~ ----
NB E =eccentricity value
e =eccentricity angle Datum point

Figure 140. Roundness parameter: eccentricity. Figure 142. Roundness parameter: runout.
 the reference plane. The term squareness is a useful
indicator of the fitment behaviour of two adjacent
=FLTt parts in aprecision assembly.
LS reference line I

Coaxiality
k--AxiS
The coaxiality parameter (Figure 145) refers to the
relationship of two cylinder axes, one of which is used
as a datum. This term is useful when attempting to
control the relative motional behaviour ofaxial and
roundness alignments between two adjacent diame-
ters.
F =flatness

Cylindricity
The cylindricity parameter (Figure 146) refers to
two, or more roundness planes used to produce
Figure 143. Associated parameter: flatness. a cylinder where the radial differences are at a
minimum. This cylindricity term will be discussed in
more detail in the following seetion.
Flatness
In the case of flatness (Figure 143) a reference plane CD ISO/DIN
is fitted such that its ftatness is calculated as the = Coax ISO,
peak-to-valley departure from that plane. In round- or Coax DIN
ness reference terminology, either "least squares
-+---~- Coaxiality
circle" (LSC) or "minimum zone circles" (MZC) can value
be utilised for ftatness determination.

Component "B"
Squareness
Once an axis has been confirmed, its squareness
parameter (Figure 144) can be described as: the
minimum axial separation of two parallel planes
normal to the reference axis and which totally enclose Axis"B"

~=sqr I
Component

~Datumaxis master

Axis"A"
(datum)

S =reference plane squareness

Figure 145. Associated parameter: coaxiality.
Radius of
measurement

Figure 144. Associated parameter: squareness.

 Roundness and eylindricity

amalgamation of a lack of straightness, poor round-

ness and taper. Furthermore, a so-called cylindrical
product mayaiso exhibit either curved, waisted or
barrelled geometrie shape, with variations in longi-
tudinal cross-section superimposed onto the part
shape (Figure 147b). For example, a component may
I
have the correct nominal size - diameter - but its
I form could be incorrect, and therefore its function
I will be impaired. Classic cases of this problem are
I encountered in automotive fuel injection systems,
where the mating cylindrieal components of valve
I
pumps and injectors must not allow a bypass of fuel
I I I under high pressure when not required, namely
I !-=l during either deceleration or braking. These fuel
: I injector systems are subject to ever-increasing and

I,r
?~/--
I~--~"'"
II~~=_-~ I
I ---- ~
stringent tolerances as greater fuel economy
becomes the norm.

Cylindricity measurement
Figure 146. Associated parameter: eylindricity.
In order to measure cylindricity, reference must be
made to the basie definitions previously described;
namely we require an instrument that will measure
4.6.1 Cylindricity radial form or roundness, axial form or vertical
straightness, dimensional uniformity or parallelism.
For many mechanical design and assembly applica- Two possible metrological sources are available to
tions cylindrieal features are crucial as they signifi- perform these measurements:
cantly contribute to the performance of a range of
products, typically revolving equipment, transmis- (a):
sion and part assemblies, precision metrology equip-
Straightness Circularity Taper Combined effects
ment, precision gauges and injection moulding dies,
enabling them to achieve their intended functions.
During the production process several factors can
cause a cylindrical feature to depart from its
intended geometrie shape or profile. Adefinition of
the measure of cylindrieity was given in Seetion 4.6,
but a simpler working description may be in order
here to appreciate its importance in precision engi-
neering applications. If a perfectly Hat plate is
inclined at a shallow angle and a parallel cylindrical
component is rolled down this plate, if it is truly
round then as it rolls there should be no discernible (b): Curved Waisted BarrelIed
radial/longitudinal motion apparent. However, if the
same plate is inclined and now a less-than-cylindri-
cal component is tested that exhibits one of the
departures from a cylindrical shape shown in Figure
147, then as the component rolls down this incline it
will either wobble or deviate in some radial or longi-
tudinal manner - this condition can be said to be
qualitative proof of a lack of cylindricity.
The schematic diagrams in Figure 147 show just
some of the combinations of geometrie departures
from cylindricity, which are invariably caused by NB Cylindrical features having similar deviation levels,
their inherent production processing - more will be aIthough with distinctly differing characteristics
said on this topic in Chapter 5. As can be seen
(Figure 147a), the combined geometrie shape is the Figure 147. Assessment ofthe geometry in eylindrieity.
 InClustrial metrology . .

1 coordinate-measuring machine (CMM); tainty for both radial and axial form has been weH
2 form/roundness-measuring instrument. documented, but what is sometimes forgotten is that
the parallelism of the instrument's spindie with
In the former case (CMM), actual volumetrie respect to the column axis needs to be addressed.
assessment of parameters such as cylindricity re- Specifically, such potential spindle-to-column align-
quires a considerable number of measurement points ment error can inftuence measurement - more will
to be made, whieh of necessity takes more time and be said on alignment checking in Chapter 6. These
provide limited accuracy. The exception to these radial slices and their axial displacement factors link
CMMs is when a very expensive CMM is utilised hav- the two-dimensional roundness and straightness
ing sophistieated continuous scanning software with measurements to the three-dimensional or total
rotational datums coupled to analogue probes. Under form analysis that is required for cylindricity
these operational circumstances, the latter case of measurement. It is therefore essential that any
using form/roundness-measuring instruments for measuring instrument used to determine cylin-
cylindricity assessment offers an ideal alternative dricity has the ability to accurately measure radial
measuring technique. form and axial form, together with dimensional
It is convenient when measuring cylindrieity to uniformity.
represent the cylinder by aseries of horizontal slices As mentioned previously, geometrie form toler-
of the radial profiles, where the axial separation is ances and measurements are described in aseries
known (Figure 148). Conventional form and round- of standards, notably ISO 1101: 1983. The inspection
ness-measuring instruments are capable of per- procedure and associated algorithm to determine
forming this type of representation, provided certain the precise measurement requirement are not pro-
criteria are observed. The contribution to uncer- vided in these standards. However, based on the

c.. ,.. !60- , _ _ ._ "'" lIt- _ "

.11iiI1'!!,,1---1S..:J ~ t 11\7 1 lr:::fhI-:-=-c.-.,-"-...-:------3''''"Ii.,.,..,..,~
~_-U -U -1 ...1....lill~.0"-1 ~..d .!!!l ~
, I I I I I I I I I

(~) ..
2883
33.011
3738
41.87 "
: :

· : · I
: :
: .
45.9&
50.25
·, · •
:
: : .
54083 : I :
!51104 , :
I
,
83.45
157.811 · :
:
I

n30 : I

78.71 " :
81 .011
85.44
·, · : : :
81178
M.OS
9834
·,:
: :
:
: :
,

10280 '. : : I

10688 I ,
111.18
115.53
11988
124 211
, :
:
·· , I
: ,
:
128.76
13317

. _-
Figure 148. Wire eage eylindricity display, utilising Windows™-based software. ((ourtesy of Taylor Hobson.)
 (a) Minimum zone cylinder (MZC) (b) Minimum circumscribed cylinder (MZC)

Cylinder axis
~-r-=-:,- __ Cylinder axis
Peak-to-valley ~---
---
Peak-to-valley ~~

-T
11

1
1 1
I 1 I
I~
I~I
I I'

I I I
I~',
l( --
",-I
--- - , ~

(c) Maximum inscribed cylinder (MIC) (d) Least squares cylinder (LSC)

___ "
~
_'

"--- - - -_/.
- I
I I
I I
I I
I I
I I
I
I
I
I
I I

I
I I
1

Peak-to-valley v""
_--_
' -",
I
~- -
~ - - /
-...!....--- '--
Figure 149. The four referenee eylinders.

standards definitions, the cylindricity of a feature is • maximum inscribed cylinder (MIC; Figure 149c);
normally assessed in two stages: firstly, a reference • least squares cylinder (LSC; Figure 149d).
cylinder is obtained utilising data obtained from
measurement of the profile; and, secondly, these A brief review will now be given for each cylin-
measured points are then compared with the refer- dricity method of assessment.
ence cylinder, allowing the derivation of cylindricity.
There are four types of reference cylinders to which Minimum zone cylinder (MZC).
data can be fitted:
The MZC technique requires information on the
total radial separation of two concentric cylinders
• minimum zone cylinder (MZC; Figure 149a); (having the same axis) which totally enclose the data
• minimum circumscribed cylinder (MCC; Figure generated during measurement and are positioned
149b); such that their minimum radial separation is known.
Due to this MZC assessment method it places equal tric. Furthermore, unlike the three previously
emphasis on the internal and external profile imper- described "cylinder fits", the axis of the LSC is very
fections to determine the reference axis. The main stable and much less infiuenced by the effects of
industrial application for MZC would be where asperities.
running (clearance) fits are required.

Minimum circumscribed cylinder (MCC). 4.6.2 Cylindricity measurement

The MCC technique relates to the minimum radius techniques
that fully encloses the profile data.
This MCC method can be utilised on precision A question frequently raised concerning the
parts where the surface of the outside diameter is measurement of cylindricity measurement, is: "How
important, such as for attribute sampling inspection many planes are required to define a cylinder?" In
techniques, typified by inspecting internal bores by theory, the minimum number of planes is two, but
limit gauging applications (plug gauges). with such a small number this can lead to problems.
For example, if a frustum (truncated cone - one of
the geometric shapes in Figure 147a) was to be
Maximum inscribed cylinder (MIC).
measured, two planes will detect any angled taper.
However, if the shape appeared in the geometric
This differs only marginally from the MCC, in that
form of those given in Figure 147(b), then this
the first terms are reversed in the following equa-
number of planes would be insufficient to detect, say,
tion, such that this MIC technique is one of where
the waisted or barrelled shapes. Basically, the
the maximum radius that fully fits into the profile
number of planes utilised must be adequate to detect
data occurs.
the variation in the vertical (axial) form. Once again,
Hence, in the case of the reference cylinder for
to define an axis a minimum of two planes is
MCC, it is the cylinder that encompasses all the data
required, although to detect variations in axial form,
points and has the smallest radius; conversely, the
or dimensional uniformity, three to five planes
MIC reference cylinder does not enclose any data
would be more realistic. The trade-off here must be
points, but simply has the largest radius. The MIC
between adequately representing a component and
method might be used for female mating part
the time taken to compute the "best fit" reference
assembly techniques, or for inspection by limit
cylinder, although with the latest instruments having
gauging using ring gauges.
significantly greater computing power than previ-
It should be noted that in all cases for the above
ously was the case this statement has less relevance.
reference cylinders they are extremely sensitive to
It should be emphasised, however, that the time
asperities, such as that caused by dirt or scratches
factor in this calculation significantly increases with
on the surface. These arbitrarily induced asperity the number of planes and/or data points per plane
effects can promote axial drift and the solution to
selected, especially when calculating the MZC, MIC
this problem will be dealt with later.
and MCC reference cylinders.
In any form of cylindricity profile assessment, it
Least squares cylinder (LSC; Figure 149d). is important to establish the relative radius of all the
planes, because this can lead to constraints with
Unlike the previous reference cylinders, which are in some stylus instruments. For example, if the only
essence a combination of mathematical algorithms means of measuring radial departure of the system
and to a lesser extent iterative trial-and-error tech- is by the pick-up, then the stylus must not decouple
niques, the LSC can be more specifically defined. (lose contact) with the component, or the radial
This LSC method is probably the most popu- reference will be lost. This is particularly relevant if
lar approach for defining cylindricity because of the pick-up had to be retracted to avoid collision
the ease of computation and the uniqueness of the with a shouldered feature on the part. This problem
solution. From a practical viewpoint, the LSC is is somewhat mitigated against by the latest round-
constructed from two or more roundness planes at ness instrumentation, due to the fact that invariably
different heights and a "best fit" line is fitted through they have an optical scale attached to the radial
the centres of the various planes. This line is termed arm, meaning that it can be moved, yet still preserve
a least squares axis. The least squares cylinder is the relative radius between the measuring planes.
then calculated by taking the average radial depar- Moreover, in this arrangement, the accuracy will
ture of all the measured data from the least squares now depend on the resolution of the radial arm
axis. It should be observed that the planes are not grating and not on the resolving accuracy of the
shifted to the least squares axis but remain eccen- pick-up.
4.6.3 Cylindrical measurement Good metrology rules of practice are extremely
important to cylindricity measurement. The major
problems reason for strict adherence to these metrological
rules is that cylindricity measurement is dependent
If a true cylinder should be tilted (Figure 145, on two axes, namely the axis of rotation of the
smaller-diameter boss) when measured, the result spin die and a linear axis. Therefore, the relationship
will appear as an ellipse; therefore it is essential between these axes must be known and stable. The
that any cylinder has been levelled prior to measure- dependence on two axes for measurement of cylin-
ment. However, it may be the case that a second dricity establishes a system requirement for the
cylinder has to be measured relative to the first straightness unit to be set parallel to the axis of the
(Figure 145, for coaxiality); in this case re-Ievelling rotating spin die. Any misalignment of the vertical
is not the answer, as the primary datum will be straightness datum relative to that of the axis of
lost. In this situation it is possible for the computer the spindie will result in a perfectly straight work-
to correct for the tilt by cakulating this tilt, then piece appearing tape red. The "coning error" of
orienting the axis and noting the radius of the the spindie will also have adetrimental effect on the
second cylinder. At this stage the computer will measured result, with a perfect cylinder giving
compensate for any error by removing the cylinder the appearance of a cone. Furthermore, if the form
tilt ovality for each radial plane, prior to performing error of the instrument's column - its straightness -
the cylinder fit. The measuring strategy of removing is significant, then this will have an effect on the
the second harmonic term or applying a 2 upr filter part's measured result and influence its peak-to-
is not adequate, as any true ovality in the component valley value.
will also be removed. The MCC, MZC and MIC axes are very sensitive
For results to have any meaning in roundness to asperities and whenever possible they should be
measurement and particularly in the case of cylin- avoided for datums, because if a large uncharacter-
dricity, various factors have to be considered; there- istic asperity occurs in one of the measuring planes
fore before any actual measurement of cylindricity this will result in a datum shift, wh ich will unduly
occurs on the instrument, three important rules influence the overall cylindrical assessment. Con-
must be addressed to obtain an accurate result: versely, the LSC axis is insensitive to asperities and
is recommended as a datum whenever possible. The
the spindie should continuously rotate during reason for this insensitivity asperity is that asperi-
measurement, which ensures that any slight shifts ties have little or very marginal effect to affect a
during the start and stop of the spindie are elim- datum in an LSC. It follows that asperities such as
inated. This operating condition, although not debris (dirt), or scratches will still affect the peak-
so critical in roundness and concentricity meas- to-valley results regardless of which reference
urements, increases in importance with the cylinder is utilised. It is therefore important to
measurement of roundness at different planes, clean both the workpiece and stylus before measure-
particularly when cylindricity tolerances ap- ment.
proaching 0.5 /-Lm are specified; A factor that needs consideration particularly
2 the maximum accuracy of the pick-up is main- with reference to the MIC and MCC (cylinders) is
tained by ensuring that it is in contact with the their inherent instability on tapered components
workpiece at all tim es. If a radial arm transducer (Figure 150). For example, if an MCC were fitted to
is fitted the resolution of measurement will be a tapered component then it must touch in at least
reduced by at least one order of magnitude worse three positions, notably at two points on the largest
than the pick-up; diameter and one on the smallest. Under these
3 the component normally requires centring and circumstances, the cylinder axis will be "skewed"
levelling with respect to the spin die axis. How from that of the component's axis. To compound this
well the part should be centred and levelled is skewness problem, there would not be a unique posi-
related to the roundness-related value that has tion, thus making its value somewhat limited in
been previously specified on the drawing for the assessment. The one case where this skewed axis
workpiece. As a general rule of thumb, the movement would have significance is when it is
centring and levelling should be within 10 times required to establish the amount of"play/clearance"
the peak-to-valley values to be measured. Any that could be obtained with a tapered shaft in abore,
residual error that occurs can be automatically or vice versa, if employing the MIC.
taken into account by the software. This auto- In all these measurement techniques that are used
matic compensation ensures that cylinder errors to determine a component's cylindricity, "sampling
are kept within the gauge range; moreover, they techniques" are utilised. Such "sampling" is essential,
allow the appropriate resolution to be selected. since it is clearly impractical to measure every single
 InClustrial metrology , .

point on the surface of a cylinder and, even if this 152 shows a three-dimensional "bed-type" plot of an
were technically feasible, the calculation itself would LSC component.
be prohibitive.
The effect of the selection of a number of planes
to be used on a cylinder has already been discussed
and, when widely spaced, results in their inability 4.7 Non-contact spherical and
to detect geometric cylindrically based shapes
such as the curved, waisted or barrelled figures roundness assessment
(Figure 147b). A potential measurement problem
can occur in the number of data points used in
each plane, because if insufficient numbers are Non-contact measurement of roundness is valid and
used then this can me an that certain profile detail has been possible for some time. Abrief discussion
will be lost, such as occasional dominant peaks of the range of instrumentation that has been devel-
and valleys. This loss of surface detail resulting oped is worthwhile, albeit if discussed in depth,
from inadequate numbers of data points means this would significantly increase the scope of the
that both inaccuracies and instability may result. current work and be somewhat counter-productive.
There are numerous ways in which the data can Despite these self-imposed objections, it is worth
be displayed by sophisticated software, notably the reviewing two techniques: one interferometry-based
multiplane screen image depicted in Figure 151. instrumental method for determining sphericity of
This figure illustrates roundness on a component at journal balls and another employing an optical
differing diametral points, indicating run out (f.Lm) sensor in conjunction with a displacement probe for
at these pre-selected locations. Conversely, Figure applications in both roundness and sphericity
assessment.

4.7.1 Sphericity interferometer

The sphericity interferometer illustrated in Figure
153 utilises in this version a monochromatic light
source along its optical path; however, as an
alternative, it would be possible to construct a
similar instrument with laser-based illumination.
As the instrument's name implies, the optical
system utilises the effect of light interference to
compare the curvature of the reftecting spherical
surface under test, typically aprecision ball, with a
curved reference surface. The interferometric prin-
ciple is as follows: when light rays are reftected
between two surfaces that are not parallel, the
difference in path lengths at particular portions
of the spherical part's surface causes phase changes
in the light reftected back to the observer. As a result
of these light path differences, some rays are can-
celled out while others augment each other. This
interference behaviour causes patterns of alternate
dark and light lines to appear across the image
(Figure 153, bottom). The shape and spacing of
these interference fringes, as they are usually known,
will depend on the regularity and spacing of the two
surfaces.
In this particular optical configuration of the
sphericity interferometer, the light is reftected
between the semi-reftecting surface of the precision
Figure 1SO. Assessment of a tapered component. The axes of the ball and a hemispherical reference optic, the result-
column and taper should be parallel, thus avoiding any potential ing interference pattern being directly related to
measurement error. ((ourtesy ofTaylor Hobson.) variations in curvature on the surface of the ball.
 Roundness and eylindricity

...
1,,-,.MtQ.l~)

, ;3.75

7.200

....
0,000

3.000
....
7<518
75.30
.. .34

2._ 8714
'.200
.,000 ....
03.30

Figure 151. Multiplane roundness/f1atness display, Windows™-based software. ((ourtesy ofTaylor Hobson.)

Figure 152. A three-dimensional "bed-type" plot of least squares eylinder results. ((ourtesy ofTaylor Hobson.)
 InClustrial metrology , .

Refereocc hemisphere

It..-___- - Sphere under test

Semi-reßccting s urface

Pitch polished steel ball: Commercial steel ball: Glass sphere:

• •
Figure 153. Sphericity interferometer and interference patterns on spherical surfaces. ((ourtesy ofTaylor Hobson.)

This type of optical interferometer can be consid- 4.7.2 Spherical and roundness
ered very much a single-purpose instrument of
limited application, but it does illustrate one method assessment by error separation
of non-contact roundness measurement. It is par-
ticularly relevant for use on precision highly Components produced to nanometric accuracy are
polished soft surfaces, which could be unintention- becoming increasingly important of late, with the
ally scratched by the action of a stylus-based round- demand for on-machine measurement to improve
ness instrument. Moreover, it is limited in the manufactured parts. The separation of potential
diameter range of spheres the technique can workpiece roundness errors from those imparted by
measure. the spin dIe error from the machine tool are important
 111
for on-maehine inspeetion. For some years, non-con- Relatively reeently, however, it has been proposed
taet measurement utilising error separation tech- that several multi-probe methods will eliminate
niques employing multiple probes to perform on- mueh of the harmonie suppression, more specifie-
maehine measurement has been undergoing devel- ally, by adopting an orthogonal mixed method
opment. As has been previously mentioned, any form (Figure 154a)j this virtually overeomes any high-fre-
of single-plane roundness measurement involves the queney suppression. In the latter arrangement, only
produet of several essential parametersj these are: single displacement and angle probes are usedj they
are arranged at an angular orientation of 90°, whieh
• workpiece roundness error - intrinsie errors has been found to yield good operation al eharaeter-
generated during the produetion proeessing isties. This orthogonal mixed method system eonfig-
technique whieh are superimposed onto the partj uration is shown in a working arrangement for
• kinematic motions - of the X- and Y-directional sphericitylroundness assessment in Figure 154(b).
eomponents. The measurement system consists of assessment of
workpieee for roundness by use of angle and dis-
In 1972, Donaldson introdueed a "Ball reversal placement probes, an air-bearing spindie motor-
technique" for error separation, whieh is analogous drive, optical rotary encoder, together with an
to that of"Straightedge reversal". This "Ball reversal analogue-to-digital eonverter and PC. In the present
teehnique" (see Figure 242, Appendix C) oecurs, eonfiguration the displacement and angle probes are
where the radial motion - given by R(8) and the ball angular-oriented at 90° with respect to eaeh other.
roundness - given by B(8) are in the prescribed rela- The roundness specimen (workpiece), is mounted on
tionship as shown. Therefore, the angular informa- the air spindie and the round/spherieal profile ean be
tion is assured to be derived from the spin dIe sampled by the prob es as it rotates. In this eurrent
position, the indicator output and the two positions configuration, the probe's output signals are input to
In(8) would produce the following relationships: a PC by a 12-bit A/D converter, as a trigger signal. To
For the two positions: avoid errors that might otherwise be introdueed as a
result of delay due to the sampling time, output
11 (8) = R(8) + B(8) signals are simultaneously sampled. The workpiece
12 (8) = -R(8) + B(8) can be adjusted in both the X- and Y-directions (via
adjustment serews), enabling any eecentrie error to
and be removed, allowing the workpiece to be inspected
within the measurement proximity range of the
R(8) = 11(8) -1 2(8) sensors. Eaeh probe is mounted on individual XYZ
2 mierostages allowing the relative positions of eaeh
sensor to be set at the desired orientation to the
B(8) = 11 (8) + 12 (8) workpiece. Results from this orthogonal mixed
2 method instrumentation are given in Figure 154(e,
d), indieating for part (c) the measured roundness
It is worth noting that this technique gives the errors of two distinet measurements and the repeata-
errors of the test ball exactlly, together with the radial bility error between them. From the test trials (c-ii)
motion of the spin dIe along the indieatorr (sensor) the roundness error was around 60 nm, with a
test direction. If the orthogonal component of the repeatability error of 5 nm. The largest individual
spin dIe radial motion is needed. It ean be found from repeatability error (e-iii) oeeurred at the third-order
a single (added) test with the indieator and ball harmonie, namely 0.6 nm, this may have been the
rotated 90°, using the known ball roundness error. result of initial set-up positioning error. Conversely,
Furthermore, with a known roundness error, the ball the relatively large measured spin dIe errors in this
ean then be employed in any of the axis of rotation eonfiguration (d) for the repetition trials gave a
test eonfigurations. spindIe error of 800 nm, with a differenee of 140 nm.
Most of the sensors used for non-eontaet round- It should be mentioned that a major contributing
ness assessment have employed a three-point factor to this spindie error was vibration compo-
method, normally using three displaeement probes nents. If a comparison is made between the plotted
that are strategically positioned to measure the results in Figure 154 for this instrumental configura-
workpieee profile. The problem with this triple- tion (small differences: c-ii and c-iiij against large
sensor technique is that high-frequency eomponents differences: d-ii and d-iii), then the roundness error
of roundness error cannot be aecurately measured (e) is separated from the large spindie error (d) with
due to harmonie suppression. Even when more high repeatability, thus eonfirming the effeetiveness
than three displacement prob es are used, this sup- of the orthogonal mixed method of non-eontaet
pression problem cannot be entirely eliminated. roundness assessment.
 InClustrial metrology , .

(a) Roundness measurement by the (b) Experimental system fiu' roundness

orthogonal mixed method

Probe 1: Displacement probe

Probe 2: Angle probe '

AdjuSlmcnl

O scrc\\'

r-"""~",,- Air spindie

Encoder

(c-i) polar plot of roundness error r( e) and (d-i) Polar plot of spindie error cx(e) and
repeatability error &(e) difference ßC x (e)

Roundncss ,.(0) 50 nm/dh' Spindie errur c.r (0) 400 nm/div

Oirrcrcncc c.• ((;/) 80 nm/div

(c-ii) Rectilinear plot of roundness error r( e) (d-ii) Polar plot of spindie error cx(e) and
and repeatability error &(e) difference ßCx (e)

,.
=5 0-
Spindie crror ex(e j
E:: =5
;
E::
... 0
.,..
; =>

..'".,
E
... e
......
=>

'"
:: Repe'llahilit~' crror t!.r (e ) ~
Dill'crcncc e... (O )
'1:0 '1:0
:: ::
::0
0 '5.
rJ)
~

0 60 120 180 240 300 360 0 60 120 180 240 300 360
Angle e deg. Angle Odcg.

Figure 154. Measured roundness and spindIe error in error separation by the orthogonal mixed method. [After Gao et al., (1997).]
 (c-iii) Spectrum of the repeatability error (d-iii) Spectrum of the difference

/' n=3

2 125
Frequency n Frequency n

Figure 154. (continued)

This chapter has briefty reviewed some of the Evans, c.J., Hocken, R. and Estler, W.T Self-calibration: Reversal,
redundaney, error, separation, and 'Absolute Testing'. Annals of
wide range of concepts, methods and techniques for the CIRP, 45, 1996, 617 -634.
roundness assessment, although for reasons of space Fan, Y., Zhang, S. and Xu, W. Kinematic and mathematical research
it has not been possible to discuss much of the other on three-point method for in-process measurement and its
valid instrumentation equipment that is currently applications in engineering. Proeeedings of the 7th Inter-
national Precision Engineering Seminar, Kobe, 1993,318-328.
available, but simply to indicate the potentialities Gao, W., Kiyono and S. Nomura, T. A new multiprobe method of
of roundness measurement to the reader. In the roundness measurements. Precision Engineering 19(1), 1996,
following chapter the surface integrity aspect of 37-45.
modifications to both surfaces and roundness will Gao, W., Kiyono, S. and Sugawara, T. High-accuracy roundness
measurement by a new error separation method. Preeision
be pursued. Attempts will be made to try and explain Engineering 21(2/3), 1997, 123-133.
how, why, where and when the surface condition can Giardini, W. and Ha, J. Measurement, characterization and volume
be inftuenced by either its production processing determination of approximately spherical objects. Measure-
ment Seienee and Teehnology 5, 1994, 1048-1052.
conditions, or through some form of material phase Gleason, E. and Schwenke, H. Spindless instrument for the round-
changes, together with any plastic deformation via ness measurement of precision balls. Proeeedings of the ASPE
mechanical working that is present. 11th Annual Meeting, 1996, 167-171.
Henke, R.P. et al. Methods for evaluating of systematic geometrie
deviations in machined parts and their relationships to process
References variables. Preeision Engineering 23,1999,273-292.
Huang, J. An exact solution for the roundness evaluation prob-
lems. Precision Engineering 23, 1999,2-8.
Journals and conference papers Huang, S. T., Fan, K.C. and John, H. A new minimum zone method
for evaluating straightness errors. Precision Engineering 15,
Carpinetti, L.C.R. and Chetwynd, D.G. A new strategy for inspec- 1993,158-165.
tion roundness feature. Precision Engineering 16(4), 1994, Kakino, Y. and Kitazawa, J. In situ measurement of cylindricity.
283-289. Annals of the CIRP 27(1),1978,371-375.
Castle, P. How to measure roundness. American Maehinist Kanada, T. Computation of sphericity from minimum circum-
December 1993,41-43. scribing and maximum inscribing centers by me ans of simula-
tion data and downhilI simplex method. International journal
Chetwynd, D.G. and Siddall, G.I. Improving the accuracy of round-
of the japanese Soeiety lor Precision Engineering 30(3),253-258.
ness measurement. journal of Physies E: Seientifie Instruments
Kato, H., Nakano, Y. and Nomura, Y. Deve10pment of in-situ mea-
9,1976,537-544.
suring system of circularity in precision cylindrical grinding.
Chetwynd, D.G. and Phillipson, P.H. An investigation of reference
International Journal of the Japanese Society for Precision
criteria used in roundness measurement. journal of Physies E: Engineering 24, 1990, 130-135.
Scientifie Instruments 13, 1980, 538. Kiyono, S. and Gao, W. Profile measurement of machined surface
Chien, A.Y. Appropriate harmonie models for roundness profiles with a new differential method. Precision Engineering 16, 1994,
with equivalent mean square energy value. Wear 77, 1982, 212-218.
247-252. Krystek, M. Formfiletrung Durch Splines. Proeeedings of the IXth
Chou, S.-Y. and Sun, C-W. Assessing cylindricity for oblique cylin- International Oberfläehenkolloquium, Chemnitz, Germany, 1996.
drical features. International journal of Maehine Tools and Krystek, M. Bias error in phase correct 2rc filtering in roundness
Manufaeture 40,2000,327 -341. measurement.Measurement 18(2), 1996, 123-127.
Chou, S.-Y. and Woo, T.C. On characterizing circularity. journal of Krystek, M. Discrete L-spline filtering in roundness measure-
Design and Manufaeture 4, 1994,253-263. ments.Measurement 18(2),1996,129-138.
Dawson, D.J. Cylindricity and its measurement. International Lai, H.Y. and Chen, I.H. Minimum zone evaluation of circles and
journal of Maehine Tools and Manufaeture 32(1/2), 1992, cylinders. Precision Engineering 36(4),1996,435-441.
247-253. Lai, H -Y., Jywe, W-Y., Chen, C- K. and Liu, C-H. Precision modelling
Donaidson, R.R. A simple method for separating spin die error of form errors for cyllndricity evaluation using genetic algo-
form test ball roundness. Annals of the CIRP, Vol 21, 1972, rithms. International journal 01 Maehine Tools and Manu-
125-126. lacture 24, 2000, 310-319.
Estler, W.T., Evans, c.J. and Shao, L.Z. Uncertainty estimation for Linxiang, C. The measurement accuracy of the multistep method
multi position form error metrology. Precision Engineering in the error separation technique.]ournal 01 Physies E: Scientifie
21(2/3),1997,72-82. Instruments 22, 1989,903-906.
Linxiang, C., Hong, W., Xiongua, L. and Qinghong, S. Full-har- Udupa, G. and Ngoi, B.K.A. Form error characterisation by an
monic error separation technique. Measurement Science and optieal profiler. International Journal of Manufacturing
Technology 3, 1992, 1129-1132. Technology 17,2001,114-124.
Moore, D. Design considerations in multi probe roundness mea- Wan, Q., Yang, S., Sun, Y. and Che, R. A new method on the evalua-
surement. Journal of Physics E: Scientific Instruments 9, 1989, tion of roundness errors. Proceedings of ISMTII'96, Japan,
339-343. September-October 1996, 174-176.
Murthy, T.S.R. and Abdin, S.Z. Minimum zone evaluation of sur- Whitehouse, D.J. A best fit reference line for use in partial
faces. Proceedings of MTDR Conference 20, 1980, 123-136. arcs. Journal of Physics E: Scientific Instruments 6, 1973,
Novaski, O. and Barczak, A.L.C. Utilization of Voronoi diagrams 921-924.
for circularity algorithms. Precision Engineering 22, 1997, Whitehouse, D.J. Some theoretical aspects of error separation
188-195. techniques in surface metrology. Journal of Physics E: Scientific
Ozono, S. On a new method of roundness measurement based on Instruments 9,1976,531-536.
the three points method. International Conference on Produc- Whitehouse, D.J. Radial deviation gauge. Precision Engineering 9,
tion Engineering, Tokyo, 1974,457-462. 1987,201.
Radhakrishnan, S., Ventura, J.A. and Ramaswamy, S.E. The mini- Zhang, G.X. and Wang, R.K. Four-point method of roundness and
max cylinder estimation problem. Journal of Manufacturing spindIe error measurements. Annals of the CIRP 42(1), 1993,
Systems 17(2), 1998,97-106. 593-596.
Zhang, H., Yun, H. and Li, J.An on-line measuring method of work-
Reason, R.E. The report on reference lines for roughness and
piece diameter based on the principle of three-sensor error sep-
roundness. Annals of the CIRP 2,1962,96.
aration. Proceedings of IEEE, 3, 1990, 1308-1312.
Reeve, c.P. The calibration of a roundness standard. National
Zhang, Q., Fan, K.c. and Zhu, L. Evaluation method for spatial
Engineering Laboratory/National Bureau of Standards Report. straightness errors based on minimum zone condition. Preci-
NBSIR79-1758, June 1979. sion Engineering 23,1999,264-272.
Shunmugam, M.S. New approach for evaluating form errors of
engineering surfaces. Computer-Aided Design 19(7), 1986,
233-238.
Spragg, R.C. Eccentricity: a technique for measurement of round- Books, booklets and guides
ness and concentricity. The Engineer September 1968,440.
Sonozaki, S. and Fujiwara, H. Simultaneous measurement of cylin- Dagnall, H. Let's Talk Roundness. Taylor Hobson Pneumo,
drical parts profile and rotating accuracy using multi-three- November 1996.
point-method. International Journal of the Japanese S()ciety for Galyer, J.F.W. and Shotbolt, C.R. Measurement of surface texture
Precision Engineering 23, 1989,286-291. <;!nd roundness. Ch. 9 in Metrology for Engineers. Cassell,
Suen, D.-S. and Chang, C.N. Interval polynomial regression by use 1990.
of a neural network for minimum zone problems. Measurement ISO RU01. Technical Drawings: Geometrical Tolerancing. Inter-
Seien ce and Technology 9,1998,913-921. national Organization for Standardization, Geneva, 1983.
Tong, S. Two-step method without harmonies suppression in Nakazawa, H. Principles of Precision Engineering. Oxford Science,
error separation. Measurement Science and Technology 7, 1996, 1994.
1563-1568. Reason, R.E. Report on the Measurement of Roundness. Rank
Tsukada, T. et al. An evaluation of form errors of cylindrical Taylor Hobson, 1966.
machined parts by a spiral tracing method. Proceedings of the Spragg, R.C. Methods for the Assessment of Departures from
18th MTDR Conference 1977,529-535. Roundness. BS 3730: 1964.
Tsukada, T. and Kanada, T. Minimum zone evaluation of cylin- Thomas, G.G. Roundness. Ch. 9 in Engineering Metrolagy.
dricity deviation by some optimization techniques. Bulletin Butterworths,1974.
of the Japanese Society for Precision Engineering 19, 1985, Whitehouse, D.J. Handbook of Surface Metrology. Institute of
18-23. Physics, Bristol, 1994.
Tsukada, T., Kanada, T. and Liu, S. Method for the evaluation of
form errors of conie tapered parts. Precision Engineering 10(1),
1988,8-12.
 Machined surface
integrity

C~rs longa, vita brevis."

Translation
"The life is so short, the craft so long to learn."
(Aphorisms, I, i; Hippocrates, 460-357 Be)

185

G. T. Smith, Industrial Metrology

© Springer-Verlag London 2002
 immediate environment via oxidation, adsorption
Outline on surface integrity and so on. Due to these complex interrelated reac-
Up to now, the discussion in the text has been prin- tions it could be said that the ultimate generating
cipally concerned with the surface topography, or mechanism would be chemical in nature. Thus, the
roundness of components. This topographical infor- unit event plays some role in influencing both the
mation is valid but disguises the fact that the subsur- component's surface and subsurface in a variety of
face material layers may have been fundamentally ways, dependent on its generating mechanism. A
altered during the production process by either complex relationship of zones could also be present;
forming or generating the part. The term surface typically the complete range would consist of the
integrity has been coined to describe the altered chemical affected layer (CAL) and mechanically
material zone (AMZ), for the localised subsurface affected layer (MAL), while a heat affected layer
layers that differ from those of the bulk material. (HAL) occurs together with a stress affected layer
These AMZs are in reality a sequence of zones that (SAL). Thus considering each in turn:
correspond to a range of subsurface features and
alterations, which can take the form of modifications • CAL - the result of chemical surface changes
promoted by metallurgical residual stresses, plastic from the production process, or by post-process
deformations, chemical changes and modifications exposure to local environment;
in hardness. They may not all be present at every • MAL - will result from factors such as material
instant of time, but any one of them can become bulk transportation (deposits, laps, folds and
dominant, depending on whether the production plastic deformation);
process and operating environment vary in their • HAL - concerned with factors like phase trans-
relative relationships and magnitudes. formations, thermal cracking and retempering;
The whole concept of surface integrity and its • SAL - results from residual stresses by a combi-
various generating mechanisms coupled to the nation of mechanical and these events.
production process is known as a unit event. If this
surface integrity unit event is considered, it may The following section discusses many of these
promote a better understanding of the mechanisms surface-related features in more depth.
behind the alterations and at the same time explain
both how and why the surface occurred. Moreover,
it enables the designer to develop surfaces that can
be engineered for specific in-service requirements. 5.1 Introduction
The unit event generation mechanisms can be
thought of as three discrete types of effect: mechan-
ical, thermal and chemical. Such mechanisms always The need to satisfy the demands of sophisticated
occur within the subsurface to some degree, their component performance, reliability and longevity
dominance altering depending upon various factors for a range of industrial, commercial and military
such as the component material, processing condi- applications where critical parts are subjected to
tions and time. However, it is often unelear what the severe conditions of stress, temperature and in
relative distinction might be between these sub- hostile environments has become increasingly im-
surface generating mechanisms. Due to the fact portant in recent years. This problem is exacerbated
that a final component's surface integrity can be by reductions in component section size, in response
quite diverse and the result of a range of inter- to the designer's need to reduce weight yet still
related factors, it is necessary to reelassify these retain mechanical strength, which means that the
surface integrity affects. Therefore, it is often more part's in-service performance is strongly influenced
appropriate to subdivide the production processes by its surface condition. Coupled to these component
into five elasses - chemical, mechanical, mechano- performance requirements, there have been contin-
thermal, thermo-mechanical and thermal - the ued increases in the development and use of heat-
order in which they are listed reflecting their respec- and corrosion-resistant alloys. One of the principal
tive power density per unit area. An increase in design factors today is that of dynamic loading
power density from the chemical end of the series and accordingly a component's design capabilities
will result in an augmented level of thermal energy are often limited by its inherent fatigue character-
entering the surface, affecting the potential for istics, particularly for structural material applica-
greater thermal damage, which in turn would lead tions. If a review is undertaken of stressed parts
to poorer surface integrity of the part. It should be and structural members that have failed in situ,
emphasised that the "chemical mechanism" is domi- their service histories indicate that fatigue failures -
nant across all elasses of production processes to a major factor - almost without exception nueleated
some degree and that surfaces react with their from a site on, or in elose proximity to, the surface of
the component. Moreover, it has been recognised phase changes and surface stresses, which constitutes
that the component's surface is a primary factor in the nature of the region termed the surface layer; in
the determination of susceptibility to attack and pre- the current chapter this is principally the result of
mature failure due to low stress corrosion resistance. machining production processes. The nature of this
Component hardness has been severely tested of altered surface region resulting from machining has
late, and to expand on this theme by inappropriate been shown to exhibit a strong influence on the com-
and somewhat less-than-informed treatment from ponent's subsequent mechanical properties. Many
some inexperienced designers the following exam- highly stressed component failures, or potential fail-
pIe highlights the problem. Many medium carbon ure modes, are associated with the part's inherent
steels which had in the past been utilised at low machining-to-service demands; this is more appar-
levels of both strength and hardness are of late being ent in some materials than others under particular
subjected to stringent heat treatment processing to machining process conditions. Below are listed just
obtain very high strength values, coupled, in certain some of the production processes, materials and cir-
applications, to greater hardness for use as struc- cumstances in which surface integrity problems may
tural parts. Such inappropriate component heat arise in surface regions:
treatment processing at these extremes goes well
beyond the demarcation boundaries initially envi- • abusively drilled or deep-drilled/bored holes
sioned for this material, which can severely limit the with poor coolant supply, which may introduce
part's performance in-service. From the discussion metallurgical changes in the material, such as
above, it can be appreciated that highly stressed untempered martensite ("white layers") most
components are influenced by at least two important notably in steels and some of their alloys;
material properties that are significantly surface- • machining-induced residual stresses promoting
oriented. Therefore, whenever these component component distortion, fatigue and the possibility
qualities are imperative, significant attention should of stress corrosion;
be made to the characteristics of the surface in its • thin component distortion, as a direct result of
pre-selected working environment. surface-induced machining stresses;
Demands by the designer for many of today's • grinding cracks resulting from either a "loaded/
components have meant that they are often subjected dull" grinding wheel (from various causes), pro-
to machining to maintain specific surface texture moting abusive grinding conditions, this being a
and/or roundness requirements, together with hold- particular problem on cast nickel-based alloys;
ing dimensional tolerances. The term surface • grinding burns occurring on high-strength parts
integrity was coined in the mid-to-Iate 1960s out of which can result in lower fatigue life for highly
the need to express a component's potential in-ser- stressed components, typically hydraulic and
vice performance from a more stringent and exhaus- stressed members on aircraft landing gear;
tive viewpoint. Surface integrity has been coined by • the influence of cutting fluids while machining,
two of the principal "champions" of the subject specifically with water-based coolants on certain
(although many people working in this field of types of materials having stress corrosion prop-
research consider themselves to be originators erties, but notably in the case of titanium;
of this important concept). The principal researchers • workpiece processing by electrical discharge
were namely, Michael Field and John F. Kahles of machining (EDM), or electrochemical milling
Metcut Research Associates Inc. in the United States, (ECM) can result in lowering of the material's
who defined surface integrity as the inherent, or fatigue strength.
enhanced condition of a surface produced in a mach-
ining, or other surface generation operation. This
important topic of surface integrity involves both the Surface alterations
study and control of two material-related factors.
These requirements can be succinctly stated as: Any surface that has had some form of value-added
activity carried out either to shape the component
1 surface roughness, or topography; from raw stock or to manipulate it from forged
2 the metallurgy of the surface region. billets or castings, typified by the following produc-
tion techniques -
First to be considered is the component's surface
texture, which principally governs the surface rough- conventional machining - drilling, milling, turn-
ness (a topic considered in Chapters 1 and 2), this ing, boring and reaming, and so on;
being in essence a measure of its surface topography. 2 non-conventional machining - EDM, ECM, laser
Secondly, a more complex inter-relationship is pre- and abrasive water jet, and the like;
sent consisting of plastic deformations, metallurgical 3 production processing - hot or cold working;
- will to a greater or lesser degree have some type Process Material
of surface alteration present. The former two, and energies properties
most notably that resulting from "conventional
machining" -induced surface integrity alterations,
\ /
Mechanical Tensile
result from the cutter's action at a zone which Electrical Ductility
indudes localised surface material rem oval. The Thermal Hardness
"conventionally machined" component's surface
layers and topography will often indude some of the
following features:

• the results of hot or cold plastic deformation; Machined

• the by-products of a too1's built-up edge (BUE), surface integrity
in du ding laps, tears and crevice-like defects; I
Roughness
• recrystallisation of the surface;
Residual stresses
• localised phase transformations; Cracks
• surface micro- and macro-cracking; Phase rangeS
• the distribution of surface layer residual stresses.

"Non-conventionally-machined" surface layers, Mechanical properties

on the other hand, may indude some of the above,
together with the following: I
High cycle jatigue
Stress rupture
• chemical absorption of elements (typified by Low cycle jatigue
hydrogen, or halogens) resulting in embrittle- Stress corrosion
ment of the surface;
• a remelted metal deposit/spattered coating on the
surface, deposited during EDM, laser machining Component reliability
or electron beam fabrication.
Figure 155. The interaction of processes with materials. [Source:
More particularly, in the case of machining oper- Bellows et al., 1975.]
ations, the principal causes for these surface alter-
ations are localised high temperatures or gradients
at or near the cutting zone, together with accom- component processing sequence is as critical as the
panying localised plastic deformation as the tool base metallurgical characteristics, due to its inherent
point "ploughs" or "burnishes" along the recently overall performance impacting on the surface
machined surface. As a result of these highly integrity. Previously, this problem of in-service
localised temperatures in the tool's cutting vicinity, performance was overcome by considerably greater
chemical reactions and possibly subsequent absorp- leeway on component design factors. The current
tion into the machined surface may occur. stringent need to conserve material, from energy
Materials are rarely homogeneous, high er- and cost-saving aspects, has led to material utilisa-
strength alloys often being sensitive to property tion doser to the original design-testing criteria,
variations in the main resulting from the presence thereby confirming the fact that a greater consider-
of minute quantities of either an element or trace ation of surface integrity requirements is demanded.
elements. In recent years complex heat treatments In order to understand the material and process
have been developed to optimise an alloy, but this interactions and potential failure mo des, Figure 155
also highlights the fact that such alloys are prone to indicates their relationship to subsequent compo-
thermal gradient sensitivity. nent reliability, while Tables 15(a and b) highlight
The amalgamation of material properties and just some of the materials having potential surface
process energies will infJ.uence component struc- alterations introduced by machining operations.
tural integrity. Often design data has been conceived
from information collated on the operational per-
formance of specially ground and polished material
specimens having a pre-selected heat treatment/ 5.2 The machined surface
thermal history. From a practical viewpoint, any
designer should consider how surface effects and
property changes have been affected by the part's When considering a machined surface, it can be
previous processing, prior to its shipment. This the result of either an abusive or gentle machining
 InClustrial metrology . .

Table 15(a). Summary ofthe potential surface alterations occurring as a result of metal removal production processes

Material Material removal methods

Conventional Non-conventional

Milling, drilling, Grinding EDM ECMandChM

orturning

5teels
Non-hardenable Roughness Roughness Roughness Roughness
Plastic deform. Plastic deform. Microcracks Selective etch
Laps and tears Recast metal Intergranular attack

Hardenable (alloy) Roughness Roughness Roughness Roughness

Plastic deform. Plastic deform. Microcracks Selective etch
Laps and tears Microcracks Recast metal Intergran. attack
Microcracks Untemp. mart. Untemp. mart.
Untemp. mart. Overtemp. mart. Overtemp. mart.
Overtemp. mart.

Tool steel (D2) Roughness Roughness Roughness Roughness

Plastic deform. Plastic deform. Microcracks Selective etch
Laps and tears Microcracks Recast. metal Intergran. attack
Microcracks Untemp. mart. Untemp. mart.
Untemp. mart. Overtemp. mart. Overtemp. mart.
Overtemp. mart.

5tainless steels
Martensitic (416 grade) Roughness Roughness Roughness Roughness
Plastic deform. Plastic deform. Microcracks Selective etch
Laps and tears Microcracks Recast metal Intergran. attack
Microcracks Untemp. mart. Untemp. mart.
Untemp. mart. Overtemp. mart. Overtemp. mart.
Overtemp. mart.

Austenitic (3 16 grade) Roughness Roughness Roughness Roughness

Plastic deform. Plastic deform. Microcracks Selective etch
Laps and tears Recast metal Intergran. attack

Precipitation hardening Roughness Roughness Roughness Roughness

Plastic deform. Plastic deform. Microcracks Selective etch

Laps and tears Over-ageing Recast metal Intergran. attack

Over-ageing Over-ageing

Maraging
(250 grade) Roughness Roughness Roughness Roughness
Plastic deform. Plastic deform. Recast meta I Selective etch
Laps and tears Resolutioning Resolutioning Intergran. attack

After Bellows et al. (1975).

regime, these being directly related to the cutting is depicted schematically in Figure 156. Here, for one
process and associated feeds and speeds. However, of the less complex machining operations, namely
this is not the complete picture for surface integrity, as turning, the surface integrity - from onlya simplistic
many other interactions influence the surface during viewpoint - is shown grouped as follows:
either its forming or generating process. Machining,
being a complex relationship of interrelated factors, • surface condition - surface texture & roundness;
affects the outcome of the production process, which • micro-structural changes - microcracks, etc;
Table 15(b). Summary ofthe potentialsurface alterations occurring as a result of metal removal production processes

Material Material rem oval methods

Conventional Non-conventional

Milling, drilling, Grinding EDM ECMandChM

orturning
--------
Nickel and cobalt-based al/or
--~======~--~==========================~

Inconel (718) Roughness Roughness Roughness Roughness

Rene (41) Plastic deform. Plastic deform. Microcracks Intergran. attack
Laps and tears Microcracks Recast metal Selective etch
Microcracks

Titanium al/ar
(Grade 5: Ti-6A 1-4V) Roughness Roughness Roughness Roughness
Plastic deform. Plastic deform. Microcracks Selective etch
Laps and tears Microcracks Recast metal Intergran. attack

Refractary al/ar
(MolyTZM) Roughness Roughness Roughness Roughness
Laps and tears Microcracks Microcracks Selective etch
Microcracks Intergran. attack

Tungsten
(Pressed and sintered) Roughness Roughness Roughness Roughness
Laps and tears Microcracks Microcracks Selective etch
Microcracks Microcracks
Intergran. attack

Abbreviations: (Machining) EDM: electrical discharge machining; ECM: electrochemical milling; ChM: chemical milling; (Metallurgy) Plastic deform.: plastic defor-
mation: Untemp. mart.: untempered martensite; Overtemp. mart.: overtempered martensite.
After Bellows et al. (1975).

• surface displacement - bulk material transporta- turning which will have harmonie departures from
tion and residual stresses; roundness characteristics present.
• surfacelsubsurface micro-hardness - plastic de- The exposed sterile surface (Figure 157) occurs as
formation and residual stresses. a result of highly localised temperatures and
transient, but cleanly cut metal resulting from this
Machined surfaces are even more complex than machining, which will instantaneously oxidise
seems at first glance; their performance can be influ- and adsorb contaminants. The outermost adsorbate
enced by external layers (chemical transformations layer is often termed the Beilby layer; it is approxi-
and plastic deformations), and/or internal zones mately 1 fLm thick and consists of many complex
(metallurgical transformations and residual stresses). factors, notably the presence of hydrocarbons and
For example, the anisotropie (periodie) turned sur- water vapour, that originated in the coolant or
face depicted in Figure 157 is influenced bythe tool tip atmospheric environment, respectively. Underneath
geometry and the regularity of the cusps (peaks and this metallic surface there is normally a plastically
valleys) is dominated by the pre-selected feedrate. A strained region that may have been metallographi-
range of other micro-topography features mayaiso be cally and metallurgically altered. The depth of this
present, superimposed onto the machined surface, layer will vary significantly, but is typically in the
such as tool wear, vibrational influences and, to a region of 10 fLm, according to the plastic deforma-
lesser extent, machine tool-induced errors. In the tion induced by the passage of the sharp tool over the
circumferential direction the lay is both periodic and surface and, is exacerbated by the metallurgical
regular, albeit this round surface is generated by composition of the metallic substrate. This plastic
 Cutting tool monitoring
I----~

Surface texture I

I
Tool wear r--------, I
Roundness I
Microstructural "------------' I
changes
r---------. I
Microcracks, etc. I
Surface
displacement I
I
Surface/sub. I
I surface Il-hardness stress I
I- - ____________
Surface integrity -.JI

Figure 156. Major factors that influence the output from a turning operation.

deformation depth, as a result of residual stresses, The physical condition of critical components
can in certain circumstances penetrate to fractions of requires to be known, including an understanding of
a millimetre.As an approximation, the depth ofhard- surface layer residual stresses l prior to in-service
ness penetration is about 50% of that for the residual applications. As has been previously alluded to, the
stress penetration depth, whereas the depth of obser- process of machining parts will generate function-
vational plastic deformation is around 50% greater ally relevant surfaces that have some significance for
than this penetration. This theme of residual stresses the development of the physical state of the surface,
resulting from machining production processes will influencing the somewhat unpredictable distribu-
be the subject of the following section. tion of its residual stresses. In many industrial appli-
cations the properties of the component's surface
dominate the functional behaviour of the part in
5.2.1 Residual stresses in machined service. The effects of residual stresses on mechan-
surfaces ical and electrical components are summarised in
Figure 158(a). This family tree of factors that affect
The residual stresses in a component are a function usage is not complete, as there are notable omissions
of the previous material process route, in combina- such as optical, acoustical and thermal results of
tion with its machining history. The fact that residual residual stresses, which are outside the current
stress levels are present may either enhance or, more narrative. Each of the residual stress effects indi-
likely, impair the functional behaviour of a machined cated in Figure 158(a) will now be briefly reviewed.
workpiece. Internal stresses in a component are gen-
erally unstable and over aperiod of reasonable time
can produce alterations in either dimensional size or Residual stress deformations
geometry. For example, in the case of certain alloy
gauge block materials, over aperiod of greater than Any residual stresses acting within a body (compo-
25 years their length has been reported to have nent) occur without external forces or moments.
changed by approximately 0.8 ,....,m with little in the Internal forces form a system that is presently in a
way of wear through usage, which is a serious dimen- state of equilibrium and if sections of this body are
sional error that cannot be ignored. removed - by machining - the equilibrium status is
 Zoncd h,)cr " in genen,l machined surfaces

~
ir
Oxides and
adsorption

Re iduul stress

Substrate

/ \
/
\

Oxides

stress

(i.e. substl"ate)

Figure 157. Cross-seetion of an anisotropie (Le., periodic) surfaee illustrating the surfaee eontaminants (oxides and adsorbates) together with
subsurfaee plastic deformation (the residual stress zone) and unaffeeted substrate.

usually disturbed, resulting in potential deforma- effects of residual stresses resulting from the previ-
tion. This distortion resulting from the machining ous production processing route.
condition is well known to industrial engineers The release of internal residual stresses must
when, for example, machining one side of a thin com- not be confused with the input of such stresses by
ponent. If either a forging or casting has not been machining, as indicated in Figure 158(b). The mach-
heat treated for stress relief and requires asymmetri- ining process gene rates residual stresses by plastic
cal machining, it will deform somewhat after deformation (Figure 159) or metallurgical trans-
unclamping from its work-holding device in the formations. As mentioned, residual stress depths
machine tool. In order to alleviate this problem an can reach several hundredths of millimetres, which
experienced machinist will release the clamping on large parts will cause little in the way of compo-
forces after roughing cuts so that stressed surfaces nent distortion when compared to the same effect
are equalised, prior to taking a finishing pass. Com- on thinner parts. For example, if a single-point
ponent deformation is approximately proportional asymetrical "light" machining operation such as
to the removed cross-section of material. Any further planing is undertaken on a 1000 mm long bar of 20
finishing is usually concerned with removing only a mm thickness, the deflection will be around 1 mm.
thin layer of material, minimising any detrimental While machining, or immediately afterwards,
 InClustrial metrology .

(a) Residual stress effects

Residual stress
effeets
1/
11
J\.
L v v
Statie Dynamic
Deformation
strength strength

Chemieal Magnetisation
resistanee
V

(b) Workpiece deformation by machining

ax

Hcat effi'cels

B -- HHHH
Heat
induction

Cross- SIre S Dcl'ormcd

section distibution \I'orkpicce

Figure 158. Effeets of residual stress and deformation of a workpieee by maehining. [After Brinksmeier et al., 1982.1

thermal influences can promote further deforma- Influence on static strength

tions. This workpiece machining can impart asym-
metrical heating to the component. As a result of the From a macroscopic viewpoint, residual stresses act
machining-induced temperature profile, stresses in a similar way to that of a pre-stressed state for the
occurring from thermal expansion are generated material. Any material that can be deformed will
which cause additional deformation; this condition have a characteristic yield point; therefore if residual
can be somewhat alleviated by subjecting the part stresses are present they will influence the level of
to subsequent temperature compensation (Figure the part's yield strength. In the case of a buckling
158b). Moreover, by vibrating the part at its natural load the critical stress is always lowered in the pres-
frequency - after rough machining (in situ), this ence of residual stresses.
allows a finish cut to to be taken and then unclamped In many industrial applications, increasing the
avoiding distortion. component's strength by pre-stressing is extensively
employed. For example, extrusion dies are often
fabricated from two cylinders that have the ability to
pre-stress the inner part, enabling the extrusion grain boundaries, or within the grains and the adja-
material to be held at a less critical state during the cent matrix, causing the precipitates to be dissolved
operation. This technique has been used to good anodically. Tensile stresses tend to open up cracks
effect in high-pressure physics and significantly developed in this mann er, promoting rupturing of
extends the pre-stressing process. the surface film at the root of the crack, causing the
"pure metai" to be attacked by the corrosive environ -
ment. In a chemical plant, one of the main failure
Influence on dynamic strength causes is stress corrosion; in a typical year around
60% of documented failures occur by corrosion,
The influence of residual stresses on fatigue strength
with greater than 20% of such detected corrosion
has been cited and proven over many years. By way of
failures being attributable to stress corrosion.
illustration, in a surface grinding operation an AISI
4340 steel component has its cutting speed varied
from 60 m/min to 180 mimin. The variables in this Magnetisation
example were for various wheel hardnesses and
coolant applications, together with either coarse or
For a ferromagnetic body, the magnetic properties
fine dressing; the resultant maximum residual
depend on its physical state. In many industrial appli-
stresses will be modified ranging from 80 to 1000
cations these dependencies are utilised for testing and
N/mm 2 , indicating how processing conditions will
measuring purposes. For instance, micro-cracks act
significantly affect induced stresses. Furthermore, if
as a disturbance in the material's magnetic flux and
the roughness of the machined surface increases, this
can therefore influence both the flux distribution and
decreases fatigue strength, as expected, although this
intensity, which can then be employed to detect sur-
textural influence for longitudinal or traverse grind-
face faults or those internally. These magnetisation
ing is considerably smaller than for that of the surface
effects may be used to investigate changes in the crys-
roughness state. Fatigue fractures are normally initi-
tallographic structure of a ferromagnetic material.
ated at the weakest point; thus it is not the average Furthermore, the coercive force is extremely sensitive
stress value that is important but the local extreme. to hardening effects; as a result it can be used for in-
This localised stress concentration value is particu-
process measurements during maehining operations.
larIy important for those machining processes that
A component's magnetic properties can be directly
generate a scattering of stress distributions. influenced by residual stresses together with any
In the example of case-hardened ge ars (16% Mn,
localised disturbanees of the crystallographic struc-
5% Cr), the effect of residual stresses resulting from
ture. The surfaee texture of, say, a soft magnetie nickel
abusive grinding operations causing burning
alloy situated in an application for the magnetic head
(without surface cracking) will lead to a 25% dec-
of a machine determines the information density of
rease in the durable flank pressure in the dynamic
the tape by the focusability of the head. This relation-
conditions associated with gearing applications.
ship is dependent on the width of the undisturbed
Under such abusive grinding operations, the surface
zone, induding the air gap between the poles, with the
residual stresses of the tooth flank in either the axial
pole surfaces being machined by a lapping process. If
or radial direction tend to be highly tensile in nature.
any loeal disloeations are present - tilting of the lattiee
structure - this will result in distinctive contrasts
Chemical resistance inside the grains. The soft magnetic properties of the
material will potentially result in degraded head
If a corrosive environment occurs and certain metals focusability as a result of these micro-mechanicallat-
are subjected to stresses on exposure to this atmos- tice influences.
phere over aperiod of time, then stress corrosion Finally, residual stresses in the machined surface
may be observed. Conditions conducive to stress layers can generate deformations that affect both
corrosion arise because of specific material sensi- static and dynamic strength, together with the mag-
tivity, in association with accompanying surface netie and ehemical properties of the surface. In order
tensile stresses in the presence of a corrosive to quantify a maehined surfaee it is necessary to
medium. Typically, if an austenitic chromium-nickel determine the residual stress tensor. This tensor
steel surface, after tension, is submerged in a solu- can be thought of as being caused by the action of
tion of sodium chloride and sulphuric acid, cracks residual strains and from further strains that result
will appear due to stress corrosion. Several theories from the component being subjeeted to eompati-
have been expounded as to what is the dominant bility conditions. In maehining processes residual
process during stress corrosion. For example, the stresses tend to be generated by meehanieal, thermal
electrochemical hypo thesis assumes that potential and transformation factors; thus characteristic
differences occur between the precipitations at the stress distributions are developed that amalgamate
 InClustrial metrology . .

in practiee across a variable and wide range. For with a highly positive and honed insert geometry is
example, shot peening intro duces an isotropie com- that the component's surface exhibits little in the way
pressive stress tensor into the surface, as does rolling, of subsurface plastic deformation, on materials that
but with distinct principal directions (anisotropie are not prone to work-hardening. When machining
surface behaviour) these are typieal of mechanical with a sharp cutting tool, where the corner intersec-
influencing processes, whereas EDM affects the part tion between the rake and clearance faces meet, the
in a thermal manner, introducing equal tensile ele- tool's edge will become progressively more rounded
ments to the surface integrity. because of the tribologieal action. This "rounding
effect" will me an that at a certain height above the
rounded portion of the tool's chip/tool interface it
5.2.2 Tribological cutting effect on will cut material and, below this contact point, it will
surface plough the surface. Ploughing is highly inefficient
and will plastieally compress the surface, acting in a
When machining a component's surface the resulting "burnishing-like" manner, as depicted in Figure 159.
machined topography will be significantly influ- Burnishing is an abusive regime of machining,
enced by the wide range of fixed and variable cutting although it is often intentionally used to produce
parameters. The ability to modify and improve this subsurface work-hardening to improve the wearing
cutting process is based on the judgement and expe- ability of the surface, yet at the same time a burnish-
rience of the operator, or on the inherent capabilities ing tool will improve the finish, by flatten-
of the computer-aided manufacturing (CAM) ing local asperities and cusps, from a previous
package. The scope to modify the cutting process is machining pass. Sometimes a material that can
considerable, with the outcome improving the work-harden will have a machined surface that gives
surface integrity, often at the expense of production the appearance of a high-quality surface texture,
output, part quality or tool life. Tool companies whieh in reality disguises the fact that considerable
expend considerable time and effort in developing work-hardening has taken place due to the burn-
tooling geometries to efficiently cut material and ishing effect of tool flank wear. A surface of this
evacuate chips from the cutting zone. Many cutting unstable type can create reliability problems at a later
inserts today have highly complex surface profiles, stage in the component's life, due to the instability of
often having a particular edge preparation - T-hone, the subsurface promoting potential for fatigue.
parabolic, etc. - either to improve tool wear charac-
teristics or to facilitate the retention of a multi-
coating for cutting inserts. Honing of the tool's rake Machinability testing
face allows a sharp cutting edge to result, which has
the ability to cleave through the workpiece material In Figure 160 are shown the results from an acceler-
in an efficient mann er. The outcome of machining ated machining test performed on Fe-Cu-C powder
Machining time
Shllrp 1001 ~=============='~\.J=============C====-1 Worn 1001

Equi·axed Hulk lransporlation SmcarcdJslrain· Elongalcd ami

grain slruclurc elf IVork-hardcncd h:lrdcncd urface dislelrlCd grain
(no meldil1 c.~ lion ) malerial; laps/lca r Siruciure
I~
')
fSurfacc
Icxlure
similur
l
I 1. .1- 10 ncw
-L ). 1001
I. 1..
- I -=-r:~ I 1 surfacc
condilion
)
-=c.
'"
I 1
Cl
"
"Abusive machining" II
[ "GenUe machining"
-<J~---- - - ---I=_

Figure 159. Distortion to the machined surface topography and integrity due to the tibological action of cutting.
(a) Sehematic illustration of "aeeeierated (b) Tabulated results from turning
machining test": 10 facing passes then test proeedure
surfaee texture and insert wear eheeked
Specimen
Saneh ik hlo('k UNII huldC'r"
2 4

"-,-,,. c•••
itN.'r1 (j.!rotd{': 1'25)

F 0 I
"-:{ii}! 11
0

t"Ull'o.lunt :<rourral'(' ~ph'd

0.10 mmJrev feedrate
Ra,~m
Rsk,~m
Rku,lJm
Rtm,IlID
0.762
0.7
2.5
3.6
75.3
0.780
0.6
2.8
4.1
69.5
0.910
0.8
3.1
4.4
97.5
0.890
0.4
2.2
3.7
90.3
r Sm,jlm
l'1:{ ruh:: . 11111 rl" :::50 IUllnili
7.4' 8.4' 5.2'
!'ort j ;.&\\ ~ DELQ 7.5"
LSC,flm 28 12 20 10

0.25 mmlrev feedrate

Ra,jlID 5.5 5.9 5.5 6.2
PI.\I bl.mk ~iLi.' : Rsk,Jlm 0.5 0.6 0.7 0.5
"-Ion )( ,\(,11) )( 60 IUII,!.! Rku,JlID 2.0 2.3 2.3 2.1
Rtm,Il ID 22 26 23 25
Sm,/lID 247 232 247 248
DELQ 13.6' 16.1' 14.3' 15.6'
LSC,flm 46 18 26 20
I

~ <D ~J' ur P:I'~C" 0.40 mmlrev feedrate

10 riu: in~·ofT 1) ; I :oiSl~ "ilh Ra,/lm 16.3 16.8 16.1 15.7
-' II U.Snun {h' llllI ur l'ul Rsk,!1m 0.2 0 0.1 0.1
Rku,IlID 1.8 2.0 2.0 1.9
Rtm,llm 64 71 71 60
Sm,jlID 401 398 402 397
DELQ 22,50 23.1' 23.2' 22.4'
LSC,flm 60 20 48 22

(e) Graphical plots of maehined surfaee texture and flank wear

1.0 1.0

0.9 0.9

0.8 0.8
r-\ e
~
0.7
/ \ 0.7
~
e 0.6 / 0.6 ~
....
5....
/!"'- \ e
....os
os
'"~
'"
0.5
l ~ --- -----
~I~
'" '''--s os
0.5 Q,
~
I::
"
e;::
0.4 ~ 0.4 .s~

II ----- ~
'"
0.3
l'l /--~- 0.3
oS"....
=
rJJ
0.2 1; /: 0.2

0.1 d 0.1

0L-~1~0~20~7.30~4~0~5=0--6~0~7~0~8~0~90~1~OO~1~10~17.20~13~0~14~0-1~5~0~1~60~1~70~17.80~19~0~20~0-2~1~0~2~2~0~2~30~O
No. of facing-off passes
1:0: 2:""': 3:0: 4: No symbol
Solid curves: refer to flank wear
Broken curves: refer to surface texture (Ra)

Figure 160. An "accelerated machining test" for assessment of tool wear and surface texture.
metallurgy (PM) compacts that were subjected to a PM surface is highly undesirable and should be the
face-turning operation at constant surface speed on cause of workpiece rejection, despite the fact that all
a CNC turning centre (Figure 160a). The results of the metrological indications state otherwise. The
four different powder mixtures are illustrated in reason for part rejection in this case would be
Figure 160(b), with graphical plots in Figure 160(c). because of surface pore closure, unique to PM
This particular accelerated machining test has the compacts, which could make these parts useless for
objective of assessing the relative machinability of porous bearing applications.
components based on a relatively short time. scale
(termed "ranking"), this being an amalgamatiOn of
the previous degraded tool and rapid facing tests, 5.2.3 Micro-hardness testing
devised by the author. The test quickly establishes
the relative machinability of differing specimens, Utilising micro-hardness testing techniques for
with a slightly degraded tool and marginally faster surface integrity assessments pro duces an indirect
rotational speed for the facing operation, this being comparison of the potential residual stress in ~he
undertaken with a constant surface speed feed func- mechanically affected layer (MAL). Such destructlVe
tion engaged on the turning centre. techniques require sampies to be sectioned - the
The tool life and wear curves are shown for the sectioning technique and the reasons for under-
four specimens in Figure 160(c), with the toollife in taking it will be discussed shortly - although the
particular showing the anticipated three stages of method should be used with caution, as any hard-
wear (Taylor curves) associated with steady-state ness plots obtained cannot show whether forces are
cutting operations. These reported wear stages are compressive or tensile in nature. A major benefit
the initial edge breakdown (tool point rounding), from utilising micro-hardness testing is that it
steady-state wear as the flank progressively ?egen- allows metallographically prepared specimen fea-
erates while in-cut, and finally catastrophlc tool tures to be optically investigated and assessed with
edge breakdown as the edge completely fails. The the test sampie in situ on the instrument. A range of
elemental additives to these powder compacts repre- micro-hardness tests are currently available, the
sent the major variable in the cutting process, this indentation being of two distinct types:
interplay of elements and their relative proportions
within the sampie significantly alte ring the shape of the geometric shape of the impression left after a
the wear curves (some specimens are more abrasive hard indentor has been pressed into the surface
than others). feature to be measured - typified by Vickers and
Superimposed onto Figure 160(b) are the circum- Knoop indentors (Figure 161b);
ferential surface finish results for each testpiece. It is 2 where a inden tor penetrates into a specimen and
curious to note from these graphical plots that orig- relative depths to which the indentor penetrates
inally a high Ra value was established during the are the measure of the specimen's hardness -
early stages of the operation during the initial tool typified by the Rockwell indentor.
edge breakdown; however, this surface .Ra va~ue
dramatically improves during the succeSSlve facmg
passes, regardless of metallurgical composition. This Vickers hardness
improvement is thought to occur because the flank
wear that is present on the cutting insert pro duces a The faces of the square-based diamond pyramid for
burnishing effect on the machined surface topog- a Vickers indentor are inclined at an angle of 136°;
raphy. Burnishing produces plastic deformation and this angle approximates the most desirable ratio of
in essence compresses the surface asperities and indentation size to that of a ball diameter in Brinell
localised surface layers, giving rise to an improve- hardness testing. Because of the shape of the
ment in the recorded surface texture. However, a indentor it is often known as the diamond-pyramid
surface condition of this type is misleading, as both hardness test. It is defined as the load divided by the
the surface topography and subsurface integrity are surface area of the indentation. The Vickers hard-
locally extremely work hardened. This work hard- ness number (VPN) may be determined from the
ening continues to progressively degenerate until the following equation:
insert edge completely fails; furthermore at the latter
stages of the flank's life the surface texture is exac- 2P sin (812)
erbated. A secondary effect of this burnishing action VPN =
by the insert's flank is the transport of workpiece
material and its sub se quent deposition in a highly
work-hardened state at new sites on the previously
machined surface. This smearing of the machined
 (a) Tapered section of component, for metallographical and micro-hardness assessment

Thrned and bored

component

I
I NB: An inverted capital "A" is the
I symbol for Surface Integrity in the
I
I USA, dealing with sub-surface features
I (American Standard B211.1).
I
I
I
\
\
\ I ' -........I<.AI
\
,' - /
/

Micro-Vickers indentations

Residual stress zone

Machined cusps

(b) Detail of "micro-hardness footprinting" across the taper section

Figure 161. Tapered seetioning of a component, for metallographieal inspeetion and miero-hardness "footprinting".

where P = applied load (kg), L = average length of Knoop hardness

diagonals (mm) and () = angle between opposite
faces of diamond (136°). The Knoop indentor has more complex facets to
The diamond indentation in the surface would its diamond indentor, having angles of 130° (short
ideally be a square shape with diagonals from corner diagonal) and Ins (long diagonal) respectively.
to cornerj however, a "pin-cushion" indentation This arrangement of the indentor geometry leaves a
(concave sides to the square) would occur due to significantly narrower and longer impression
sinking into the material's surfacej conversely a (approximate diagonal ratio of 7:1) in the surface
"barrelled" indentation (convex sides to the square) than that of the Vickers indentation. The Knoop
occurs by ridging. The test has found wide accep- hardness indentation test was developed by the
tance and can be utilised from relatively soft surfaces National Bureau of Standards (USA), the Knoop
up to the hardest surfaces yet produced. hardness number (KHN) being defined as the
applied load divided by the unrecovered projected
area of the indentation. The following equation
relates to the surface indentation:
 Table 16. RockweIl hardness test indentor scales, geometries, loads
KHN and prefixes

RockweIl Indentor Shape of Load (kg) Prefix

P test indentor
L2 C
A 60 HRA
where P = applied load (kg), A p = unrecovered ( Diamond 120° cone 150 HR(
projected area of indentation (mm 2 ), L = length of 0 100 HRo
long diagonal (mm) and C = a constant for each B 100 HRB
indentor supplied by manufacturer. F Hardened
The unique shape of the Knoop indentor enables steel sphere 1/ 16"<1> ball 60 HRF
it to position adjacent indentations much doser G 150 H~
E Hardened
together than, for example, the Vickers inden tor, steel sphere 1/8" <I> ball 100 HR!
allowing measurement of a steep hardness gradient
(Figure 161b). Moreover, the Knoop's long diagonal
length of indentation at a given depth is only 15% of
that for an equivalent Vickers impression of identical
with a letter from the alphabet, indicating the partic-
length. These width and depth advantages can be
ular combination of load and indentor for the hard-
used to good effect when assessing surfaces with a
ness scale utilised, as illustrated in Table 16.
thin layer, by utilising the footprinting technique.
This footprinting entails taking dosely packed
indentations across a specimen's transverse seetion
of subsurface features (Figure 161b) to determine
Micro-hardness applications
the hardness profile, or taking hardness readings
In Figure 162 is illustrated how important informa-
into anormal surface,but increasing the pre-Ioad,
tion resulting from work hardening related to
until the thin layer is pierced, thereby gaining an
localised plastic deformation and thus, residual
indication of the depth of the hardened layer.
stresses, due to the machining process and tool
geometry can be established. Here, the drilling of
RockweIl hardness "blind holes" by a jobber drill with a conventional
drill point geometry (118°) and split-point drill with
The Rockwell Hardness test was invented in 1919 by web-thinning and modified point angle (135°) has
Stanley P. Rockwell and is a widely used testing been assessed, employing consistent cutting data.
procedure, particularly in North America. The test The indirect measurement of the residual stresses in
employs a variety of scales and either hardened steel the work-hardened region is confirmed by obtaining
balls (for softer materials) or a diamond cone (120°) hardness plots through a "taper seetion" (more on
indentor with a slightly rounded point termed a this topic in the foUowing subsection) at three pre-
Brale inden tor - the latter being used for micro- defined hole depths ("slice levels"). The results of the
hardness testing. The operation differs from the drilling process show increases in hardness in the
previously discussed methods in that hardness is proximity of the hole's surface, their values steadily
correlated with the depth of penetration by exerting decreasing the further such readings are taken into
apre-load that reduces the amount of surface prepa- the substrate, indicating the limit of the localised
ration, while at the same time minimising the work-hardening effect. More specifically, as the
tendency for the indentor to ridge or sink. After pre- jobber or split -point drill penetrated into these
load has been applied, then the major load is compacts, the local hardnesses were modified as
engaged and the depth of indentation is automati- foUows:
caUy recorded on a suitable dial gauge scale; the scale
is reversed, so that a high hardness corresponds to • jobber drill with a penetration at low feed (Figure
smaU indentor depth, relating to a high hardness 162a) - the mean micro-hardness increased from
number. The whole process can be automated for 193 H y @ top, by 16% and 33% at a depth of
movement of the indentor to obtain a pre-selected lOmm and 20 mm respectively;
surface hardness profile across the specimen being • jobber drill with a penetration at high feed
tested. (Figure 162c) - the mean micro-hardness in-
Due to the fact that the RockweU hardness value creased from 248 H y @ top, by 10% and 15% at a
is dependent on the load and indentor, it is im- depth of 10 mm and 20 mm respectively;
portant to specify the combination that is employed. • split-point drill with a penetration at low
This is achieved by prefixing the hardness number feed (Figure 162b) - the mean micro-hardness
 Low compaction PIM components:
(a) Jobber drill (feedrate: 0.127 mmJrev)

r:=E
(b) Split-point drill (feedrate: 0.127 mmJrev)
115mm • 115mm
l~,rn ~ 150llm
:1 l
,
I 100
1lm

,50 llm<l---- I
I
,
:~100Ilm 1
,50 llm-Q- I l
225 198 157 (Hv ) 251 222 198 (Hv )
~ ~ ~ ~ ~ ~
8 8
8 8
....
0
~

286 212 178 298 264 229

8 ~ ~ ~ 8 ~ ~ ~
8 8
0 0
N N

306 272 194 314 288 218

~ ~ ~ $ $ ~

High compaction PIM components:

(c) Jobber drill (feedrate: 0.254 mmJrev) (d) Split-point drill (feedrate: 0.254 mmJrev)

8 115mm

I
'''',rn
100 11m J
115 mm
~100flm1
150 11m

1
I I
50llm<l--
I
! 1~50Ilm~ I

Top 84 260 202 (H v ) Top 284 260 202

e
8
f- ~ ~ ~

8
8
f- ~ ~ ~

....
0
....
0

300 276 248 300 276 248

8 ~ ~ 4 8 ~ 4 4
8 8
0 0
N N

310 292 262 310 292 262

$ ~ ~ ~ $ $

Figure 162. Influence on localised residual hardness, near the drilled hole edge, with variations in drill geometries and component hardnesses.
 increased from 234 H y @ top, by l3% and 8% at such as phase transformation, plastic ftow zones,
a depth of 10 mm and 20 mm respectively; localised cracking, and bulk transportation of
• split-point drill with a penetration at high feed material (redeposited material, re-entrant angles
(Figure 162d) - the mean micro-hardness in- and under-cutting) can be undertaken that would
creased from 264 Hy top, by 11% and 16% at a otherwise have been missed if only profilometry
depth of 10 mm and 20 mm respectively. assessment had been used. As its name implies,
"tapered sectioning" overcomes the limitation of
These hardness data plots illustrate that for a perpendicular sectioning by modifying the mag ni-
given feedrate and, in nearly all cases, the jobber drill fication of the subsurface features without undue
did not promote significant plastic deformation in distortion to the transversely cut surface topog-
the hole's surface region when compared with to a raphy. The procedure is illustrated in Figure 161 (a),
split-point drill. This hardness increase globally where previous production processes, in this case
ranged from 6-7% at higher feeds to 21 % at the turning and boring, have been produced and the
lower feedrate. The mean difference in micro-hard- subsurface detail would otherwise have been lost (as
ness, at its extremes, ranged from 47% from the top these modifications may occur in a relatively small
surface of the low-compaction specimen to the and localised surface zone), if a conventional per-
bottom of the high-compaction variety, with this pendicular section had been taken. Taper sectioning
difference increasing to 56% in the case of the high- can overcome the potential distortion problem, by
feedrate/compaction specimens at the deepest hole. selecting a section angle that increases the vertical
The overall increases in plastic deformation values magnification without unduly inftuencing the metal-
from the jobber to the split-point drills are thought lographical and topological features of interest.
to be due to three factors: Figure 161(a) illustrates the transverse component
cutting procedure, showing that an 11 ° section has
the greater bulk hardness of the high-compaction been made. This 11 ° sectional cut improves assess-
PM varieties increasing the overall hardnesses of me nt discrimination by increasing the vertical
the compacts after drilling; section magnification by around five times. Thus, the
2 as the point angle increased from 118 (jobber
0
taper section angle (TSA) will be 79°, the vertical
drill), to 135° (split-point drill), this induced magnification being obtained from the following
greater plastic deformation - hardness - in the expression:
hole's surface region;
3 as penetration of the respective drill types TSM = secant (TSA)
occurred, the abusive regime of drilling also
increased because of higher drilling tempera- where TSM = taper section magnification and TSA
tures, toward the hole bottom, and this was exac- = taper section angle.
erbated by poor ejection of the work-hardened Such vertical magnification by taper sectioning,
chips and compounded by the drill's margin increases the ability to obtain valid hardness infor-
inftuence on the hole's wall- "burnishing effects". mation on, for example, a machined surface, by
taking staggered hardness readings across the cut,
In all of the above drilling operations, the PM polished and etched surface as illustrated in Figure
compacts were drilled without coolant, which if 161(b). This subsurface feature of interest that has
utilised would have reduced the abusive hole been plastically deformed/mechanically altered is
generation procedure considerably, towards a "gende often quite small in width, somewhat less than 0.1
regime". However, this discussion illustrates the mm wide, requiring a small micro-hardness inden-
information that can be gleaned from undertaking tor such as the Knoop, as this indentor geometry and
hardness data studies and the effect that abusive associated readings can be more closely packed than
machining plays in just one aspect of a machin- the equivalent micro-Vickers impressions. It is
ability/surface integrity study. normally advisable to obtain several micro-hardness
readings in the subsurface area of interest, as an
otherwise spurious result can heavily inftuence the
Taper sectioning overall hardness profile result; therefore it is advis-
able to take severallocalised hardness readings and
In order to improve the metallographical assessment then average them to obtain mean values to
of a sectioned surface and its subsurface features, minimise such inftuences.
while achieving greater discrimination in hardness Caution and some degree of care must be taken
profile analysis across the viewed area - after appro- when attempting to section a metallurgically altered
priate polishing and etching - tapered sectioning can subsurface workpiece feature prior to assessment, as
be employed. Interrogation of subsurface features when sectioning the component heat can be acci-
dentally induced into the transversely cut surface by • an increase in the susceptibility to stress-corro-
the previous process of cutting and polishing the sion;
surface for etching and then testing. This undesir- • increased probability of surface material break-
able heat induction due to sampie preparation could out and generation of debris;
"swamp" the potentiallocalised subsurface hardness • higher wear rates resulting from "three-body
values and give significantly misleading and unrep- generation";
resentative hardness profile results. • surface delamination and fatigue.

Cracks may be considered as either separa-

5.2.4 Surface cracks and white II tions or narrow ruptures that interrupt the surface
continuity and usually include sharp edges, severe
layers" directional changes or both. Macro-cracks can
normally be visually seen unaided, whereas micro-
Surface cracks cracks require microscopic examination. Often com-
plex metallurgical interactions are compounded
Any cracks that are present at the free surface which with an abusive regime in the production processing
extend into the substrate are potential sites for that contributed to the unacceptable surface con-
premature component failure for highly stressed dition. A crack's origin can be the result of some
components. It has been shown in the UK railway extremely complex phenomena; typically they can be
industry in recent years that despite track being promoted by an inter-granular attack that might be
precision machined and occasionally inspected by exacerbated by surface dissolution, via chemical
non-destructive testing (NDT) techniques, instances processes. Whenever preferential intergranular
have occurred when the rails (employed on high- attack takes pI ace, this additionally intro duces a
speed corners) have delaminated, thus causing a grain boundary network of micro-cracks that can
passenger train to lose contact and crash, resulting extend beneath the surface, following the underlying
in significant loss of life. The method of machining grain boundaries. Micro-crack sites cannot be
can contribute to the susceptibility of surfaces to ignored, as they can affect the component's func-
fail. In the case of milling operations, it has been tional performance because they act as a potential
recognised that conventional or up-cut milling source for macroscopic crack failure. Once a crack
(Figure 163a) can introduced surface tensile residual has been generated it cannot be successfully re-
stresses into the surface layers of a component. If this sealed, owing to subsequent contamination and
component is then subjected to both an arduous chemical reactions. Just because a manufactured
and potentially fatigue-inducing environment, then surface visually appears to be perfectly smooth and
these tensile layers can open up and may result in of high integrity, it may not be without fault and
premature failure. Conversely, a machined compo- should be treated as a potential failure site.
nent that has been climb-milled or up-cut milled
(Figure 163b) will induce surface compressive
residual stresses. This type of stress concentration "White layers"
has invariably been shown to remain closed and
avoid crack growth, under identical circumstances Certain ferrous-based work-hardening materials, if
to those previously mentioned. It has been recom- they have been subjected to an abusive regime, say,
mended for CNC milling and machining centre of machining, from a metallurgical viewpoint can
applications that climb-milling on a component not exhibit an undesirable and unwanted trait termed
only generates a compressive stress into the surface "white layers". This metallurgically unstable sub-
but also has the added benefit of normally drawing surface, refers to localised hard surface layers that,
less spin die power. when etched and viewed through a microscope,
Both craters and pits in a machined surface do appear as white, featureless and bland metallo-
not pose too great a problem in terms of influencing graphical areas, hence the term white layer (Figure
fatigue life of parts, because their depth-to-width 164). Previously, they were known as "white phase",
ratios tend to be shallow, although from the cosmetic "white etching" or "hard etching" and, depending
appearance viewpoint they are unacceptable. Cracks upon the variety of white layering production,
are normally classified as either being micro- or several other classifications have been identified.
macro-cracks; these cracks have depth-to-width The classification depends upon whether a mechan-
ratios greater than four; typically they can promote: ical, chemical or thermal event has transpired, which
in turn directly relates to factors such as strain,
• a reduction in mechanical strength, fatigue and strain rate, heating and cooling rates, plus environ-
creep; mental conditions.
 (a) ConventionaVup-cut milling

Rotation
Milling cutter
Surface tensile residual stresses
Plastic deformation

[ ~~<>:::7_1
~~~~~~-l
Depth
Surface cracks ofcut
tend to open up

"Abusive machining"

Crack

Rotation

rPlastic deformation
ne
Surface compressive residual stresses ,k!lr""(""1=ri=:-T--c-,-",- ~

~~~~tf1ttifj}-1
Unaffected substrate Depth
Surface cracks ofcut
(if present) tend
to dose up

Less "abusive machining"

(b) Climb/down-cut milling

Figure 163. Fatigue characteristics of the surface region being influenced by the mode of machining.

In Figure 164(b) a white layer exists beneath a be very hard and in this example (Figure 164b) the
re cast and redeposited layer, resulting from abusive recorded hardness was 62 HRe> with a significantly
drilling with a dull drill. Due to ihe fact that the softer layer beneath this zone. In this situation, the
re cast layer has a similar metallurgy to that of the white layer-generating unit was probably a combi-
white layer, the delineation of their respective zones nation of both thermal and mechanical events.
is not clearly defined. Underneath the white layer is The thickness of a white layer in the drilling exam-
a complex metallurgical zone consisting of some pIe is strongly infiuenced by both the potential plas-
white layering, untempered martensite (UTM), tic deformation that can be created in this vicinity
together with over-tempered martensite (OTM), and to a lesser degree by the thermal infiuence of the
while beneath these layers the bulk material of the passage of the dull drill as it penetrates through the
substrate remains unaffected. White layers tend to workpiece. The problem is compounded by lack of
 (a) Surface produced by a new drill, indicating no surface thermo-mechanical generation. The point has
aIterations already been made that many machining processes
impart residual stresses into the machined surface
layer; this can significantly increase the tendency
toward component distortion. For example, when a
face-milling operation occurred on 4340 steel that
was quenched and tempered to a bulk hardness of 52
HRe> a tensile residual stress was the less-favoured
machined surface condition, as illustrated in Figure
165. Even with cutting inserts on a sharp tool, a cer-
tain degree of tensile residual stress was apparent in
the immediate surface region, directly under the sur-
face (zone width of around 50 fLm); this stress con-
centration changed to one of compression. After
successive milling passes, the flank wear increased in
proportion to the Taylor wear curves:
(b) Surface produced by a dull drill, indicating "white layering"
together with surface cracking. NB untempered Taylor's general cutting tool wear relationship:
martensitic phase of the workpiece 62 HRc
C= VTn

where V = cutting speed (m/min), T = tool life

(minutes) and C and n are constants, being tool- and
material-derived.
As the cutter became steadily more worn and the
flank wear increased, as did the subsurface compres-
sive layer, so this would lead to even greater poten-
tial for distortion.
When the forces involved in the production
process exceed the flow stress, plastic deformation
occurs and the structure is deformed. In the case of
ductile materials, the plastic flow can create a range
of degenerative surface topography characteristics,
including burrs, laps, a residue from the tool's built-
(c) Blind hole drilled under an "abusive" machining regime, up edge (BUE) and other unwanted debris deposits.
indicating the "white layer" at hole's bottom
If this deformation becomes severe because of exces-
sive plastic flow, any grains adjacent to the surface
may become fragmented to such an extent that little
or no structure can be metallographically resolved,
hence white layering will result. Normally, a white
layer region extends to quite a small depth beneath
the surface, typically varying between 10 and 100
fLm, depending upon the severity of the abusive
regime of surface generation. Although the graph
illustrated in Figure 165 has residual stress along its
. ..... vertical axis, it is possible to superimpose a micro-
( : ; . ' ' :' . hardness axis here; this is because the shape of the
hardness profile plot closely follows that of residual
stress. Care must be taken when employing this
Figure 164. Surface integrity resulting fram an "abusive" drilling
machining regime. [Saurce: Field and Kahles, 1971.] analogous treatment, as residual stress and hardness
are two distinct quantitative values. As has been
mentioned, the hardness profile closely follows that
coolant during the hole's generation; this creates very of the residual stress graph; however, instead of a
high levels of friction at the drill margins because of tensile stress at the surface resulting from the by-
inefficiency in the dull drill's cutting lips. Virtually all product of a machining process, the subsurface layer
tooling, even the most sharp, have a finite tip radius here could equally be one of compression.
of approximately 8 fLm, which results in increased If a sharp or new tool is employed with little if
wear that can trans form the surface metallurgy by any flank wear land present (Figure 165) in the
 Q
690
0
'(jl
Q

::0 ~ 345

.,'"'"
.l: 0
~ ",,- //
--...----
--
---
~~ ~
f--

\, / ./
-;'" .",/ --~
:::
"CI
.,
'(jl
c
0
'(jl 345 /
,,~ . J /
/' /
.,...'"
~/
i:.::
c- /'
E
0 690
//
U ,/

o 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

Depth below milled surface (rnrn)
- - - - - Zero wear land

- - - 0.2 rnrn wear land

0.4 mrn wear land

Cutting speed: 60 mlmin Tool: 0100 rnrn fly-cutting face mill
Feed: 127 mmltooth with Caboloy 370 (C-6) carbide
Depth of cut: 1 mm AR: 0° TR: _10° RR: _15° Ind: 10°
Width of cut: 19 rnrn CA: 45° ECEA: 5°
Cutting fluid: dry Peripheral dearance: 8°
Specirnen size: 1.5 rnm x 19 mrn x 108 rnrn

Figure 165. Residual stress in a milled surface of 4340 steel quenched and tempered to 52 HR c [After Field and Kahles 1971.1

machining of a potential work-hardening material, fied heat-affected zone (HAZ) normally occurs, this
then some plastic deformation will be present in the being a product of either (UTM) or (OTM). As the
grains immediately adjacent to the surface. A certain effect of either of these HAZ's conditions diminishes
amount of over-tempering may be associated with further into the workpiece, then the original sub-
this machining operation that extends a Httle way strate condition will predominate. The white layer
into the material's substrate. In this example, as the effect tends to be significantly harder than the
tool's flank wear increases from 0.2 to 0.4 mm during substrate, with typical values for the substrate being
successive in-cut operations, then significant over- 50 HR c, while the UTM can quite easily reach hard-
tempering occurs, with the effect extending consid- nesses of >62 HR c, with a local tensile residual stress
erably further into the substrate, until it diminishes. present. This purely thermal event that occurs
Such significant plastic deformation of the substrate during the EDM process results generally from a
combined with the unstable metallurgical condition non-contact process, occurring as a result of the
may possibly lead to premature failure of a highly extremely rapid heating and cooling of the EDM
stressed component member in service, which might continuous cyclical operation. The white layer is
have a catastrophic influence on an assembly, with strongly dependent upon the energy density, result-
disastrous consequences weIl beyond its original ing from the interplay of two process parameters,
and intended purpose. namely the current magnitude and associated pulse
Electrical dis charge machining (EDM) can ini- duration. As both of these parameters increase, so
tiate a "thermal event" such as white layering and its does the depth of these metaIlurgically altered states.
associated martensitic transformations, which exist The influence of the unwanted by-products of the
as re cast and redeposited layers. Due to the fact EDM process can, on some electrical dis charge
that the re cast layer has a similar metallurgical machines, be minimised by switching on a surface
formation to that of a white layer, the delineation integrity (SI) generator (Figure 166). The advantages
between them is not readily apparent. Beneath this of utilising this SI generator are significant and
white layer the existence of a metaIlurgically modi- include the following:
 Without SI WithSI
- parts machined with a - parts machined with a
standard generator: surface integrity generator:
CARBIDE
-micrographic cross-sections
of workpieces in carbide

STEEL Presence of an affected layer (cracks) Absence ofan affected layer (cracks)

-micrographic cross-sections
of workpieces in steel

Presence of a white layer Absence of a white layer

TITANIUM
-slots machined in titanium
workpieces

Presence of an oxidised area Absence of oxidisation

Figure 166. 5urface integrity of electrical discharge machining (EDM) with/without an 51 generator. [Courtesy of Charmilles Technology.l

• surface integrity of the material improves above, then the consequences are for a poor surface
through areduction of unwanted and undesirable integrity workpiece to result. These problems may
metallurgical effects; include shorter tooling life, through impaired
• using the SI generator, the electrical dis charge mechanical characteristics of the metal; reduced
machine tool allows fine surface finishes? Ra 0.2 resistance to wear, promoting surface fragility on, for
f.1m up to component heights of 100 mm; example, dies and punches; and bad adherence of
conversely an Ra of <0.1 f.1m can be achieved at surface covering, on materials such as titanium
around 5 mm component heights); nitrate (TiN) and titanium carbide (TiC). Other
• output from the whole production process can be metallurgically related surface integrity EDM prob-
increased, while obtaining the equivalent level of lems may include the following:
surface roughness and integrity;
• as a result of utilising the SI generator, the compo- • Thermal damage - during the process a large
nent's parallelism is improved for high work- proportion of energy contained in the spark is
pieces. transmitted to the material being machined,
causing localised heating. The temperature differ-
If the electrical dis charge machine does not have ence inside the material generates constraints,
enhanced features such as the SI generator discussed resulting in surface micro-cracks;
 InClustrial metrology . .

• Corrosion effects - on ce micro-cracking has range of surface topographical effects and sub-
occurred, small quantities of water stagnate in surface integrity modifications, depending upon
these cracks and are not deionised. Hence, the whether the operation was of agende or abusive
properties of these minute reservoirs of water are nature. In Figure 167 the abrasion performance of a
modified and the liquid corrodes the material by conventional grinding wheel consists of a number of
dissolution; variable factors; these include its abrasive, grit size,
• Electrolysis - during the EDM process, an electric grade, structure and bond type. Today, grinding
current passes through the water, which in turn wheels can be manufactured from ultra-hard abra-
increases the quantity of OH- and H 30+ ions, sives, but for brevity and to simply illustrate the
exacerbating the surface integrity still further. problems that occur in grinding operations only the
conventional grinding wheels will be addressed
The use in EDM of enhanced control features, typi- here, to highlight their infiuence in the overall
fied by these SI generators, significandy reduces the grinding process. Prior to this discussion of the rela-
likelihood of white layering and other unwanted tive merits of wheel selection and their anticipated
process-induced surface defects. effect on the ground surface topography/integrity, it
The process of grinding, whether of the surface, is worth iterating that several factors infiuence
cylindrical or centreless variety, can introduce a grinding wheel selection; however, they are outside

Crcsceot wcights (adjustable)

Waviness -
out-of-balance

_l

Deptb ofeut

Grit size of
grinding wheel

NB Cross-feed travel will also iofluence the traosverse roughness of the surface

Figure 167. Influence on the ground surface promoted by both the wheel's grit size and out-of-balance effects, when surface grinding.
the scope of the current topic, but nevertheless are integrity effects in combination with a range of
worthy of amention. These grinding factors are the residual stresses, depending upon whether the
workpiece material, condition of the machine tool, grinding process was gentle or abusive, the latter
wheel speed, work speed and angle of contact regime significantly increasing residual stresses in
between work and wheel. the workpiece surface layers. Under an abusive
A conventional wheel's abrasive make-up tends to regime, the dominant feature will be of a thermal
be of two distinct types - either aluminium oxide or nature in combination with tensile stresses in the
silicon carbide - the former abrasive wheels, being localised surface layers. Equally, if a gentle grinding
softer than those of silicon carbide, are employed for regime occurs with copious amounts of coolant
harder workpiece materials, as new grains are present, then reduced frietion resuIts and the "unit
broken and sharp cutting edges are continuously event" tends toward the mechanieal, promoting a
exposed. The grinding wheel's grit size influences compressive residual stress in the subsurface. If the
the ground surface as illustrated in Figure 167, as grinding conditions utilise more conventional
does the wheel out-of-balance. The wheel out-of- cutting and grinding data, this introduces a thermal
balance in surface grinding causes it to "bounce" event that has residual stress magnitudes approach-
across the ground surface, introducing medium- ing those of the abusive regime, but tend to be of
frequency waviness, the relative surface roughness shallower depth below the surface. The peak residual
being a function of the grit size that is superimposed stress in conventional grinding virtually coincides
onto this waviness. If a fine or small grit is selected, with that from abusive grinding regime and the
then the anticipated ground surface will also be fatigue endurance limits are alm ost identieal. It has
smoother than if a coarser grit had been selected. been reported that there is an inverse relationship
Both an isotropie and anisotropie surface can be between the fatigue limit and peak residual stress.
produced by the action of grinding. Wheel grade In Figure 168 white layering is shown to be
selection is important as this can affect the overall present in the subsurface regions for both surface
strength of the wheel, enabling it to be used for and cylindrical grinding operations. Even if the
either low or high production output. The structure selection of wheel, work and associated cutting data
of the grinding wheel is an indication of the propor- are optimised, then if the wheel is not "self-sharp-
tion of bond to abrasive, an open-structure or ening" and the wheel wear pattern is toward the
porous wheel having around 30% bond, whereas a outer edges, rather than at the centre of the wheel's
closed wheel may only have 10% porosity. Bonds can width, it will "glaze". A glazed wheel will introduce
vary considerably, rubber-bonded wheels being the abusive machining and in its worst condition it can
strongest, whereas shellac bonds are used for fine become "loaded" with totally inefficient cutting,
surface-finishing operations. Generally, a good guide further exacerbating the workpiece surface integrity.
as to the type of wheel and grinding conditions to Therefore under either a glazed or loaded wheel
recommend in obtaining efficient workpiece output local hardness will increase, as will the white
and surface topographies would be as follows: layering tendency, which further degrades the work-
piece surface texture and integrity.
• "harder work, softer wheel" - relating to work-
piece production;
• "larger angle of contact, softer wheel" - relating in Altered materiallayers
the main to either internal or external cylindrical
and centreless grinding operations; In order to gain an overview of the aItered material
• "good rigid machine, softer wheel" - relates to layers (AMLs) that occur in diverse surface and
machine tool rigidity and its associated vibra- subsurface topographieal features ranging from
tional tendencies, which may be superimposed differing metallurgieal processes, mechanieal appli-
onto the workpiece surface; cations and uses, see Table 17, whieh highlights their
• "high wheel speed, softer wheel" - this refers to the particular influence on functional performance.
fact that when wheel speed is high the work In the majority of cases shown in Table 17 the
moves a relatively small distance while the abra- influence of these subsurface defects tends to be of
sive grains pass, hence its grinding forces are low; notable importance, specifically with respect to an
• "high work speed, harder wheel" - when the work abusive regime producing a machined white layer. In
speed is high an abrasive grain has to remove some instances the aItered material zone (AMZ) can
more workpiece material during each pass, which affect component in-service performance in a
increases the grinding forces. variety of ways. For example, where the in-service
tribological situations produce either redeposited or
The resuIt of all of these variables in the grinding recast layers in the surface region, it has been known
process will be to introduce inconsistent surface that such defects will influence wear and affect reli-
 ni:tustrial metrology .

SURFACE GRlNDING ,

-+ -
<;, Whccl
--8 -i3-
wear:~
I Wheel :
Whccl:
out-of- balance '

t
I >~
r
Localised Minimal Localised
white layers distortion white layers

CYLINDRlCAL GRlNDING

I r- >~
Wheel : Feed

- ~-3--

Minimal Localised
distortion white layers

HAZ Heat-affected zone

Wh·Itc ayer

K\,I HAZ

-
- - - Abusive
_ _ Gentle

_.\ - ,
0
/
U
o 0 ·25 0·50

Distance from surface (mm)

Figure 168. Effeet on maehined surfaee integrity du ring surfaee/eylindrieal grinding of high-strength steels and alloys.
 Im
Table 17. The influenee of subsurfaee features on funetion

Surfaee integrity: subsurface features

Function Metallurgy Deformation Deposits Stress

UTM OTM Aust IGA WL Plast Burrs Cracks Tears Tool Redp Res
orWL rev defn and frags matl stress
laps

Wear
Strength
•• • • • •• •• •• •• •• • • ••
0 0
0 0

•• •• • •• •• •• ••
Chemical attack 0 0
Fatigue
Magnetism
• • 0
0 0
0

Bearings
Seals 0• • • •• •• •• •• 0
0 0 0
0
0
Friction
Bonding and adhesion
0
0
•• •• •• •• • •0
0

Forming
• • • • • • • •
Key:.: strong influence on function; 0: some influence on function; (>: possible influence on function.
0

Abbreviations: UTM: untempered martensite; OTM: over-tempered martensite; Aust rev: austenitic reversion; IGA: interangular attack; WL: white layer; Plast defn:
plastic deformation; Tool frags: tool fragments; Redp matl: redeposited material; Res stress: residual stress.
After Griffiths et al. (2001).

ability. This often undetected subsurface condition surfaces and particularly a turned surface, refers to
degrades the functional performance; because they any surface undulations that are regular/periodic
are the product of hard, brittle and unstable layers (anisotropic) that are deemed typical for that surface
with tensile residual stresses present, combined with condition. Such roughness can be considered to be
an acute change to the bulk substrate, they are likely the constant and predictable nature of the surface
to spall (delaminate and break away). Conversely, if and not the random variations and disturbances that
a subsurface feature pro duces severe plastic defor- may be present overlaying the fundamental surface
mation, evidence has shown that in the case of the geometry. Conversely, a machined surface's topo-
die and tool industry some dies benefit from graphy describes the real surface including the
increased life due to this enhanced abrasion resis- texture and any disturbances present; therefore if
tance. Table 17 indicates to the design engineer that there are no surface disturbances then the topog-
simply selecting a production process without an raphy equates to its texture or roughness. Further, if
intimate knowledge of how it is to be manufactured no disturbances occur within the macro/micro-
will affect the subsequent part's performance. textures of this machined surface, then its finish can
Moreover, due regard must be given to the compo- be considered as the quantified description of the
nent's potential subsurface state, as this condition topography defining the basic structure.
will inevitably lead to problems when the part is In Figure 169, if consideration is given to the
utilised in service with its potential reliability being "idealised" turned surface topography, then for
impaired. constant tool no se geometry and an undeformed
chip thickness (often wrongly termed the "depth of
cut" in turning), as the feed per revolution is in-
5.2.5 Machined surface topography creased the surface texture is degraded. The residual
cusps that periodically occur on the turned surface
after the tool's passage along the part are the product
Turning of two phenomena: the so-called "moving step
effect", in conjunction with the associated condition
When a component has been machined this can be of the "emerging diameter". This relationship
considered to be a permanent re cord of the "unit increases the notable height of the cusps with larger
event" and, as such, can be diagnostically investi- feed per revolution and diminishes in height with
gated to obtain invaluable information of the reduced feeds. If proportionally larger feeds per
production process and the data utilised in its revolution are selected, this increases the residual
manufacture. The term "roughness", relating to most influence of the tool nose contact region on the
 15° Plan
approach 40° Point angle
angle with 1.5mm

Undeformed chip thickness "" 3 mm (doc) Cusp height (smalI) =lower Ra

approach 40° Point angle

angle with 1.5 mm
nose radius
Feed
-<J-------

Undeformed chip thickness '" 3 mm (doc)

Cusp height (large) =higher Ra
Figure 169. How feedrate influences the machined cusp/surface roughness value Sm and its effect on the waviness parameter ß q.

surface and as a result heightens the turned cusps, raphy will markedly differ. This topography differ-
promoting a larger recorded value in Ra and greater ence is shown to good effect when turning ferrous
angles for 6.q values. The opposite effect takes place PM compacts as depicted in Figure 170, where two
in the case where feed per revolution is reduced. extremes of cutting insert no se radii are utilised. In
Namely, as smaller tool no se contact region occurs, Figure 170(a) the no se radius is 0.8 mm, whereas in
this results in a smaller cusp height and accom- Figure 170(b) it has been increased to the equivalent
panying Ra; this also gives a shallower 6.q due to the of a 6 mm no se radius, by employing a large button-
partial curvature of the tool nose tending to zero as type of insert geometry. In the case of Figure 170(a),
it approaches tangency with the workpiece axis. The it is visually apparent that the periodicity of the
height, profile shape and periodicity of the cusps surface is indicated by the regularity of these cusps
resulting from the feed and tool geometry signifi- and, despite the fact that a new tooling insert was
cantly influence the measurement and magnitude of employed in this turning operation, the surface
the surface topography, and hence resultant para- shows significant signs of tears, laps and burrs. By
meters. Therefore if the Ra value alone was utilised, way of comparison in Figure 170(b), the surface
it cannot adequately describe the nature of the topography appears significantly smoother in pro-
surface topography condition in any meaningful file, with no noticeable cusps present, although the
way. overall surface topography is marred by similar te ars
If the cutting data is standardised (rotational and so on which could be cause for the part's
speed, feedrate and undeformed chip thickness are potential rejection. This smoother surface due to the
constant) and only the tool nose geometry is significantly larger cutting insert tool no se radius is
changed, then the resulting turned surface topog- not unexpected, as the feedrate was considerably
smaller than the nose curvature of the insert; there- (a) Thrned surface iIIustrating partial tool nose
fore it could blend out and as a result obliterate any geometry, from a 0.8 mm tool nose radiused insert
potential cusps on the turned surface. This tech-
nique of improving the surface by using a large tool
no se geometry has been employed by precision
turners over the years and can be summarised in
terms of the resultant cusp height by the following
relationship:

Therefore
f2
Ra=-
32r

where H = maximum peak-to-valley height, r = tool

no se radius and f = feedrate.
In the literature there are traditional formulae for (b) Identical cutting data as in (a), but now utilising a
predicting both Hand Ra; they are produced by a button-style insert
given feedrate fand tool nose radius r, for rounded
geometry turning inserts. With some cutting inserts
having almost a sharp point, when in effect the no se
radius equates to zero. For finish turning feedrates
in conjunction with a rounded or button insert
(Figure 170b) the relationship for Ra derived above
gives an approximate agreement. Although the value
of Ra is an average by definition, this occurs through
a combination of both feedrates and no se radii,
allowing optimised conditions to be achieved. In
reality, the value obtained from the Ra calculation
predicts a smoother surface topography than actu-
ally occurs in industrial conditions.

Milling
Milling operations represent the main produc- Instrument:JSM-5600
tion processes for the manufacture of aerospace Acc.Voltage: kV 15
Photo mag. x200
and many high-quality precision parts requiring S.E. image
prismatic features to be incorporated into their Vacuum: Pa
design. Basically, a prismatic feature on a component WO: 46
consists of a machined face or facet - not necessarily
at 90° to an adjacent surface - which normally acts Figure 170. SEM photomicrographs of typical turned ferrous P/M
surface topography with constant feeds and speeds, but varying tool
as a location face or datum in aprecision assembly. nose geometries. [(ourtesy of Jeol (UK) Ltd.)
Predominantly, the aerospace industry would tend
to manufacture these prismatic features from
wrought material by utilising one of two machining enabling HSM machining centres to cover similar
strategies: areas even quicker than their counterparts
described in the former case.
face milling - allowing sizeable surface areas to
be generated by large-diameter face mills, In fact, many more industrial processes and
running at conventional speeds and feedrates; applications tend to be utilising the latter approach
2 high-speed milling (HSM) - employing smaller- of HSM. Major production benefits accrue from
diameter end-milling cutters rotating at very high milling surfaces by this technique, not least of which
peripheral speed and moving at higher feedrates, is the minimal cutter deftection - enabling thin-
 InClustrial metrology . .

walled machining to be undertaken - together with milled surface topography will also vary. Returning
negligible subsurface damage inflicted by the HSM to the previous scenario using a standardised
strategy and in many cases enhanced milled surface feedrate, the rotation in combination with the feed
texture. for a given cut-off will change the milled surface
The end- or face-milling process is an interrupted topography, introducing various cusp height effects
cutting operation that imparts an isotropic surface along the surface. Here, the periodic nature of the
topography, as schematically illustrated in Figure surface topography is regular (Sm), but its period-
171. If in Figure 171 (top) stock is removed by icity changes according to whether the surface is
milling, the resultant milled surface exhibits quite a measured at the centre or at some point across the
complex surface topography, due to the "recutting surface; this in turn modifies the relative height
effect" of the cutter's trailing edge as it moves over of the cusps. Conversely, across the milled surface
the surface at a periodic and set speed. This periodic at arbitrary positions denoted in these examples as
surface topography will not be the case if some form "X-X" and "Y-Y" (Figure 171), the topography fluc-
of milling incorporating adaptive control constraint tuates at a predetermined quantifiable interval
(ACC) is employed. The reason for this variation is depending upon where the surface trace was pos i-
that as the cutter progresses over the surface - in, tioned. Milled surfaces with a non-directional or
for example, torque-controlled machining (TCM) - undefined lay, as is the case for the recutting effects
its torque is monitored and the stock height may introduced by either end- or face-milling opera-
vary. The adaptive control system constrains the tions, should not simply have an Ra quoted on
machine tool's feed function, thereby protecting the the engineering drawing. Such design information
cutter from damage, but more importantly from a concerning the milled surface becomes somewhat
surface's viewpoint, the feed varies and hence the meaningless and at best only indicates the worst

-<r-- [ Stock rcmoved

,.,m'•• ,••", - - - - f [7777Z7L2=~==ZZ=ZZZ=ZZ=Z Z=Z =Z Z=ZZ=!'lc{"=:tn

Workpiccc I

Fccd

---
X, C ut-off I V
r
I
Cut-off
'I I
Cusp
hClghts
--~: =

-- ~::::::

• - --- ~~~~~~~~ -- ~:=

Rotation Varying
X, ,Y urface
topographies

~l-
I I I I

~~
I I

Surface topography Surface topography

@x-x @y-y

Figure 171. Milling operations impart an isotropie maehined surfaee topography to the workpieee. [After Whitehouse, 1999.]
 Category (enlarged Examplesof Posslble reasons for variation
profile view) kindof
variation
1 Variations in Unevenness Errors in the guideways of the machine tool or of the
shape Out ofround component
Incorrect clamping of the component

~ Distortion due to hardening

Wear

2 Waviness Waves Eccentric clamping or errors in shape of a cutter

~
Vibrations of the machine, tool or component

3 Grooving Grooves Shape of the cutting edge, feed or supply of

~
the tool

4 Minor imperfections Scores How chips are produced (tear chip, shearing

-
Flakes chip, built-up edge), deformation of the
Arches component at the sand-blasting, serration
(after galvanic treatment)

5 (Cannot be shown Structure Crystallising processes, alterations of

on a picture) surface by chemical influence (e.g. corrosion)
6 (Cannot be shown Lattice Physical and chemical reactions, in the structure
on a picture) structure of of the material, strains and shearing strains
the material in the crystallattice

There is a general overlapping of categories 1 to 4

Figure 172. Typical surfaces obtained by face or peripheral milling operations.

surface roughness that the designer will tolerate, ability to the functional performance of milled
saying nothing about the surface topography or its surfaces in critical and highly stressed component
functional performance. environments. Therefore, "wiper inserts" are some-
In Figure 172 several milled surface topography times incorporated into the face-milling cutter body
categories are indicated, with examples of the kind (Figure 173b), where they are accurately preset to
of topographical variation that might occur and the few micrometres below those of the normal cutting
possible reasons for this disparity. Milling opera- inserts (50-100 fLm). Firstly, these wipers will
tions can be performed in a number of different minimise cusp effects, thereby enhancing the sur-
ways and the resultant milled surface topography is face topography, and secondly, they gently remove
heavily influenced by the numerous varieties in a small depth from the transient milled surface,
cutter insert geometry, plan approach angles and minimising the plastically altered subsurface
tool path cutting data selected, as simply shown in layer.
the few examples depicted in Figure 173. Not only The term spindie camber describes a slight incli-
will the relationship of these interdependent milling nation of the milling spindIe axis normal to that of
factors influence the surface topography but also, as the machine's table, or zero inclination (Figure 174).
was shown previously in Figure 163, the subsurface This camber technique is used to avoid the recutting
integrity is heavily influenced by residual stresses effect on the surface previously mentioned and,
induced by the cutter's passage over the surface additionally, it minimises cutter insert wear, partic-
being milled at conventional cutting data. This ularly at the cutter insert's periphery. In reality, the
problem of subsurface modification is one reason spindIe camber tends to be very slight, generally
for the upsurge in the popularity of HSM strategies, only amounting to between 100 and 300 fLm over a
as the technique imparts minimal influence on the length of 1000 mm. If this camber is converted to
plastic deformation of this subsurface region after angular measurements, the value ranges between 20
milling. It is often recognised that the recutting and 60 seconds of arc, this effect being greatly exag-
effect introduces a certain amount of unpredict- gerated in Figure 174 (top).
 (a) Surfaces obtained with radiused edges True centre line

Facemill
Cusp
height :;:::::~'7-~>-::.-?-?--r'~

Insert radius

(b) Surfaces obtained with plane chamfers

Feed
direction

1000mm

(c) Face milling with wiper blades

e

25 0=100
(d) Surfaces obtained from peripheral milling e~ 20
18
..c 16

~
14
-lli Variation in
12
."
10 0=160 shape,f, caused
.5 9 by spindIe
] 8
7 ./ camber
e 6 ,/ ./ /' 0=200
Ei 5 /' V rD=2&

Cusp height
Q

~
/ /1/ V V 0=315

/ / /1/ V ~
"...
...:;
/ / r//~ v
..
oS 2

~
.5 IIiL ~

rt
c
=
<=
Figure 173. Examples of variations in the surface topography
·2
~
resulting from milled surfaces. 0.01 0.02 0.03 0.04 0.05

SpindIe camber in 1000 mm (mm)

D=cutter diameter (mm)
When machining with a relatively large-diameter
face-milling cutter, a concave surface is generated as Figure 174. Influence of spindle's relative squareness on concavity
shown in Figure 174 (middle): clearly in this case it of the workpiece surface.
is not possible to produce a plane Hat surface. The
surface concavity generated by the spindle camber
depends upon the relationship between the cutter where De = effective diameter of the cutting circle, e
diameter, the width of the workpiece surface and the = width of workpiece surface and q = lOOO.tan e,
depth (mm) of this concavity 1", which can be calcu- where e is the spindle camber.
lated using the Kirchner-Schulz formula as follows: Rather than calculating the surface concavity 1",
a reasonable estimate can be obtained from the
Surface concavity: graph in Figure 174 (bottom), showing the variation
in surface shape for a variety of spindle cambers and
face-milling cutter diameters. These concave surface
modifications introduced by the spindle cambering
effect are never large deviations from the "true" Drilling
plane surface. Even under the extreme conditions of,
for example, a small cutter diameter (100 mm) and By far the most popular machining process is that of
a large q value (50 /Lm), the deviation only amounts drilling holes in components and, apart from the
to 25 /Lm over a workpiece width of 100 mm, this more recent designs based on indexable/solid drills
being weIl within the allowed tolerances for most (Delta and U-drills), where high penetration rates
commercial situations. can be achieved, the twist drill has hardly changed.
The twist drill's basic construction was conceived
during the latter part of the American Civil War
(1864), when Steven Morse perfected its design.

Drill diame er (varies)

I
/'

depth of b ody

web centrality =!w. - wal

lip centrality =(I, -1& ]
point centrality = !c,-~]

* I
chisei edge III.Ist be web-thinned as regrinding occurs - due to web thickening.

: hel ix angle c hosen tor specitic material to be dritt ed.

Figure 175. Drilling geometry and chisei point shape: influence on hole generation.
 Despite the relatively simple design of the most p
popular type of drill, this being the twist drill, its
material rem oval mechanism is quite complex and a
slight digression to give some understanding of how
the hole generation technique occurs will aid in the
interpretation of hole surfaces. Near the bottom of
the drill's flutes where the radii intersect with the
chis el edge (Figure 175), the drill's clearance surfaces
form a cutting rake surface that is highly negative in
geometry. As the centre of the drill is approached,
the tool's action resembles that of a "blunt wedge-
shaped indentor" (Figure 176). Under the chisei edge
the cutting process is inefficient and a region of
severe deformation occurs. These deformed prod-
ucts are extruded, then wiped away into the flutes,
whereupon they intermingle with the main cutting
edge chips (from the lips). The chis el edge in a
conventionally ground twist drill has no "true
point", this being a major source for a hole's dimen-
sional inaccuracy. The twist drill geometry of a
conventional point is shown in Figure 175, together
with associated nomenclature for critical features
and tolerance boundaries. From the relatively
complex geometry and dimensional characteristics x
shown in Figure 175, the obtainable accuracy of
holes generated while drilling is dependent upon
previously grinding the drill geometry to within
prescribed limits. Any variations in geometry and
dimensions, such as dissimilar lip lengths and
angles, or chisei point not centralised, will have a
profound effect on both the hole dimensional accu-
racy and roundness. "Helical wandering" of the drill
as it passes through the component will occur with
these geometrie drill inaccuracies present (Figures
177 and 178). Hole accuracy, particularly the "bell-
mouthing effect" (Figure 177), is minimised by
previously centre-drilling prior to drilling to size.
One of the main causes of a "bell-mouthed hole"
is inconsistency in either the lip lengths/angles
(Figure 175), or employing a straight (conventional)
chisei point, or a combination of both. This "bell-
mouthing" at workpiece surface entry by the drill is
attributed to the chis el point. This effect is produced
by a line contact as the point initially touches the Figure 176. Cutting force generation and the effect that thrust and
torque play in hole quality.
component surface, causing it to "walk" until the
drill feed/penetration stabilises at the outer corners
(margin) ente ring the part and guiding the drill
(Figures 177 and 178). Such degraded hole effects are metrical and the cutting load and torque on the lips
exacerbated by longer series drills because of the vary (Figure 176). Galloway defined the drilled
"rigidity rule", which states that cutter rigidity hole slope angle cf> and hence the hole's subsequent
decreases by the square of the distance. Namely, if a roundness and circularity in the following manner:
drill is twice as long, then it is four times less rigid.
It follows that the greater the drill penetration into Slope in drilled hole:
the workpiece, the progressively larger the deflec-
tion and the further from the true axis will be the
drill's subsequent path. This drill deflection is com- 3 R(
(cf»=2 1y 1- (ktank)1
I)
pounded when the drill point geometry is not sym-
 (a) Relative lip height difference (i.e., 0.125 mm), (b) Relative lip height difference (i.e., <0.25 mm),
without centre-drilled hole with previously centre-drilled hole

0-5
0·5
revs 5-7

0-5

before contoct
contoct
0·5

Figure 177. The roundness of a hole is influenced by the initial drill's progression into the part, causing "bell-mouthing"
[After Galloway, 1957].

Drill: "A" "lr' "D'·

y

10o"um

1OOtum x

Cutting conditions: speed In): 173 rpm fee d IfI = 0·012 mm/rev depth Id) =0·5 mm

La r ge-=:::::Jf--_ _[_C_H_IS_E_L_EO_u_E_LE_N_u_TH--,-l_ _ _ _--1Jmoll

12·4.mm) 10·7 mm)

Figure 178. Influence played by an increasingly larger chisei point on the resulting Lissajous figures of drill deflection, as hole depth is
generated. [After Onikura et al., 1984.]
 where I = length of deftected element, R = ratio of 5.2.6 Machined roundness
the transverse re action at the drill point, T = thrust
force, I = system's "moment of inertia" and k =
...J(T/E)I. Harmonie departures from roundness
As this slope suggests, the error is initiated when and surfaee texture eorreetion
the chisel edge begins to penetrate into the work-
piece, and unless the feed is discontinued or the In the previous seetion it was mentioned that oper-
error is corrected in some way, the magnitude of ations such as drilling can impart a range of hole-
deftection will increase as drill penetration con- gene rating mechanisms that could be the cause for
tinues. Typically, the magnitude of deftection can part rejection, such as roundness, cylindricity and
reach up to 60 f.lm under exaggerated conditions and surface topography errors. In this seetion roundness
this could be the source of component rejection. will be the main point of discussion, as it will be seen
Due to the drill cutting edges and lip lengths that machining operations and their accompanying
(Figure 175) not being ground correctly, the cutting tool geometries can seriously inftuence the resultant
forces - thrust and torque (Figure 176) - will be both shape of the roundness profile. If, for example, a drill
unbalanced and uneven. This imbalance can lead to has an asymmetrie cutting action (unbalanced),
the drill spiralling down as it penetrates the work- then this condition is likely to impart changes to the
piece, causing the surface profile to exhibit the generated hole as penetration continues through the
dassical effect of a "saw-toothed profile" on the part. In Figure 179(a) just such a situation is
resultant surface topography. This "saw-tooth effect" depicted, where at the top an almost three-Iobed
on the topography is the result of the partial geom- roundness profile is shown; this becomes more
etry of the lip corner and its adjacent margin, regular in the middle, but the lobes are partially
repeating itself at the periodicity of the feedrate, rotated through an angle, and at the bottom the
which could cause problems if the hole is subjected polar plot has become almost round. This improved
to cydical stresses in service. roundness at deeper depths of penetration is prob-
The chisei edge has been shown to playa crucial ably the result of increased support by the drill's
role in drilling hole quality, not least of which is the margin as it progresses through the workpiece.
result of the inefficient behaviour of the chis el In order to improve the drilled hole quality in a
point's extruding operation as the hole is generated. number of significant ways, such as in its apparent
If the chis el point on the twist drill with respect to roundness, surface texture and integrity, together
its diameter is relatively large, then the Lissajous with the correction for any geometric errors present,
figures, as depicted in Figure 178, will show large then reaming may well be employed. Drilling, being
ftuctuations in the drill's axis of rotation; conversely generally considered to an abusive regime, will
at a smaller chisei point size, this motional behav- inevitably introduce more undesirable errors during
iour is significantly smaller. One of the reasons why hole generation than the complementary and more
the chis el point length and its profile play such a gende reaming operation. The advantage of utilising
major role in producing "bell-mouthing" of the hole reamers after the drilling process is shown to good
at entry is the result of much greater thrust forces, effect in Figure 179(b), where several different
with a proportional increase in the smaller torque reamers are utilised on identical drilled holes,
values (Figure 176). When the forces on the drill with the roundness being assessed at the same
increase as a result of the higher cutting force and vertical "slice level". On the left-hand illustration,
torque levels, this tends to cause the twist drill to a six-edged carbide-tipped machine reamer of the
unwind slightly. This subsidiary drill behaviour conventional type was firmly supported in an appro-
introduces the potential for dynamic instability, priate adaptor that directly fitted into the spindie
resulting in an increased tendency toward vibra- nose taper. This positive location and reamer
tion and additional harmonie out-of-roundness restraint mean that the reamer will introduce into
effects, coupled to the likelihood of a degraded the hole any inaccuracies apparent in the machine's
drilled surface topography. If a through-drilled hole spindIe, although even here the accuracy is a signif-
is required rather than a "blind hole", then there is icant improvement - five times better in fact - than
the potential for the generation of "bell-mouthing" the previously drilled hole (Figure 179a). If the
at the exit point of the workpiece, resulting from the reamer is allowed to "ftoat" (having lateral/radial
so-called "trepanning effect" as the drill breaks play) in the machine's spindie, any spindie errors
through the part's underside. here will not be superimposed onto the reamed
surface and the reamer, with its longer cutting edges
than the previous drill, will correct for many of the
geometrie inaccuracies introduced during the
drilling operation. In Figure 179(b, middle) a "ftoat-
 (a) Typieal roundness error (i.e., "displayed profiles") produced by asymmetrie cutting action.

Hole position/s:

(b) Roundness profiles produced by a six-edged carbide-tipped helieal reamer.

Inaccurate reaming. Floating reamer- roller drive. Acc rately reamed hole.

Figure 179. The effect of harmonie departures from roundness change with different drilling hole depths or reamer types.

ing reamer" replaeed the eonventional type and In some maehining eases, the errors introdueed
was driven through the drilled hole at too high a by the drilling proeess are anticipated to be quite
penetration rate, introducing sixth upr harmonie large, as shown by the displayed profiles in Figure
equating to the equispaeed six-edged reamer. In 180 (bottom left) , illustrating the relative radial
Figure 179(b, right), an identieal re am er was used to movement in the displayed profiles and their
ream a similar hole, but at a slower penetration rate, associated harmonies. Henee, it is often prudent to
producing an aeeurately reamed hole with minimal introduee aboring operation prior to reaming
harmonie disturbanee. (Figure 180). After previously drilling a slightly
 PRODUCTION OPERATION
DISPLAYED PROFILES I

Floating
reamer

Floating reamed hole-

circularity and hole's Boring
dimensional size improved bar

Twist
drill
Bored hole-circularity and
geometrie alignment errors
corrected

l-
I
I
,I

Harmonie departures from

roundness of drilled hole at Bell-mouthing at the entrance and exit points-
different slice levels with helieal wandering

Figure 180. Exaggerated errors caused by incorrect drill geometry and the technique for its subsequent correction.
 smaller predetermined diameter hole, aboring oper- • thermal growth effects - changing the both the
ation will then correct any abnormalities (roundness spindle's growth (axially) and modifications of an
and helical wandering effects) as indicated in Figure elastie nature to the relative axis orthogonalities
180 (left, towards the top). Aboring operation does of the maehine tool;
not follow the hole's contour and, as such, will elimi- • working clearances and motor drive configura-
nate these drill-induced errors by machining away. tions - these are neeessary to allow for relative
Finally, the reaming operation will impart the fin- thermal growth and eomponent "running fits"
ished dimensional size to the hole and simultane- within the spindie assembly, which are com-
ously improve both the circularity and surface pounded by arrangement of the motor drive
texture (see polar plot in Figure 180 (top left). system.
In order to improve the boring process still
further, adjustable twin- or triple-boring heads with This latter feature of spin dIe inaeeuraey is present
indexable inserts can be utilised. These special- in headstocks on either lathes or turning centres, as
purpose multi-boring tools balance out the cutting illustrated in Figure 181. Direet-driven headstocks
forces associated with the boring operation and, in for turning centres offer eonsiderable advantages
so doing, improve surface text ure and hole accuracy, over their belt-driven counterparts. The working
together with repeatability and consistency of bored clearanees and drive eonsiderations found on
holes. In fact, triple-boring heads in partieular can conventional belt -drive maehines suffer from a
sometimes eliminate the need for the succeeding combination of the effeets of spin dIe motor drive
reaming operation, giving a significant saving in the plus associated drive belts on these clearanees. This
value-added machining costs. headstoek's motor drive arrangement eauses an
Reaming technology has radically improved in undulating and irregular harmonie rotational
recent years, reamer design and development con- motion on the work-holding equipment, whieh is
sisting of asymmetrie replaceable blades having a then translated onto the resultant roundness and
floating action with accompanying burnishing surface texture of the workpiece as turning opera-
pads for cutter edge support. Reamers of this level of tions occur. The influenee of this irregular harmonie
sophistieation are typieally employed in the high- rotational belt-driven rotation can be gained from
.volume end of production for automotive engine the schematic representation shown in Figure 181,
part Irtanufacture, such as when reaming out the high where a "tumbling three-Iobed harmonie" shape is
silicou:aluminium alloy camshaft bearing supports. reprodueed on the workpieee. The irregular but
In this production situation the high silicon content periodic nature of the rotational action of the head-
in the aluminium alloy produces ideal (small) grain- stock is reproduced on the workpiece by aseries of
refining effects. As a result of additions of silicon, it kinematie combinations of headstock rotation and
can introduce hard and abrasive inclusions, requir- linear motion supplied by the eutting too1. If a
ing either a cemented carbide or synthetic diamond direct-drive headstoek configuration is used instead
reamer bhlde to be fitted, with guidance through the of the conventional belt-drive variety, then there is
camshaft's location length provided by support from virtually no harmonie influence on the part and
the bearing pads. Today, "hard reaming" is emerging more consistent turned eomponents will result, in
as an important technique in some industries. terms of their geometrical and linear dimensions.
Hard reaming allows components made from mater- Direet-drive systems offer other benefits, not least
ials such as austempered ductile iron (ADI) to be of whieh are lower maintenanee problems, as there
machined directly after initial heat treatment, intro- is no need to periodically adjust belt tension and
ducing not only an appreciable cost saving but also the bearings operate under a more eonsistent and
enhancing the surface texture and roundness levels uniform loading, with better thermal growth char-
of the finished part. Moreover, it has been alluded to aeteristies, resulting in high er spindie aecuraey and
previously that the machine tool's spin dIe and to a improved damping capacity. These positive merits in
lesser extent its overall structural configuration can using a direct -drive spindie promote significant
impart significant errors to the manufactured com- improvements in the overall harmonie roundness
ponent, due to a variety of machine tool-induced throughout the eomplete length of larger turned
inaccuracies. These can be identified as: parts, with less variation in the medium-to-Iarge
frequency surface texture components, resulting
• spindie imbalance - introducing dynamie lower- from this turning operation.
frequency harmonics on the part; If one ignores the fact that either a conventional
• cutterforces - that dynamieally affect the machin- or direct-drive system might be incorporated into
ing process, causing aseries of high-frequency the headstock's design, then the problems associated
harmonics to be superimposed on the lower- with the former type can be ignored. Instead it is
frequency harmonic resulting from imbalance; possible to eoncentrate purelyon the rotational
In(justrial metrology

(a) SpindIe

v v v
(b) SpindIe plus motor and drive

(e) ResuItant

(d) The "tumbIing harmonie effeet" of a beIt-driven headstoek on a turned eomponent

(e) The "harmonies" are virtually eIiminated if direet-drive spindIes are employed

Figure 181. Improvement in roundness quality, via "harmonie suppression" oeeurs if direet-drive spindies are utilised.
effect of the workpiece, allowing the problem to be E.I = flexural stiffness (I = cross-sectional moment
somewhat simplified. In Figure 182 the overall of inertia), Mt = boring bar mass and L = length of
machine-workpiece-tool system can be isolated to cantilever.
consider the simple effect of a tool that is either Therefore, for large length-to-diameter ratios,
inadequately supported or the unlikely occurrence the boring bar's rigidity will influence either the
of too small a cross-section making it under- harmonics of the hole profile or its surface texture.
strength. The main cutting force in turning opera- If the boring tool is too long and unsupported by
tions is the tangential force, resulting from several either bearing/burnishing pads, or via additional
factors, such as: end support - as is normally the case for "line-
boring" operations - then geometrie hole and form
• resistance to rotation - caused by the workpiece deviations will inevitably occur.
material's shear strength; If moderate bar overhangs are utilised in the hole
• undeJormed chip thickness - resulting from the generation process, then the operation can correct
radial depth of cut selected; hole deviations resulting from either the manu-
• orientation oJ cutter rake angle geometry - this facturing process (such as cored holes in castings)
being a combination of either a positive, neutral or drilled hole deviations (promoted by a drilling
or negative rake angle, plus the effect of tool edge operation). In Figure 183 a schematic representation
preparation (if any) and the shape and size of the of the helieal wandering of the drilled hole is
tool no se radius; depicted, along with the correction for geometrie
• Jeedrate - the feedrate, in combination with the error resulting from boring hole generation.
depth of cut, will heavily influence the size of the Notably, as the drill progresses through the work-
effective chip thickness and playa dominant role piece material, minute variations in its geometry
in affecting the resultant turned surface texture. cause the drill to helically wander, producing a
regular but undesirable eccentricity to the hole,
In the top diagram of Figure 182 the tangential whieh needs to be corrected by another operation
force is simplistieally shown contacting the cutting such as boring. This hole correction is necessary
insert at the point. This cutting force application because the drill's cent re line follows the path as
causes a large bending moment to occur at the pivot, indicated" "visiting" the four quadrant points as it
or fulcrum point, as shown. The resultant dynamic spirally progresses through the part. Hole eccen-
action is represented in an exaggerated form in the tricity along with harmonie departures from round-
lower diagram of Figure 182, where the tool is elas- ness can be excessive if lip lengths and drill point
tically deflected in a downward manner by this angles are off-centre. Moreover, the combination of
bending moment. As the resistance to deflection drilling faults pro duces drill out-of-balance forces;
increases with the tool's downward direction, this these can result in an oversized hole, this effect being
causes increased pressure from the inherent tool exacerbated by high-volume production demands
body mechanical strength, enabling a certain degree and necessitating significantly greater drill penetra-
of recovery and hence a partial upward motion of tion rates. The cross-hatched circular areas represent
the too1. This cydieal upward and downward tool the excess stock material removed in the passage of
point motion is repeated at a periodic medium the boring bar along the hole's length, as the boring
frequency, causing sinusoidal motional effects to be insert corrects for harmonic departures from round-
reproduced on the turned surface. High-frequency ness. In this case, if the boring bar is relatively rigid
harmonie behaviour can be superimposed onto the it will not be unduly affected by the variation in
medium-frequency harmonies and can be shown to material wall thickness as it bores out the eccen-
good effect in the power spectrum analysis of tricity caused by the drill's wandering action.
harmonie behaviour during machining. This subject
will be discussed later in a relevant seetion in this
chapter.
Boring operations tend to be somewhat less Cutting forces and tool geometry affects
rigid than turning operations because the rigidity on roundness
decreases by the cube of the distance, as the
following equation predicts: The cutting forces resulting from a combination of
the material's shear strength, undeformed chip thick-
'IT~ 3 EI ness, tool geometry and accompanying no se radius
Jo = U(Mt + O.23Mb ) in turning operations have a significant affect on the
harmonie departures from roundness of the compo-
where Jo = normal force acting on the "free end" of nent. In order to show the effect of these variables in
the cantilever, Mb = modulus of elasticity of the bar, the cutting generation process and to simplify the
 Tangential
force

/
-----_--t---_
Cutting insert
// I

I
-t----- -
\
+-- I

I
~~-'------------1~r

\" I
/
Stock to be removed
Tool post

"True circ\e"

"True circ\e"

1- _ _ _ _ _ _ +_ _ -__ _
I
I

I
~
Radial
force
r-----l_Deflection

_::",:\~~§S:JL_ _t-_/V_i_b_ra_t_io_n_ _ _ _ _ _--,

-c

Harmonie behaviour-resulting
from force variation henee defleetion
and vibration

Figure 182. Harmonie departures from roundness resulting from rigidity/damping effeets while turning.
 "True" diameter-after boring operation

Zone of eccentricity

Harmonie departures
from roundness

Length of machined cut

Line of bored path

Boring insert

Figure 183. Harmonie and geometrie correetions by aboring operation previously eaused by the helieal drift of the drill's path through the
workpieee.

discussion, only external-diameter turning opera- piece speeds that can be employed, but this is at the
tions will be discussed. In Figure 184 a typical expense of less rigidity within the work-holding
precision turning operation is shown partially arrangement.
completed, with the "light-turning and finishing" As the orthogonally oriented insert (zero plan
tool progressing along the part. The turning situation approach angle) turns along the part, a "moving
here shows that a long slender part is held in some step" is seen to be present as the "emerging dia-
work-holding device such as a collet, or a chuck at meter" occurs to the desired dimensional size.
the headstock end, with additional support supplied However, if a very high quality part is required, then
by the "dead centre" held in the tailstock. The general it is necessary to look at the turning operation more
trend today is to fit a "rotating centre" in the tail- criticallyas some unexpected and unwanted features
stock, but this can introduce its own eccentric error are present in the final part. As the turning insert has
into the turning process. Rotating centres have an orthogonal orientation to the axis of rotation of
become popular because of the high rotational work- the part, it might be thought that no radial force
component occurs, but this is not the case as the tool component eccentricity/runout. On many sophisti-
nose radius can introduce a radial force that can cated turning centres "programmable-steadies" can
affect the turned surface. The radial force has little be employed to overcome the problem of turning
effect onthe part's harmonics elose to the support long length-to-diameter ratios, where there is a
provided by the tailstock, as shown by the cross- significant tendency to introduce negative out-of-
sectional harmonie and surface texture effect indi- roundness effects into the production process. If the
cated in section "C-C" (Figure 184). As the tool turning centre is equipped with twin turrets, then
progresses along the part the influence of the contri- the technique of "balanced turning", using a tool
bution of the tailstock's support lessens and hence situated in the top and bottom turret with one
the effect of the radial force component increases, as cutting edge slightly ahead of the other, virtually
shown in section "B-B" and to a greater extent in eliminates the radial force affects.
section ''A-X'. Here, the harmonic departures from If the tool is inadvertently set either too 10w or
roundness is significant and has been recognised for high then this will cause either "barrelling" along the
many years by precision turners. Such experienced part, or the "candlestick effect", as it progressively
machinists fit either a "fixed-steady" or the more narrows toward its centre, before increasing in size
preferable "moving-steady" elose to the tool cutting in the same manner as the tool travels along the
zone - on the opposite side of the workpiece - to workpiece's length. Similar effects can occur if the
counteract this radial force problem and to minimise tailstock is not directly aligned to coincide with the

Rotation Transient surface

"Moving step" A B c

-+---~-
Headstock
end
Feed
<:J~--

DISPLAYED PROFILES
- exhibiting harmonies - exhibiting large - some harmonie - indicating
andRa harmonie departures departures from little departures
from roundness roundness from roundness

Displayed profile (LSC)

taken at sections: A -A B-B C-C

Figure 184. Effect of a combination of several and factors influencing roundness, including lack of support and a non-integral headstock
spindie drive system.
headstocks centre. These setting-induced errors able in the process will be the shape of the cutting
compound the problem occurring as a result of insert. The forces significantly vary in magnitude
the inadequately supported cutting process. The due to the plan approach angle. In the case of the
machine tool alignment problem can be simply orthogonal insert (0°) plan approach angle the axial
rectified by using calibration artefact-based or laser- force dominates, this being associated directly with
based techniques, but this topic is outside the scope the feedrate. A very small radial force is present,
of the present discussion. resulting from the small nose radius on the insert,
Cutting forces can be directly linked to the geo- and its effect on the harmonie trace (A-A) is virtu-
metrie shape of the cutting insert, this effect being ally negligible. If a triangular cutting insert geom-
illustrated in Figure 185; in these cutting force repre- etry is utilised, in this case with a 15° plan approach
sentations the tangential or cutting force is not angle, then there is a slight reduction in the axial
considered in the diagrams. If the overall cutting force component and a corresponding increase in its
conditions remain the same - rotational speed, radial counterpart. This minor increase in radial
feedrate, undeformed chip thickness, workpiece force in combination with a slightly longer cutting
material and insert rake angle - then the only vari- edge in contact with the workpiece's transient
Displayed profiles at sections but in the same position on part:

A-A B-B c-c D-D

A C

I
I
I
,
I
- - -,- - ----1-
,
I
I
I I

I I

Feed
<J----
Curved = Plan
Approach
Insert shape: Angle
Weak Strong
~~----------------------------~~

~--------------------------------~~
Less prone to vibration Vibrational tendencies
NB Tangential, or main cutting force not shown

Figure 185. The effect that insert shape and its approach angle has in influencing the cutting forces, hence harmonie departures from roundness.
surface (the workpiece surface to be removed during the result of two principle factors: firstly, the length
the next rotation) leads to a marginal increase in of the transient surface has increased; and secondly,
harmonics on the polar trace (B-B). As the obliquity as a result of this first condition, the part rigidity is
of the insert's plan approach angle increases, as compromised, leading to this exacerbated workpiece
shown by the square-shaped cutting insert inclined roundness.
at an angle of 45°, then the axial and radial force If a round insert geometry is selected, this leads
components equalise. The considerable radial force to a vast increase in the radial force component,
component in this case has a significant effect on the which in turn causes significant harmonic out-of-
polar plot as shown in C-C, where the harmonics roundness in the resulting polar plot (D-D). Here,
have increased along with a notable vibration the polar plot shows major sinusoidal tool motion
tendency that is superimposed onto the primary (see Figure 182, lower diagrammatic representation)
harmonic. An increase in vibration during cutting is with a significant increase in vibration present on

(a) Forces acting on a round geometry insert at a variety of depths of cut: constant feedrate
(i) (ii) (iii)

Resultant
force
(mean) Radial
force

Fine depth of cut Medium depth of cut Roughing depth of cut

Cutting Constant feedrate

insert
~~-------------------

(b) Harmonie departures from roundness on machined component resulting from differing cut depths
Sm all harmonie effect Large harmonie effect

I I

-+- -+- I
-+- I

Finishing cut Medium cut Roughing cut

Figure 186. Harmonie departures from roundness resulting from the radial cutting force pressure on the component.
 Im
this trace. The reason for this notable increase in square rule", which states: milling cutter rigidity
harmonics using round inserts is the product of decreases by the square of the distance from its
several factors. The transient surface in contact with holder. In reality what this is attempting to infer is
the tool has markedly increased and the plan that if a cutter was originally 50 mm long and a
approach angle at the tangency point is at maxi- cutter of twice this length is fitted instead (100 mm
mum; the combination of these two factors leads to long), then the rigidity of the complete assembly
a momentous deterioration in workpiece rigidity will be four times less rigid. This lack of milling
and, as a result, vibration especially is increased. cutter rigidity will cause a number of unwanted
The problem associated with the harmonie effects to appear on the final part feature. Cutter
behaviour of cutting insert geometry is of less deflection will introduce distortion to a square-
importance for large stock removal operations; milled shoulder and harmonie variation, degrading
roughing cuts as a strong insert can be employed, the departure from roundness of the workpiece, as
hence the use of either square or round inserts. If illustrated in Figure 187. In order to introduce
vibration is a potential problem, then the light minimal changes to the milled profile cutter lengths
turning and facing insert of the "trapezoid geom- need to be kept to aminimum, conducive with
etry" shown in this case, with a 0° plan approach correct operational practices.
angle, may be the production solution. In order to fully appreciate the significance of
The round geometry insert is worthy of closer the schematic diagram shown in Figure 188,
investigation and a more detailed sketch of its featuring milling by circular interpolation and its
effect on the harmonie machining behaviour is indi- relationship to the circular feature produced, a slight
cated in Figure 186. Here, identical rotational speed, digression into basic machine tooI-induced errors is
workpiece material but differing undeformed chip necessary. Machine tools that are typified by the
thicknesses (depth of cuts) are utilised to isolate popular three-axis vertical machining centre config-
variability in the machining process. In Figure uration of conventional orthogonally orientated
186(ai) a small undeformed chip thickness is used, slideways (square to one another) can introduce
creating a significant radial force in combination considerable error into the quality of the final part.
with a minute axial force component, leading to the It has been established that three orthogonal slide-
polar plot shown in Figure 186(b, for a "finishing ways - X- and Y-axes in the horizontal plane,
cut"). The argument put forward above, concerning together with the Z-axis in the vertical plane - can
the influence of the large radial force component introduce up to 21 kinematic errors into the cutting
leading to poor harmonics in Figure 185, is not process. The kinematics are quite complex for any
contradicted here, because the cut depth is very machine tool having the ability to move all its
smalI, hence transient surface contact will also be of axes simultaneously, but these small though im-
litde influence. portant errors can be simplistically said to occur as
To give an impression of the influence that the cut a result of:
depth plays in the magnitude of cutting component
forces, the following discussion needs consideration. • six linear motions - produced by displacement of
As the undeformed chip thickness increases from the forward and backward motion of the X-, Y-
the left in Figure 186(ai, aii to aiii), then the are of and Z-axes slideway movements, introducing
contact along the transient surface also proportion- particular non-linearities into slideway posi-
ally increases. This larger effective cut length creates tioning;
a significant departure from roundness harmonie • three rotational motions - yaw, pitch and roll
behaviour on the polar plot, as depicted in Figure on each axis. Yaw is the side-to-side "crabbing
186(b, for a "roughing cut"). motion" along the slideway. Pitch occurs through
waviness in the slideway, introducing a backward
and forward rocking (pitching) action, normal to
Interpolation and its effect on harmonics the slideway, as the moving element traverses
along the axis. Roll may be introduced by two
In CNC milling operations circular features on pris- adjacent ways on the slideway not being coinci-
matic components such as bosses can be produced dent (laying in the same respective plane), causing
in a number of ways, most notably by milling using an upward and downward pivoting action with
the circular interpolation function. This CNC func- respect to the "line-of-sight" along the axis, as the
tion allows accurate circular control of two slideways moving element traverses along its length;
simultaneously, while the cutter mills around the • three squareness errors - these errors occur due
part, as illustrated in Figures 187 and 188. Cutter to the fact that each axis may not be normal
rigidity plays an important role in the quality of the (square) to one another.
final machined feature, being based on the "rigidity
 Constant pull-stud force

Spindie
taper

Gauge line
Milling
Milling cutter
cutter -deflected
-no deflection
e
e
0::>
lI'l
e Milling
e
0::>
cutter
:: -undistorted

m~,
VI

I
I
Workpiece
distortion

+-
I

Circular interpolated profile

Circular interpolated profile

-indicating harmonie oscillation
with a superimposed vibrational
signature

Figure 187. Exaggerated effect of cutter length on the resultant circular interpolated profile on the workpiece.

These kinematic machine errors can be appre- campensatian system have been shown to extensively
ciably reduced by the application of calibration reduce the effects of the variety of errors that can be
through laser-based techniques, or to a lesser extent present on the machine tool, but on ce again this
via baUbar artefact-based methods. These machine topic is outside the scope of the present discussion.
tool calibration techniques are outside the scope of Considering the circular interpolation of milled
the present discussionj the same can be said of the profiles as shown in Figure 188, the departures from
thermaUy induced errors and how they can also roundness of the workpiece is a function of the
influence the machined part surface and profiling previously discussed kinematic machine-induced
qualities. Moreover, errar-mapping techniques and and thermally induced errors, together with load-
in-process control by an associated dynamic errar induced errors. The diagrammatic representation
 True circular
path Tolerance of linear
approximation to
programmed path

Yaxis ~
Circulation interpolation
(Down-cut milling)

"-

Rotatio. ~
\
_+-____ -~;.t-C: =-! X ni,

I
I
/' "\
\~ -
True circular
-Yaxis V Servo-lag

Actual profile
path

Backlash Y axis
Servo-spike

Figure 188. Generated circular interpolation exaggerating the calibration/machine tool errors associated with climb milling.

shown in Figure 188 indicates that several of these Servo-spikes result from areversal of one of the
errors on the circular profile of the workpiece are axes at this angular position and a motor power
present. As the simultaneous motions of the two axes surge (spike) occurs with a corresponding local-
occur (to produce the circular feature), then by ised slack here, as take-up begins (see Figure 188
increasing the interpolation speed for the cutter the at the quadrant points on the circular feature);
roundness will degrade; this is the result of several • backlash - this is present here in some form at
interrelated factors, some of the most notable being: the axis reversal position, because of the forward
and backward motion at the axis transition
• servo-spikes - these occur at the axis transition points, resulting in "play" (backlash) in the ball-
points at their respective 90° angular intervals. screw (attached to and controlling the slideway
 motion). This backlash is present despite the fact (in-cut) until it becomes completely worn out. In
that a recirculating ballscrew on each axis has Figure 189(b, top) the surface profile appears as
been utilised that should have previously been aseries of periodie and regular machined cusps,
pre-loaded, which was meant to negate such indieating that an efficient turning operation has
backlash problems; occurred, whieh can be confirmed by the power
• servo-errors - when both axes are moving simul- spectrum in Figure 189(c). This turning operation's
taneously, their respective linear speeds should power spectrum, as expected, indieates distinct
have been matched, allowing either a perfect frequencies, the fundamental frequency relating to
circular feature or partial arc to be reproduced. the feed spacing of the cusps and the harmonies to
If any non-synchronised motion occurs (servo- the non-sinusoidal tool geometry - its shape. After
mismatch) between these two axes, then an ellip- successive passes along consecutive workpieces the
tical profile normally set at a 45° angle would tool begins to show signs of localised insert flank
occur. If the contouring interpolation is changed and face damage, with the ratio of the height of these
from a clockwise to an anti-clockwise circular harmonics to that of the fundamental frequency
feed motion, then the angular elliptical profile increasing. This harmonic increase is a direct result
shape will "mirror-image" (flip) that of the oppo- of the imposition of the insert's imprint in the
site profile; turned surface. The increased harmonie activity
• squareness - if orthogonality (squareness) is not here occurs due to the wear scars that correspond-
maintained between the two axes, then this will ingly occur on the insert's cutting edges; these are
result in yet another similar milled angular ellip- then faithfully reproduced on the surface topo-
tical profile (as discussed above, for servo-mis- graphy. Moreover, at this position the baseline of the
match), but the difference here is that the shape fundamental frequency can be expected to rise; this
orientation does not change (that is, reverse its increase will probably be because of random effects
shape) if either clockwise or anti-clockwise feed due to the formation of arbitrary chips and the
motion is applied. turned surface exhibiting micro-fracturing. If it is
adjudged that these surface random effects of
Considerably more factors can affect a milled machining are significant, it is probably a more
circular interpolated shape and many of these appropriate strategy to consider their examination
machine tool-induced errors can be diagnostically by utilising the auto-correlation function.
interrogated by using dynamie artefacts, such as the In Figure 190 (right) harmonie information
ballbar. Such instrumentation can find the sources within the power spectrum is depieted, in this case
of error, list their respective magnitudes and apply to the left-hand side of the fundamental frequency
corrections that can be fed into the CNC controller (feed). Here, periodicities occur which are of consid-
to nullify the errors, enabling circular contouring erably greater wavelengths than that of the feed.
milling performance to be significantly enhanced as These harmonics are the result of machine tool-
a result. induced problems, such as bearing chatter and
slideway errors, together with other unwanted
effects. The position of the harmonies related to the
bearing problems appear closest to that of the
5.2.7 Power spectrum analysis of fundamental frequency, whereas the very long wave-
machined surfaces lengths, for example, can be attributed to form errors
in the slideway, or perhaps are the result of struc-
In many situations of a stochastic (variable) tural distortion of the machine tool itself.
machining nature, the power spectrum is a useful A major advantage of using the power spectrum
aid in process control monitoring of the cutting as a diagnostic aid is that it can separate out the
capabilities and gives a good interpretation of the process-related problems, as depicted by the power
anticipated surface topography. Once again the spectrum diagram shown at the bottom left-hand
turning operation, because of its relative simplicity side of Figure 190. Hence, to the left of the funda-
of a single-point machining operation, will be used mental frequency (feed) in this diagram, in region
to show how the power spectrum can be employed "/\', the harmonics of the machine tool problems can
to identify tribologieal factors such as tool wear and be isolated out; and to the right of region "B" those
how this influences the surface topography of the of the feed frequency can be established, with the
workpiece. In Figure 189 the main feature here is workpiece material properties occurring displaced
that it is possible to identify and then quantify tool even further to the right in region "C". The separa-
wear by an examination of the power spectrum. To tion into discrete power spectrum frequencies can
the left-hand side of Figure 189(a) a tool is shown be established, although such diagnostic interpreta-
when new, then it progressively degrades with time tion cannot be readily detected in surface profiles at
 (a) New tool (b) Surfaee profile (e) P(w)

~-----~__~'__'_~-L__~____~

Stightly worn
toot
~-----l~_ _F_~LI ____L -_ _ ~ ~_e~
___

~ A+ 1\ I\j

~-
o A+ 1\: 1\
I
+
+
I
I
+ I
I
I A~ ______~A~____~~
Badtyworn
toot

:I
( l I<~~~~~

~
+
I II<~~~ Wearscars
Surface micro-fracture

I
------c:::::- Process
! ~
~.
BuHt-up edge I I Feed, wear scars and
+ microfracture I I surface micro-fracture
~ Sampling length ~
(Long) .... Frequency.... (Short)

Figure 189. Power speetrum analysis harmonies for a maehined surfaee. [After Whitehouse, 1997.]
 Bearing chatter
surface profile

~ Feed w
Freq
I Slideway error
I

~ Feed w
Long-wavelength summed
land scaled

~ Feed
Subharmonies w
Added roughness

..
I

P(w)
I Region A-Machine tool
I Region B-Process
I Region C-Material properties
I
I

Figure 190. Power speetrum analysis, sub-harmonie and total speetrum. [After Whitehouse, 1997.]

this initial stage in the life of the tool by the senses 5.2.8 Manufacturing process
- either visually or by its sound during machining.
This application of the power spectrum to process- envelopes
related manufacturing problems gives an early
warning of the performance of the complete produc- The principal feature of manufacturing process
tion process, enabling remedial action to be taken to envelopes and, indeed, for many amplitude distrib-
rectify a minor problem before it becomes critical. ution curves is that they can be approximated by the
A note of caution is required here, as there are occa- so-called "beta function" (Figure 191a). Here, the
sions when even the application and interpretation function has two parameters that are independent
of the power spectrum may prove inadequate to of one another, enabling them to be utilised as a
identify problems in production. This limitation to means of characterisation. The notation a is the allo-
the use of such powerful diagnostic aids can arise in cated weighting for the profile ordinates measured
situations where either time or space varies; further- from the lowest valley and above, with the notation
more, if workpiece variability through unanticipated b being given to weighting the profile from the
defects is present in the production batches this will highest peak down. Hence, peaks and valleys have
nullify the technique. accordingly different weights. One of the problems
(a) The beta function: (i) symmetrical and (ii) asymmetrical case. (After Whitehouse, 1994)
(i) (ii)
0=b=5 3.0 0=5
b =1.5
o =b=3 0=2
b =1
0=3
o =b=2 b =1.5
a =b =1 0=1.5
b=3
a =1.5
b=5
0 0.5 1.0 0.5 1.0

(b) Skewness and kurtosis relationship, illustrating different

manufacturing process envelopes
Kurtosis
('\ 14

\ \ Plateau honing
12
\ ~ Milling

10

-2 -1 o +1 +2
Skewness

(c) Skewness: kurtosis graph, showing the functional envelopes

Kurtosis

12

-2 -1 o +1 +2
Skewness

Figure 191. "Beta function" and typical production process envelopes.

that has arisen from using this technique for a topo- open to the "free surface". Moreover, with the rem-
graphical profile, which has discredited it for certain nants of the periodic saw-toothed profile (produced
applications, is how and in what mann er can one by the effect of the partial lip and margin of the
determine a and b. The beta function is normally drill), this gives a negative skewness via the pores
defined within a set range of 0 --7 I, being expressed and positive skewness introduced by drilled cusps
in the following manner: (saw-tooth effect). In Figure 192 the dispersion or
scatter of the results for individual envelopes varies,
ß(a, b) = f za-l(a - Z)b-l dz
depending upon the frequency of the feedrate and
the type of drill employed. Generally, the finer the
feedrate the less the scatter, this also being true for
If by changing the range of the beta function equa- many turning and milling operations. The geometry
tion above, from 0 --7 1 to Rp + Rv, or indeed with of the drill's lips, but more specifically the profile of
Rt, then substituting (F (the standard deviation of the the chis el edge and whether it is of the self-centring
distribution) with Rq, the beta function parameters type, such as will be the case for a "split-point" drill,
a and b become
creates less drill deflection and hence the results
introduce smaller scatter to the envelope. These
RVRP - Rq2) machining conditions introduce an intermediate-
a = Rv ( type of manufacturing envelope that is neither a
Rt Rq2
"bearing" nor a "locking" variety.

b = Rp (RV Rp - Rq2)
Rt Rq2 Ternary manufacturing envelopes (TMEs)
The reality that any dominant peak or valley within In machining operations the dominant factor that
the assessment length is only raised to a unit power influences both the resultant cutting forces and the
infers additional stability over the skewness/kurtosis surface topography has been shown to be the tool's
approach. The trouble with this method is in accu- feedrate. In Figures 193 and 194 the feedrate, in
rately determining sound results from Rv and Rp, conjunction with the principal factors of surface
which confirms the problem that obtaining infor- texture (Ra) and roundness (by least squares cirde
mation from peak/valley measurement and then - LSC), are used to define the limits for these TMEs.
deriving valid information from them can be fraught Utilising these diverse surface texture, roundness
with difficulties. In Figure 191(a) both symmetrical and processing parameters (feedrate) for the major
(i) and asymmetrical (ii) graphs for the beta func- axes on the ternary diagram enable a machined
tion are illustrated, based on a dass of Pearson surface to be characterised in a new manner. As
distributions. In the symmetrical case the skewness might be anticipated, these TMEs differ markedly
equates to zero; conversely, for an asymmetrical from the more usual and restricted manufacturing
series of results, skewness can be either positively or envelopes previously alluded to above; the latter type
negatively skewed. Nevertheless, even allowing for of skewness and kurtosis axes might otherwise
these limitations, an example of a range of the manu- mask crucial information. The "TME approach"
facturing process envelopes that can be produced gives a pseudo three-dimensional graph that can be
using this technique is illustrated in Figure 191(b). exploited to illustrate how the influence of changing
Here the production processes can be simplistically a parameter such as feedrate modifies the relation-
dassified and grouped into either a "bearing" or ship of the surface texture and roundness values in
"locking" surface topography (Figure 191c). This the final machined result. Furthermore, such a three-
bearing/locking grouping shows that certain pro- dimensional graph can be exploited to illustrate how
duction processes can achieve specific functional the influence of changing the feedrate modifies the
surfaces for industrial applications. However, the interrelationship of surface texture and roundness
general picture is not as distinct as the grouping values.
shown in Figure 191(c), because certain processes As an example of the usefulness of this TME
cannot only run across each dassification but may approach to complex machining analysis, Figures
fall between both the bearing and locking envelopes. 193 and 194 for dissimilar production processes are
A typical "intermediate envelope" that normally presented. The TME graph for a particular range of
occurs in just this vicinity of the skewness/kurtosis turning and boring processes, indicated in Figure
graph would be machined surfaces manufactured by 193, shows how at low feedrate (0.10 mm/rev) the
the PM route, as shown in Figure 192. Here, in both surface texture is dosely confined to a relatively
a twist and split-point drilling operation, the porous small spread of values - nominally around 0.5-1.5
nature of the PM compacts creates pores that are f.Lm Ra, whereas its associated roundness lies
 Kurtosis: Rku (!-tm)

Feedrates (mmJrev)
Jobber Split-point
0.127 GJ--IZ>'I--
(SAMPLES: 1 to 8)
0.254 o ~--

-4 -2 o 2 4
Skewness: Rsk (!-tm)

Figure 192. Manufacturing envelopes for jobber and split-point drilling operations: skewness v. kurtosis.
 Surface texture: Ra (11m)

16

15

14

13

12

11

10

Feedrates (mmlrev)
Turning Boring
0.10 8-- ®--
0.25 0 - - (n/al Feedrate (mmlrev)

0.40 8. - - A - -

Figure 193. Ternary manufaeturing envelopes of the produetion processes: turning and boring - feedrate, roundness and surfaee texture.
between approximately 5 and 30 /-Lm LSC. As the is only the result of linear motion (being caused by
feedrate increased in an arithmetic progression to the fee ding motion along the part) and it is not sur-
0.25 mm/rev, the range of the surface texture band- prising to note that a direct relationship exists
width proportionally expanded to l.5 at approxi- between selected feedrates and tool nose geometries.
mately 5-6.5 /-Lm Ra, with a corresponding increase The boring proportional bandwidth can be ex-
in roundness from 8 to 48 /-Lm LSC, giving a propor- plained, allowing for its overhang, which must be
tional bandwidth of l.6. As the feedrate was raised present to avoid fouling on the workpiece. This over-
even high er, to 0040 mm/rev, it was not surprising to hang creates a reduction in stiffness based on the
note that this also produced increases in both the cube of the bar's overhanging length; when com-
surface roughness and its proportional bandwidth, pared to the process of turning this is its OD opera-
with similar values with respect to its roundness. tional counterpart. Aboring tool intro duces a
Similar trends occurred for the boring operation "phase-damping" effect on the out-of-roundness
on Figure 193, but here only two feedrates were and, as such, reduces the overall magnitude of the
employed. At low feedrate (0.10 mm/rev) the surface LSC value obtained. When utilising the 0040 mm/rev
texture range was approximately 0.5-2 /-Lm Ra, with feedrate, boring in this case will generate a propor-
roundness being between 8 and 36 /-Lm LSC. Once the tional bandwidth increase in LSC of l.7, while an
boring feedrate had been raised to 004 mm/rev, this equivalent OD turning operation produces a value of
produced an increase in the surface texture to l.92. The more compliant and less rigid boring bar
around 13-18 /-Lm Ra, with a corresponding propor- will have a greater tendency to deflect the cut, leading
tional bandwidth of 33.33. Furthermore, the round- to slight tool bending termed a "spring cut", intro-
ness also degraded to 15-63 /-Lm LSC, which resulted ducing a form of "phase-damping" behaviour that
in the bandwidth being proportionally increased to tends to smooth and blend out the harmonie factors.
l.7. Discussion of these proportional bandwidth dif- Conversely, the opposite is true for the behaviour of
ferences can be used to explain why either the the proportional bandwidth in machined surface
outside diameter (OD) turning or internal diameter texture parameters, where an increase was observed
(ID) boring operational performances vary. In the in boring with respect to turning at identical feed-
case of the OD turning operation (a schematic of the rates. This difference can be explained in terms of the
turning process is shown in Figure 169), as the fee- associated boring bar overhang and its reduction in
drate is progressively increased both the surface rigidity. Increased tool deflection would tend to be
texture and roundness will degrade. This expected superimposed onto the bored surface topography,
deterioration occurs because as the feedrate in- leading to greater dispersion in data points, whieh
creases it results in coarser machined surface topog- then increases the proportional bandwidth if com-
raphy, and as the cutting forces are proportionally pared to a similar turning operation, with both
high er this provides a destabilised cutting-edge cutting inserts having identical nose radii.
condition, which subsequendy leads to greater The drilling TME depicted in Figure 194 shows a
departures from roundness, resulting in secondary limited range of drilling operations using both an
harmonie roundness error. There was a larger uncoated jobber drill and a TiN-coated split-point
proportional bandwidth expansion for surface drill. In this case the TME shows how the generation
texture as a result of the influence of feedrate rather process of dynamieally drilling holes influences the
than for departures from roundness. This bandwidth simultaneous procedure of surface roughness and
increased in the same fashion to the increases in roundness production at different feedrates. The
respective feedrates, namely in an arithmetie pro- normal technique is to consider each metrologieal
gression - lengthening at the rate of l.5. Such condition in isolation and thereby potentionally
increases in dispersion of resultant machining data missing some vital information. In Figure 194a "com-
can be substantiated by earlier work, whieh suggests parison of extremes" can be made to establish how
that there is a geometrie relationship between the differing tool lengths combined with changes in
insert nose geometry and selected feedrate on the point geometry influence the resulting surface
respective machined surface texture. However, this texture and roundness parameters. Most surpris-
proportional behaviour did not extend itself to ingly, at a 0.127 mm/rev feedrate the split-point drill's
machined roundness values, which are a function of mean range of surface texture was worse
a more complex interactive cutting path action. Such than that obtained with a jobber drill, ranging over
action is promoted by combined rotational and 2.8 /-Lm Ra; this was possibly due to the "rounding
linear motion (as the workpiece rotates and the tool effect" on the drill's lips and point edges. Equally, the
is fed along the part), leading to differential primary, LSC roundness range showed that the jobber drill
secondary and tertiary (an so on) harmonie depar- was far worse than that for the split-point, its vari-
tures from roundness. The machined surface topog- ability increasing by 33%, being present over 30 /-Lm
raphy, indicated through the Ra surface texture trace, LSC. What this variation illustrates is that the jobber
 Surface texture: Ra (~m)

4
I
I
I
3

Feedrate (mmJrev)
Feedrates (mmJrev)
Jobber Split-point
0.127 8 - - 181- -
(Sampies: 1 to 8)
0.254 0-- ~ - -

Figure 194. Ternary manufaeturing envelopes of produetion processes: jobber and split-point drilling - feedrate, roundness and surfaee
texture.
drill, at low feed, had a smaller and more consistent The in-depth discussion on the use of the TME
spread of results for surface roughness than that of technique with their related production processes
the split-point drill, although its level of recorded and associated metrological parameters, together
roughness was high er. This drilled hole variability with subsequent results above, was included to illus-
suggests that the slightly larger drill margin on the trate how machined components and specific
jobber drill burnished the resulting saw-toothed processes can be combined into pseudo-three-
profile topography and, in so doing, reduced the Ra dimensional diagrams. Such graphs may give a new
range. Hole roundness deteriorated using the jobber insight into what constitutes a machined surface,
drill. This deterioration came as no surprise, as its indicating that surface text ure parameters alone
longer and flatter chisel-edge profile, combined with should not be separated out from those of round-
a greater overall drilliength to that of the split -point, ness. The combination of these previously disasso-
inevitably meant that the drill produced a degraded ciated surface texture and roundness parameters
self-centring action. Coupled to the jobber's greater may offer a new approach in the manner that the
length, reduced drill integral rigidity occurred; this machined surface condition is assessed, allowing an
led to inferior drilled hole roundness. Once the fee- appreciation of the whole surface and not just some
drate was increased to 0.254 mm/rev, a similar trend associated parameters in isolation.
occurred for drilled hole roughness. The split-point
drill's range was 41 % larger and of similar magni-
tude to that of the jobber - ranging over an Ra of 3.9
/-Lm - illustrating that drill burnishing was present in
this case at higher feedrates. However, the trend for 5.3 Surface engineering
roundness at this high er feedrate was reversed. The
split-point drill showed a larger error in the range
of departures from roundness, being 25% worse, Historical context
coupled to a similar magnitude - ranging over an
LSC of 32 /-Lm - indicating that at higher feedrates the The currently accepted definition of the subject
self-centring action was a negligible factor. of "surface engineering" was adequately coined in
If a comparison is now made of the surface the late 1980s by Professor Bell of the University
texture and roundness parameters against feedrate, of Birmingham. It states that surface engineering
in the first instance for the jobber drill and then involves the application of traditional and innovative
considering the effects produced by the split-point, surface technologies to engineering components and
a significantly different pieture emerges for the materials in order to produce a composite material
drilling process. When utilising the jobber drill, and with properties unattainable in either the base or
as the feedrate increases from 0.127 to 0.254 mm/rev, surface material. Frequently, the various surface tech-
there is a slight improvement in the range of surface nologies are applied to existing designs of engi-
texture from 2.6 to 2.3 /-Lm Ra, for the low-to-high neering components but, ideally, surface engineering
feedrates, respectively. This 11.5% improvement in involves the design of the component with a know-
the spread of Ra at higher feedrate is somewhat ledge of the surface treatment to be employed.
misleading as the magnitude is greater, which Historically, some of the earliest references on the
conforms to the published literature. This partial yet subject of surface engineering are somewhat vague
slightly confusing improvement in surface texture allusions to the metallurgical treatment of armour,
using the jobber drill at high feedrate also occurred notably quench hardening of weapons by the
with respect to roundness, as the LSC ranged from process of blinding the Cyclops in the Odyssey, by
8 to 39 /-Lm at high feedrate. Once again, it would Homer. This Greek saga was written about 880 BC,
seem that the wider jobber drill's margin influenced but referred to an earlier Bronze Age period (1400-
the burnishing both linearly with respect to variance 1200 BC). However, in the Iron Age the carburising
in asperities and circumferentially, by minimising of weapons by some form of case hardening, or more
harmonie departures from roundness. The drilling specifically carbonitriding, was widely practised.
behaviour of the split-point was more predictable. This heat treatment process was in reality an austen-
In the case of surface roughness, this ranged from itic thermo-chemical procedure involving the diffu-
2.8 /-Lm at low feedrate to 3.9 /-Lm at high feed, with sional addition of both carbon and nitrogen to
similar magnitudes with respect to Ra levels. The austenite. This was a somewhat long-winded process
trend in roundness variations was much the same; and was followed by rapid and agitated vertical
at low feedrate the range was 20 /-Lm, while at high quenching to harden and improve the wear charac-
feed it degraded to 32 /-Lm, with similar magnitude teristics of the weapon, followed by tempering to
in LSC level. This degradation is wholly consistent enhance the toughness. Notably, in the eleventh
with increasing penetration rates. century AD a German monk, Theophilus, in his work
 Schedule diversarum artium, mentioned the word nous and impracticable within the scope of the
temperamento, meaning to quench and self-temper current work; however, only an overview can be
in goat urine - after certain complex pre-treatment given here. The level of technological development
of the goat! This prior management of the goat was for surface engineering technologies across various
necessary to increase the acidity level of the sub se- manufacturing industries will be markedly different,
quent fluid media necessary for heat treatment and although their natural sequence of events for
hence enhance its severity, enabling the liquid to be component production may be summarised in the
utilised for severe quenching applications. Although following manner:
they did not know it at the time, such practical heat
treatment enhanced the critical cooling velocities Application => Properties (required) => Design =>
through the time-temperature transformation (TTT Materials (selection) => Engineer (substrate) =>
curve), obtaining the desired mechanical properties.
In Japan the swords of the samurai were uns ur- Engineer (surface) => Lubrication =>
passed in their keen edge and cutting ability. Their Performance
success relied on very complex manipulation of the
sword's grain structure (up-setting and folding by (After Bell, 1990)
hammer blows of the weapon on the anvil) and intri- Recent and traditional surface technologies con-
cate heat treatment processes, in order to obtain the tribute to a multifarious assortment of treatments
desired metallurgical and mechanical properties. enabling design and manufacture of a diverse range
Contemporaneous with the Japanese were the early of metallic-based composites. By modifying a com-
metallurgical developments by Chinese dynasties, ponent's structural characteristics/composition, it
whereby soya bean pro tein that had decomposed intrinsically changes its engineering properties. In
was employed to enrich the red-hot edges of steel simplistic terms, non-mechanical surface treatments
swords with carbon and nitrogen, and in so doing can be classified into various groupings, as follows:
enhance their mechanical properties.
In Britain a 1727 publication entitled Husbandry • thermal treatment;
and Trade Improvement by John Houghton FRS • thermo-chemical treatment;
made reference to surface hardening. In the West • plating and coating;
Midlands techniques for surface improvement "... • implantation.
as being done at Wolverhampton in a different
manner; with burnt hoofs and horns, bag felt, old For an engineering designer, the technological
burnt leather and tartar all mixed together with selection of a means to manipulate and change a
urine" again seemingly found the latter ingredient surface becomes an integral part of component
essential! Pre-dating this time was the first scientific design. The decisive step taken to enhance a compo-
investigation of surface treatments undertaken by nent's material surface signifies the fact that one of
Reaumur in France in his magnum opus published the existing base material properties needs manip-
in 1722. In fact, some rather dubious materials ulation in some way, in order that it satisfactorily
(burnt leather in particular) were being employed fulfils its functional performance, while at the same
for case hardening, by pack methods, prior to the time being economically acceptable. An engineered
First World War. These natural forerunners of later surface can have its coating thickness varying from
surface engineering processing and treatments for several millimetres, by weId overlays, to just a few
enhancement of metallurgy, typified by techniques micrometres - for either physical vapour deposition
such as grain refining and surface quench hard- (PVD) or chemical vapour deposition (CVD) coat-
ening, could be utilised without reliance upon elec- ings - and surface depth modifications to approxi-
trical heating/phenomena to induce modifications mately 0.1 /-Lm - by ion implantation (Figure 195).
to surface properties. Only with the advent of an In a similar manner the hardness of coatings spans
active research programme into electrical pheno- a wide range; typically these might be:
mena in the late nineteenth and early twentieth
centuries was it possible to envisage surface tech- • spray coatings - ranging from >250 to 350 Hy ;
nology applications, many of which are now becom- • nitrided steels - around 1000 Hy ;
ing mature commercial technologies. • detonation gun (D-gun) carbide-metal cermet
coatings - ranging over 1300-1600 H y ;
• titanium nitride (TiN) coatings by the PVD
(ontemporary surface engineering process - up to 3500 Hy •

If the subject of surface engineering was only Initially when selecting a surface modification
partially developed, then the text would be volumi- treatment, it is important to comprehend what the
 10

==!~~f/Wß
cv~t0WA1
I W~ Electroless ~ickel
I 01)
=
~

Composi~e~ 8
"0
=~

~M Thermal SfraYingi
c:::

~Urface Weld!ng /2/'j/ß

AnO~iSing Wh'2
- I I

r-:-
~ Ion Implantation
-- I I

BorO~iSing ~
I

~~~ carboNitridi;n g

Carburisi~g ~~
I

Surface AI!OYin g ~~"S3

----
Thermal ~ardeninglL._ _ _ _ _.......J

Figure 195. Typical thicknesses of surface layers for engineering components. [After Bell, 1990.]

anticipated surface and subsurface requirements • scuffing resistance, or resistance to seizure;

might be for the component's in-service function. • torsional, or bending strength;
Often these decisions involve only one of the • torsional fatigue/bending strength;
following set of properties, but may embrace several • resistance to surface contact fatigue (mechanical
for complex surfaces: pitting);
• resistance to surface collapse (case crushing);
• abrasive wear resistance - for either high or low • corrosion, or erosion resistance.
compressive loading;
 In recent years, in order to assistance the designer times unwittingly, or intentionally a surface can be
in their choice of the correct decision related to the either enhanced or degraded by the actual production
surface's functional performance, the development process. Moreover, some surface engineering treat-
of databases utilising intelligent design systems has ments and processes have reached a significant level
been of some success, although a comprehensive all- of sophistication, offering exciting challenges in the
embracing design system is still some way off as a future for potential industrial applications. In the
practical reality. final chapter the reader's attention will be drawn to
the reasons why, where and how it is important to
ensure that instrumentation is appropriately cali-
Future surface engineering applications brated at planned periodic intervals, thereby ensuring
that any results obtained are both valid and traceable.
The widening market acceptance of the previous
"first-generation" surface technological innovations
and their assimilation into design and manufacture
have allowed many technically enhanced engi- References
neering surfaces now to be considered as economi-
cally viable. It has been reported that at least eight Journals and conference papers
widely diverse activities will dominate surface devel-
Allen, A.F. and Brewer, R.C. The influence of machine tool vari-
opment in the future, exduding those currently in
ability and tool flank wear on surface texture. Proceedings of
use for microelectronic surfaces and treatments; the Sixth International Machine Tool Design and Research
these areas are: Conference, 1965,301-314.
Bell, T. Surface engineering: past, present and future, Surface
1 non-ferrous metal surface engineering; Engineering 6(1),1990,31-40.
Bellows, G., Field, M. and Kahles, J.B. Influence of material and its
2 polymer and composite surface engineering; metallurgical state on surface integrity. In Influence of
3 surface engineering of ceramic materials; Metallurgy on Machinability, ASM. 1975,272-295.
4 mathematical modelling of components for Bonifacio, M.E.R. and Diniz, A.E. Correlating tool wear, toollife,
surface enhancement; surface roughness and tool vibration in finish turning with
coated carbide tools. Wear 173, 1994, 137-144.
5 "second-generation" surface engineering techno- Bousfield, B. and Bousfield, T. Progress towards a metallography
logies; standard. Metals and Materials March 1990, 146-148.
6 surface engineering in material manufacture; Brinksmeier, E., Cammett, J.T., König, W., Leskovar, P., Peters, J. and
7 statistical process control (SPC) in surface engin- Tönshoff, H.K. Residual stresses: measurement and causes in
eering; machining processes. Annals of the CIRP 31(2),1982.
Chadda, Y. and Haque, M. Finish-turning surface roughness.
8 non-destructive testing (NDT) of surface-engin- Cutting Tool Engineering June 1993,95-99.
eered components. Chou, K.Y and Evans, c.J. White layers and thermal modeling of
hard turned surfaces. International Journal of Machine Tools
Several of these surface-engineered activities and Manufacture 39, 1999, 1863-188l.
Davies, N., Spedding T.A. and Watson, W. Auto-regressive moving
are now becoming accepted across a wide field of
average processes with no normal residues. Time Se ries Annals.
specialisms, with others yet to become more than 2,1980,103-109.
simply research topics as they are at present. It is EI-Wahab, A.I. and Kishawy, H.A. A new method to improve
undoubtedly true that the capability of producing surface quality during CNC machining. International Journal of
complex duplex or multiple surface treatments, Production Research 38(16),2000,3711-3723.
Fathailal, M., Danai, K. and Barber, G. Effect of flank wear on the
providing an intricate array of novel surface- topography of machined surfaces. Tribology Transactions 36(4),
engineered products, is almost unlimited in its 1993,693-699.
scope. A typical example of this sophisticated Field, M. and Kahles, J.F. Review of surface integrity of machined
surface treatment occurs when attempting to components. Annals of the CIRP 20(2), 1971, 153-163.
Ford, D.G. Machining to microns: error avoidance or compensa-
improve the machining process, this being provided
tion? Proceedings of Lamdamap II. Computational Mechanics
by the multiple coating treatment of cutting tool 1995,277-286.
inserts, where non-metallics and metallics are Ford, D.G., Postlethwaite, S. and Morton, D. Error compensation
deposited as hard and soft-glide coatings that are applied to high precision machinery. Proceedings of Lamdamap
utilised to significantly enhance a specific produc- I. Computational Mechanics, 1993, 105-121.
Galloway, D.F. Some experiments on the influence of various
tion process. Oflate, surface engineers and designers factors on drill performance. Transactions of ASME February
must be encouraged to cooperate ever doser, in 1957, 191-23l.
order to realise from a commercial viewpoint the Gillibrand, D. and Heginbotham, W.B. Experimental observations
combinations of surface technologies that currently on surface roughness in metal machining. Proceedings of the
18th International Machine Tool Design and Research Con-
exist and their relevant manner of exploitation.
ference, Las Vegas, 1977,629-640.
This chapter has attempted to explain the impor- Griffiths, B.J. Deficiencies in surface specifications. Proceedings of
tance of component surface integrity and how some- Lamdamap III. Computational Mechanics, 1997,465-474.
Griffiths, B.J. Manufacturing surface design and monitoring for Tönshoff, H.K. Eigenspannungen und plastische Verformungen
performance. Surface Topography 1, 1988,61-69. im Werkstück durch spanende Bearbeitung. Dr.-Ing. disserta-
Griffiths, B.J. Mechanisms of white layer generation with reference tion, TH Hannover, 1965.
to machining and deformation processes. Transactions of ASME Turley, D.M., Doyle, E.D. and Samuels, L.E. A structure of the
- Journal ofTribology 109, July 1987,525-530. damaged layer on metals. Proceedings of the International
Griffiths, B.J. and Furze, D.C. Tribological advantages of white Conference on Production Engineering, Tokyo. Japanese Society
layers produced by machining. Transactions of ASME - Journal of Precision Engineering Part 2, 1974.
ofTribology 109, April 1987,338-342. Vajpayee,S.Analytical studyof surface roughness in turning. Wear
Griffiths, B.J. Problems in measuring the topography of machined 70,1981,165-175.
surfaces produced by plastic deformation mechanisms. Wear Watson D.W. and Murphy, M.C. The effect of machining on surface
109,1986,195-205. integrity. Metallurgist and Materials Technology April 1979,
Grum, J. and Oblak,A. Use oflinear discriminating function for the 199-204.
description of the effects of microstructure on surface rough- White,A.J.,Postiethwaite, S.R. and Ford, D.G.An identification and
ness after fine-turning.Acta Stereol, 18(3), 1999, 313-318. study of mechanisms causing thermal errors in CNC machine
Huynh, V.M. and Fan, Y. Surface-texture measurement and charac- tools. Proceedings of Lamdamap IV. WIT Press, 1999, 101-112.
terisation with applications to machine-tool monitoring. Whitehouse, D.J. Surface measurement fidelity. Proceedings of
International Journal of Advances in Manufacturing Technology Lamdamap IV. WIT Press, 1999,267-276.
7(2),1992,1-10. Whitehouse, D.J. Conditioning of the manufacturing process using
Koster, W.P, Field, M, Fritz, L.J., Gatto, L.R. and Kahles, J.F. Surface surface finish. Proceedings of Lamdamap III. Computational
integrity of machined structural components. Airforce Mater- Mechanics, 1997,3-20.
ials Laboratory Technical Report AFML-TR-70-11, MMP Project Wu, D.W. and Matsumoto, Y. The effect of hardness on residual
No. 721-8. Metcut Research Associates Inc., Cincinnati, OH, stresses in orthogonal machining of AISI 4340 steel.
March 1970. Transactions of ASME - Journal of Engineering and Industry
Lamb, A.D. Some aspects of the character of machined surfaces. 112,August 1990,246-252.
Metals and Materials March 1987,75-79. Yeh, L-J and Lai, G-J. Study of the monitoring and suppression
Ng, E-G. and Aspinwall, D. Hard part machining AISI ((50 HRC) system for turning slender workpieces. Proc. Instn. Mech Engrs.
using Amborite AMB90: a finite element modelling approach. 1995,209,227-236
Industrial Diamond Review 60(587, part 4),2000,305-312. Zhu, C. How to obtain a good quality surface finish in NC machin-
Patz, M., Dittmar, A., Hess, A. and Wagner, W. Better hard turning ing of free formed surfaces. Computers in Industry 23, 1993,
with optimised tooling. Industrial Diamond Review 60(587, part 227-233.
4),2000,277-282.
Rakhit, A.K., Sankar, T.S. and Osman, M.O.M. The influence of
metal cutting forces on the formation of surface texture in Books, booklets and guides
turning. International Journal of Machine Tool Design Research
16,1976,281-292. AmericanNational Standard (ANS) B211.1 - Surface Integrity, 1986.
Rhoades, L. Understanding edmed surfaces. Cutting Tool Engin- Behaviour of Materials in Machining. 10M Communications,
eering April 1996, 22-31. November 1998.
Shiraishi, M. and Sato, S. Dimensional and surface roughness Boothroyd, G. Fundamentals of Metal Machining and Machine
controls in a turning operation. Transactions ofASME 112, 1990, Tools. McGraw-Hili, New York, 1975.
78-83. Cardarelli, F. Materials Handbook. Springer, Berlin, 2000.
Smith, G.T. Ternary manufacturing envelopes (TME's): a new Dieter, G.E. Mechanical Metallurgy. McGraw-HilI, New York, 1976.
approach to describing machined surfaces. International Griffiths, B.J. Manufacturing Surface Technology: Penton Press,
Journal ofMachine Tools and Manufacture 40(2), 2000, 295-305. 2001.
Smith, G.T. Surface integrity. Machinery Market October 1998, Halling, J. Introduction to Tribology. Wykeham Engineering and
28-30. Technology Series, 1976.
Smith, G.T. Beauty is only skin deep? Metalworking Production Höganäs Handbook for Sintered Components: Machining Guide-
February 1998,7-8. lines. Höganäs AB, 1998.
Smith, G.T. The influence of insert wear on machined component Harris, W. T. Chemical Milling. Clarendon Press, Oxford, 1976.
surface texture whilst turning. Proceedings of Lamdamap II. Influence of Metallurgy on Hole Making Operations. ASM, 1978.
Computational Mechanics, 1995,355-366. Influence of Metallurgy on Machinability. ASM, 1975.
Smith, G.T. The surface topography resulting from boring powder McGeough, J.A. Advanced Methods of Machining. Chapman &
metallurgycomponents. Tribologia 13(4), 1994,44-62. Hall, 1988.
Smith, G. T. Hard-metal turning of crankshafts. Proceedings of the Momber, A. and Kovacevic, R. Principles of Abrasive Water Jet
Third International Conference on the Behaviour of Metals in Machining. Springer, Berlin, 1998.
Machining, University of Warwick. 10M Communications, Metalworking Fluids. Verlag Moderne Industrie, 1991.
November 1994. Modern Metal Cutting, Sandvik Coromant, 1994.
Smith, G.T. Some aspects in the surface integrity and tool wear Rao, B.K.N. Handbook of Condition Monitoring. Elsevier Advanced
associated with turning powder metallurgy compacts. Wear Technology,1996.
151,1991,289-302. Finish Turning. Sandvik Coromant, 1995.
Smith, G.T. Secondary machining operations and the result- Shaw, M.C. Principles of Abrasive Processing. Oxford University
ing surface integrity: an overview. Surface Topography 3, 1990, Press, 1996.
25-42. Shaw, M.C. Metal Cutting Principles. Oxford University Press, 1996.
Smith, G.T. The surface integrity of turned ferrous powder metal- Smith, G.T. CNC Machining Technology. Springer, Berlin, 1993.
lurgy components. Powder Metallurgy 33(2), 1990, 155-165. Smith, G.T. Advanced Machining: The Handbook of Cutting
Smith, G.T. and Maxted, P. Evaluating the high-speed machining Technology. IFS/Springer, 1989.
performance of a vertical machining centre during milling Surface Technology. Verlag Moderne Industrie, 1990.
operations. Proceedings of Lamdamap II. Computational Trent, E.M. Metal Cutting. Butterworths, 1984.
Mechanics, 1995, 123-138. Tlusty, G. Manufacturing Processes and Equipment. Prentice-Hall,
Stout, K.J. and Spedding, T.A. The characterisation of internal Englewood Cliffs, NJ, 2000.
combustion engine bores. Wear 83,1982,311-326. Williams, J.A. Engineering Tribology. Oxford Science, 1996.
 Qualityand
calibration techniques

"Si parva licet componere magnis."

Translation
"l! one may measure small things by great."
(Virgil, 70-19 Be)

249

G. T. Smith, Industrial Metrology

© Springer-Verlag London 2002
 111
Artefacts such as the Elizabethan yard are termed
Dimensional measurement: "end standards", being engraved with an initial "E"
historical perspective and the Royal crown to denote universal acceptance
in the country. However, until 1824 this and other
measures remained legitimate until superseded by
Archaie length measurement an Act of Parliament under King George IV, when
the imperial standard yard was introduced, to
Probably the oldest units of measurement employed minimise the inaccuracies generally associated with
in the ancient world were related to either body mea- linear measurement at the time. This imperial yard
surements or physical entities that were readily avail- had a relatively brief li fe as a comparative end stan-
able in daily life. Notably, in Egypt around 3000 BC dard because it was destroyed in 1834 in the fire
the basic form of measurement was the cubit, this that swept through the Houses of Parliament, where
being a measurement taken from the arm relating to it resided. Restoration of this linear standard was
the distance from the tip of the finger to the elbow. undertaken from an alloy composition of copper
The cubit could then be subdivided into shorter (16 parts), tin (2.5 parts) and zinc (1 part). It was
units, such as the foot or the hand - this latter term colloquially known as "Bronze No. 1" by its creator,
today still being used for expressing the height of the Rev. Sheepshanks, the principal design feature
horses. Conversely, the cubit can be extended by being that engraved gold plugs were recessed into
adding several together to form longer units such as the artefact at the plane of its neutral axis to
the stride. Unfortunately, these physical measure- minimise Hexure and hence measurement inaccu-
ments would vary considerably owing to the differ- racy. Duplicates of exacting dimensions were made
ence in the size of people. In England, as earlyas 1000 and they were legalised by Parliament in 1855, one
the Saxon King Edgar kept a yards tick at Winchester being presented to the Government of the United
as the official standard of measurement. However, it States of America in 1856.
was not until the reign of King Richard I (the
Lionheart) that any form of standardisation of units
of measurement was first documented. In his Assizes International Bureau of Weights and
of Measures in 1196 it was stated: "Throughout the Measures: the Metre
realm there shall be the same yard of the same size
and it should be of iron." This metallic artefact was In 1875 in continental Europe parallel development
the first permanent standard measure to be utilised was taking place creating the metric system, where
in an attempt to control the vagaries of linear mea- at Paris the "Metre Convention" of participating
surement. In 1215 in the reign of King John, the nations established the International Bureau of
Magna Carta also considered units of standardised Weights and Measures (BIPM). In 1889 the metal-
measurement - including those of wine and beer! lurgy and design of measurement artefacts had been
The yard (or ulna, as it was sometimes known) improved from the rather crude designs and
together with its linear subdivisions and aggregated elementary metallurgical compositions to those of a
divisions, came into existence during the reign of more uniform cross-section and greater stability of
King Edward I (1272-1307). Here, it was stated that artefact materials. In particular, of the number of
"It is ordained that three grains of badey, dry and artefacts produced, No. 6 manufactured from plat-
round make an inch, twelve inches make a foot, three inum-iridium, which replaced its more primitive Hat
feet make an ulna [yard], five and a half ulna make a cross-sectioned forerunner (Metre des Archives), was
perch (rod), and forty perches in length and four to become known as the International Prototype
perches in breadth make an acre." The perch, or rod as Metre. üf the remaining bars, they were distributed
it was known, was a traditional Saxon land measure to the representative nations, Britain receiving
that still survives into the twenty-first century, bar No. 16. Britain did not actually sign the Metre
although it was originally defined as "the totallength Convention until 1884 and even then was reluctant
of the left feet of the first sixteen men to leave church to implement the clause referring to the intro duc-
on Sunday morning", this statement being somewhat tion of metric measures into signatory countries. In
open to variation in its determination of length! fact, the use of the metric system of weights and
In 1588 the Elizabethan yard or ell as it was other- measures by trade and industry only became law in
wise known was on ce again based on a large man's Britain in 1897, but was not universally adopted -
stride. Alternatively, the measurement could be and then with certain res ist an ce - until 1971.
taken from the distance between a man's nose and It had been known for some time that a natural
his thumb when standing and stretching the arm - constant measurement value for the metre was
this technique often being used as a rudimentary required. The solution for its appropriate length was
form of measurement for the length of cloth. developed from a portion (quadrant) of the circum-
ferential line of longitude through the Earth, theo- any well-equipped metrology laboratory. In order to
retically at sea level. Ihe French National Assembly get an accurate determination of the metre in terms
meeting's 1791 definition of the metre was: "one ten of the wavelength of light, a comprehensive testing
millionth of the polar quadrant of the earth passing programme was initiated - nine times - between
through Paris". In practical terms what this meant 1892 and 1940 at various laboratories, typically twice
was that a line running from the North Pole to the at the National Physical Laboratory (NPL) in 1932
Equator, passing through Paris, would represent a and 1935. Ihe me an of these nine results became the
quarter of the Earth's circumference and, if this was basis of a new definition for the metre: the length
broken down to a ten millionth part of this linear equal to 1,650,763.73 wavelengths in a vacuum of the
length, the amount would represent a metre. A team radiation . .. of krypton-86. In 1960 the met re was
of surveyors was charged with the difficult task of defined in this mann er, thereby replacing the
establishing this geometricalliinear relationship International Prototype Metre as the fundamental
and took six years to complete the work, during the standard of measurement at the time.
war between France and Spain. In practice, only a At the same time (1960), the first laser was con-
relatively small portion of the geographical terrain structed and by the mid-1970s lasers were being
could be surveyed, between Dunkirk (France) to used as length standards. Iherefore, in 1983 the
Barcelona (Spain), minimising the affect of altitude definition of the met re using krypton-86 source
whenever practicable. Many of these surveyors were became: the length of the path travelled by light in a
charged as spies, nearly losing their heads in the vacuum during the time interval of 11299,792,458
process of this work, the outcome of the exacting of a second. Ihis metre length is realised by employ-
and testing survey being the 1799 Metre des Archives. ing iodine-stabilised helium-neon lasers having
Ihis artefact became the "master standard" for the a reproducibility of ~ ±2 parts in 100,000,000,000,
first entity to be known as a means of global or alternatively to ±0.00000002 mm. Although as a
measurement classification, the so-called "metric wavelength source the iodine-stabilised helium-
system". neon laser has excellent characteristics, in the ne ar
future it should be possible to construct lasers
with vasdy superior wavelength accuracy. Ihe new
Optical and laser length measurement laser instrumental design is necessary because the
current iodine-stabilised helium-neon type is
Several major problems exist when using conven- limited by the fact that the iodine molecules are at
tional artefacts as the basis of comparative measure- room temperature and collide into each other and
ment. Most notable among these problems are: fluctuate in and out of the laser beam. Ihis relative
motion causes them to absorb the laser light over a
• the inability to manufacture linear-based arte- comparatively large range of wavelengths. Ihe prob-
facts to identicallengths; lem can be rectified by substituting the iodine mole-
• transferring these measurements precisely to cules with a single atom of a metal such as ytterbium
employ them elsewhere; or strontium. Ihis atom is given an electrical charge
• instability of the material resulting from move- (changing its characteristics into an ion), which is
ment/creep as internal stresses are released and held in an "ion trap". Once the ion has been captured
gravitational and weight loss occurs with time; in this mann er it can be cryogenically treated
• part accuracies and tolerances becoming tighter (cooled down), so that its temperature creates an
with chronological time, necessitating an even almost stationary atomic motion and either an ultra-
more accurate me ans of establishing length. violet light or blue laser illuminates the ion. Ihis
single atom source is a considerable improvement
Ihis latter point was weIl known but needed a for the wavelength reference over that of current
new means to determine even greater accuracy and technique, as it absorbs light over a range of one
precision than was apparent using the International millionth that of iodine, giving the potential for
Prototype Metre and its contemporary artefacts vasdy improved performance in the future.
around the world. In order to minimise the problems
associated with metallic artefacts for standards,
during the early to mid-twentieth century another
natural standard was established for dimensional 6.1 Size and scale
measurement, namely the wavelength of light. Once
the metre had been defined in terms of the wave-
length of light from an atomic dis charge lamp, the In recent times, some form of length measurement
necessary instrumentation could be reproduced in has been used in every sphere of life, to enable fair
 111
trading conditions and the development of new and
improved products and processes to raise the stan-
dards of living of the populace. The range of preci- Metres
sion products is vast and covers items such as
nanometric electronic devices, with adjacent circuit
dimensions of approximately 0.000006 mm. Con-
versely, at the other extreme similar measurements
are required for geographical distances of many
kilometres having aresolution of just a few milli- • Three dimensional work-zone;
metres - this latter measurement accuracy was - an uncompensated CMM
necessary when railway tunnels were constructed • Accuracy of 25 mm hand held micrometer
and when the Channel Tunnel needed to meet in the • Three-dimensional work-zone accuracy
- compensated CMM
middle between England and France. Accurate
length measurement is a fundamental concept for
the vast majority of technological products, ranging 10'{} ~
from video recorder heads to automotive fuel
injector nozzles. 10"7 ~~ • Resolution - best air gauges, inductive
To gain a better understanding of the relative size sensors (i.e., variable areal
and scale of measurement at the nano-and micro-
technology end of ultra-high-precision manufac-
ture, Figure 196 has been included and illustrates • Resolution - best optical gratings
the generally accepted and logical divide between capacitive sensors (i.e., variable areal
these two technologies, nominally quoted as around
10- 8 m. In Chapter 3, describing notable technologies
and features of surface microscopy, it was shown that
instrumentation such as the atomic force microscope 1O"9~'§ ~Ei
-<... .-'"
Q

(AFM) could both resolve and, indeed, manipulate

structures in the atomic resolution range (10- 9 to
10- 10 m). This atomic resolution sits at a level of
ultra-high accuracy and resolution far below that of
10·\0
{ ~anostep resolution (i.e., stylus instrumentation)
• Resolution - best capacitive sensors
• Field emission tunnelling

visible light, which is nominally quoted as ranging

between the values of 10- 6 and 10- 7 m. In fact, until 10"\\

the middle of the twentieth century high resolution

was considered by many as the thickness of a hair
(around 10- 5 m), people quoting a "hair's breadth"
as being an exceptionally small measurement. If a
group of hairs are pulled from several people's
heads - the author's included - then it can be visibly Figure 196. The relative size and scale of things. [After McKeown,
appreciated (Figure 197a) that they vary consider- 1986.]
ably in diameter. The largest hairs in this group
are around 100 ""m and the smallest approximately
30 ""m; therefore, this term for a hair's diameter is
open to some degree of variability. What is apparent
when viewing Figure 197 collectively is that although
6.2 Predictable accuracy: its
their individual diameters might vary in size, all evolution
of these hairs have one constant feature: diametral
consistency. This natural sizing by our bodies is
to be expected, as our evolutionary development At the end of the nineteenth century ultra-precision
over millions of years has led to hair follicles being of engineering meant machining to around 0.1 mm
consistent size. The often quoted hair diameter tolerances; such tolerancing had steadily improved
is nominally agreed as being about 75 ""m, which to this level due to the earlier efforts of inventors in
gives an indication of its apparent size, but is subject the Industrial Revolution. Pioneers such as John
to a degree of uncertainty in its measurement, this Wilkinson (1774) produced the first boring mach-
being one of the main topics discussed later in this ines capable of producing abored hole of 1270 mm
chapter. diameter to an error of approximately 1 mm, while
Henry Maudslay (1771-1831) invented many
 InClustrial metrology , .

(a) Hairs vary remarkably in size - orten used as a stan- consistent parts more economically enables signifi-
dard of minute measurement (magnification x270) cantly higher levels of precision and the attendant
higher value-added tolerances, surface texture and
other exacting features to accrue.
In Table 18 the currently achievable tolerances by
contemporary precision CNC machine tools and
specialised associated equipment are listed.
When it is considered that Taniguchi made his
prediction of machined accuracy in 1974, he was
remarkably accurate in his extrapolated data for
normal, precision and ultra-precision machining
accuracies (Figure 198). The advent of the term nan-
otechnology, coined by Taniguchi in 1974, heralded
the concept of actually machining at nanometric
accuracies (10- 9 m), which is very dose to the atomic
lattice spacing. Today nanometric machining is a
reality, with components of quite large dimensional
size being held to diameters and surface textures of
(b) CIose-up of an individual strand of hair, iIIustrating nanometric accuracies and finishes respectively, or a
good diametrical consistency (magnification x1500) combination of both. Conversely, minute compo-
nents can be manufactured to within nanoscale
resolution and this technique is fast becoming
increasingly popular, as part miniaturisation be-
comes essential in many industrial applications.

Nanotechnology instrumentation
The general rule of thumb that inspection instru-
mentation should be at least ten times more accurate
than the equipment manufacturing the component
is exceedingly difficult to achieve at present for parts
made in the nanotechnology range. Nanotechnology
instruments have been developed to both explore
and measure surface properties in the vicinity of
the atomic range, typically investigating surface
topography, grain boundary coherence, atomic dis-
locations or similar - some of which were briefly
touched on in Chapter 3. These nanometric require-
Figure 197. Hair size has often been seleeted to illustrate the magni-
tude of size. [Courtesy of Jeol (UK) Ltd,]
ments exceed the concept of performance of dimen-
sional tolerances on a machine tool, or the attainable
surface text ure for apart. Notionally, the surface
production and its accompanying measurement can
precision machine tools, but particularly the engine be categorised as follows:
lathe. Notably others such as Sir Joseph Whitworth
(1803-1887) developed the 55° induded angle Vee- Functional performance for specific topographical
form screwthread, allowing precision feed motion to and surface features
be accomplished through suitable ge ar trains on
machine tools. These technological advances and • surfaces of magnetic heads and compact disks for
fundamental work by others, for example Joseph R. storage capacity;
Brown (1852), who designed a "dividing engine" for • coatings on cutting inserts and tools, for wear
the production of precision engraving, further properties;
enabled a wider range of machine tools, such as that • tribological investigation on bearings, or adhe-
by Eli Whitney (milling machine) to be designed and sion and adsorption problems;
developed, and then refined. With the considerable • surface effects produced by non-conventional
technological strides in precision engineering in the mechanisms (EDM, ECM, USM, laser and WJM,
last hundred years, today the ability to reproduce etc.);
 Quality and calibration techniques

Table 18. Acceptable tolerance errors in products

Tolerance Mechanical components Electronic components Optical components

(accuracy) (11m)

200 Normal machine and home-made General-purpose electrical Camera and telescope
parts parts (switches, motors and bodies
connectors)
50 General purpose mechanical parts Packages (electronic parts) micro- Camera shutters, lens
(gears and threads), typewriter motors, transistors, diodes and holders (for cameras and
and engine parts magnetic heads (video/type microscopes)
recorders)
5 Mechanical watch parts, accurate Electrk relays, condensers, disk lens, prisms, optical fibres and
gears, threads, machine tool memory, silicon wafers, TV colour connectors
bearings, ballscrews, rotary masks, video head and cylinder
compressor parts, shaver bl ades
0.5 Ball and roller bearings, precision Magnetic heads (video cassette Precision lens and prisms,
drawn wires, flapper servo-valves, recorders, CCDs, magnetic scales, optical scales, elastic deflection
gyro bearings, air bearings, quartz oscillator, magnetic bubble mirrors, IC exposure masks
precision dies, roll thread dies, memory, ICs, magnetrons, thin (photo/X-ray), laser polygon
ink jet noule film-pressure transducers, thermal mirrors, X-ray mirrors
printer heads, memory, electronic
video disks
0.05 Gauge blocks, diamond indentors, IC memory, electronic video disks, Optkal flats, precision Fresnel
high-precision X-Ytables, high- large-scale I('s, micro-vacuum lens, optical diffraction
precision stamping and dies, tubes, TFHCD gratings, optical video disk
microtome cutters (dia mond) (CD)
0.005 Ultra-precision parts (plane, ball, VlSI's, super-Iattices (synthesis) Ultra-precision diffraction
roller, thread) thin films gratings
(Special features) Shape (3-0) preciseness Pattern (2-D) fineness Mirror grating (1-0) accuracy

After Taniguchi (1994).

• damage monitoring of both pure and hybrid graphie al processes, a large number of instrumental
materials at the atomie scale; techniques have been developed to encompass these
• semieonductor and similar electrically based investigative procedures. Each instrument in this
surface properties, with respect to either charge large array of equipment needs to be calibrated at
injection or storage; periodic intervals to ensure that data obtained is
• surface catalysts and reagents concerning chem- traceable to known international and national stan-
ical processes and reactions; dards, so that some form of consistency in inves-
• surfaces of biological molecules in both liquids tigative procedure is assured. The following section
and membranes and their changes in real time. will explore both the underlying philosophy and the
route in achieving traceable measurements.
Specific measurement requirements

• height, or topographieal features;

• profile shapes and edge sharpness; 6.3 Traceability of
• position and relative position of features; measurement
• volumetrie analysis;
• flaw, or defect detection;
• structural characterisation of lattice parameters; In order to substantiate and validate any measure-
• atomie motion; ments taken on metrological instrumentation, it is
• time changes of atomic, or molecular structures. vital to ensure that these measurements are correct
and that they relate to international or national stan-
Due to the comprehensive investigative and diag- dards, or both. Prior to describing the mechanism
nostic assessments that are routinely required on for obtaining valid measurement from these instru-
such a diverse range of surface and subsurface topo- ments, it is worth pursuing the confusion that often
 Achievoble I
MaChining (systematic) error .• d ) • •
accuracy ( + (random) error. ,.(<1) Machlne :ools , ~e_asurl~g..!.!!.str,umen~
,. \. (processlng eqUipment) (Inspechon eqUipment)

um \.~o... 1 I ,Lathes. milling machines Vernier calipers

O.lmm. '1021
------~----------~~~r -T-----T;,.~;:;~~'~h:S~ :,~:g-m~c~,~:.---:.~:,:;:,~c~:'~'~O~S~ -
n '" ~~ I I lapplng machines. homng machines mlcrometers. dial indlcators

O.Olmm.IO' __\:E~~-- --x~~·~~ - - - _______ ~~ --:~~}I~ ---~_ --------------__ -------------___ _

r.;~ '- ~ ,5 m p generators) (Electric or pneumatic
~+ ~ , . te)
~~:q..~ II'"
" X ~C"<1-,,' III (CurvePrecislon jig borers. mlcrome rs
~ "<r". "~,?_ ':'': jig grinding machines. .

t"m. I
____ ~C'::,,>~ -x~~~- __ ~:1~. _---L __ ___ ~ ~~~~~~~_ :u~~ ~~s~n~ ~~~~~ ______O~~c~I~~~a~a~o~s____ _
OI,.~ (\.~~:9 I ' , Preclsl.o~ gnnd,~g machl~es.
': Optical or magnetic scales
">~ 1> " l I
precision lapplng machlnes, electrical comparators
"1", I 1 : ' , (opticallens grinding machines)
~? X I: I Electronics comparators
"...
______ -Allowab~ ~__ ':'~I'_"2. ~~1~~ ___ :______ ~r:.c~.i~n_d~~o~~~~~ __ j~<3~a5.t~_._____ _
0.lpm.l0- 1
achi 'ng accurac y "'..... I 'O.051'm
-N-J
I (D.U.V. mask ahgner). laser measunnglnslruments
~rror~) (r) : . . . : (opticallens finish grinding machines). Optical fibers, Talysurts,
,-..-f-L---.j I :', 'precision diamond grindlOg Talyronds
I ' I " ..
(systematic I" mac h'Ines, uItrapreclslon
0.0Ipm.10 4 --: ~ errors «(A» --- -~N--
~ -'-'-~~':'O_I~~_g.!:.i~d~[~~c~i~~ -- --- - - - - -- - - - ------
.... a. I '0. 005 m I" (diHraction grating ruling engine)
p High precislon laser
I101:)
, ... 41 random errors «(.)
I
I I
--- I ,
,(electron beam hthographer)
. . •
measunng Instruments
:.8.~ / ' I " " Super high precision grindingmach. (Doppler,
I § :il . : I ',: ' super high precision lapping multi·reflection)
:z E Vartance (2<1) I I ' .... I and polishing machines Talysteps
O.OOI"m. 10-3
1nm. --, - - -;- --:0,0011';;,1:::::- ÄiOm,-mC1~cule-örion oeam -SCanning electrö'n" mi~oscopes,
" , .......... machlOlng. atom or. . transmission electron microscopes.
O.3nm ..... -;tIl • 'I =". - - - - , - - - - - - .1......
I' - - - - ........ -molecule de~sltlon
- _________ el e ctr on d'ff .
I rac t'loneqUlprnen. t
Average Dimension : I : .... , . ion analyzers.
(Atomic I Specified
dimension I I I (Substance syntheslzlOg) X.ray miero analyzers
lattice dirrynsion, , n I I I I •
distance) 1900 1920 1940 1960 1980 2000 year Auger analyzers, ESCAR

Figure 198. The prediction of achievable accuracy. [After Taniguchi 1974.]

 ABSOLUTE Definition of the metre
ABSOLUTE

Realisation as an iodine-stabilised
10- 11 laser wavelength

NATIONAL PRIMARY STANDARD

10-9 Calibrated Zeeman-stabilised

lasers

NPL Primary Gauge Block NPL Primary Length Bar

10-8 Interferometer Interferometer
05-300mm 100-1500mm

Major calibration laboratories' Major calibration laboratories'

UK Accreditation
master standards master standards
Service
(UKAS)
05-300mm 100-1500 mm

Gauge blocks and lengths bars used as standards, e.g. validation of

coordinate measuring machines, transfer standards, calibration standards

Figure 199. Traceability of end standard length calibrations. [(ourtesy of NPL.]

arises over the subject of accuracy and repeatability, by the manufacturer of the instrument is not wise,
as they do not have the same origin; hence: without.ascertaining this fact for oneself. Returning
to the current theme of traceable measurement, if
• repeatability is a function of the instrument's the traceability chart illustrated in Figure 199 is
design; considered, the route from metrological and inspec-
• accuracy is a function of calibration. tion equipment in daily use, through the traceablity
chain to the primary standard of dimensional
For example, an instrument may repeat with great measurement, can be shown. In each case, as cali-
precision, but it is only by calibrating the equip- bration progresses toward the primary standard,
ment that we know whether to believe the results. more exacting measurements are made, thus
This seems an obvious point to make, but it is often ensuring that each piece of instrumentation in the
a widely misunderstood distinction that is very "chain" is calibrated to a high er order of accuracy
important whenever a new instrument is purchased. and precision. This process of calibration against
To take for granted the positional accuracies claimed known artefacts is both costly and, more important,
a time-consuming process, taking valuable and often proliferation an alternative option might be to
delicate instrumentation out of the metrological describe the surface by way of its Fourier compo-
chain, but is essential to ensure traceability. nents. Anyone familiar with the complex waveforms
Today, many of the primary standards are now existing in acoustics will be able to comprehend
"derived" and do not exist as discrete entities; this is Fourier concepts. For instance, a musical instru-
particularly true for many electrical and physical ment's range can be described by its relative prop-
standards. As a result of such "pseudo-standards", a erties at different frequeneies in a harmonie series,
move in recent years in the United Kingdom has in the same manner that areal surface is described
been an attempt to mitigate the losses in cost and in terms of a succession of sinusoidal corrugations,
time as essential calibration occurs, by employing each having particular frequeneies, amplitudes and
virtual gauging concepts, which have yet to be intro- phases. When a surface has a periodic profile, this
duced into conventional dimensional measure- can be described by sine waves of infinite series,
ments. By way of illustration of the potential for this whose actual periods are integral fractions of the
technique for dimensional measurement, that of an fundamental; alternatively, if periodieity is not
electrical calibration will be briefty reviewed. Here, represented on the surface, then all spatial frequen-
users obtain calibration through on-line Internet eies are probably present. If for simplieity, consider-
access to the National Physical Laboratory (NPL), by ation is given to the cross-section of a random
way of its appropriate website - via encrypted se cu- surface, two facts can be established; these are:
rity stages.1t is important that in order to gain access
to the calibration facilities the sensitive nature of the the surface can be represented by a graph of
data to the site must be secure. Once access has been height -versus-position;
confirmed, users then enter information such as the 2 another graph can be made to represent ampli-
type of equipment they wish to calibrate, together tude versus spatial frequency.
with the calibration kit and frequency ranges. This
data is transferred to the NPL, where it is sub se- The relationship between both graphs is in effect
quently evaluated and the correction values are a Fourier transform. To obtain a complete descrip-
displayed on the PC's screen. Thereafter, a need to tion of this arbitrary surface, consideration must
evaluate any imperfections that can arise via elec- be given to the relative phases of the individual com-
tronic noise and other outside approval by the user ponents. In effect this is the power spectrum, but
needs to be substantiated, with the values sent back as the power is proportional to the square of the
to the instrument, which will automatically calibrate amplitude it will not have any phase information.
itself. A major benefit in utilising the Fourier transform
Although Figure 199 shows the traceability route for surface description is that surface data can be
to be relatively straightforward, in reality the devel- simply extracted and as a consequence metrology
opment of the route to effieient calibration can equipment can be more easily assessed as the instru-
involve some quite complex practical problems and mental response function, or more notably on Sted-
philosophical interrelationships, as depicted in man diagrams, yet to be described.
schematic form in Figure 200. The traceable route from the concept of an ideal
sinusoidal surface to its practical realisation, and
from there to more complex surfaces to provide
Surface description standards against which to compare real surfaces of
functional performance, is depicted in Figure 200.
Any given surface can be described in a variety of Achieving this goal represents considerable seien-
ways (Figure 200), the most obvious being where the tific research in both fundamental and applied fields.
height of each data point is plotted in relation to a Even when much practical metrological work is
reference plane. For example, in two-dimensional undertaken, it cannot be guaranteed that it will be
sectioning this is what the surface texture instru- possible to accomplish instrumental calibration to
mentation would in the main be measuring. How- the level of accuracy one might expect, nor even that
ever, such information tends to be quite speeific and this accuracy will co in eide with the sensitivities of
relates to both the surface stylus probe interaction typical instruments. However, it should be possible
and the measuring conditions. These inspection to establish uncertainty levels and, in the majority of
parameters make it problematical to extract from instances, the causes of this uncertainty. Further, it
the surface the required information to enable com- cannot be certain that this exereise in establishing
parison of surfaces to each other. This problem has uncertainty will lead to improvements in measure-
led to roughness parameters considerably multiply- ment determination, but it enables the metrologist
ing in number, many with quite limited practical to obtain a clearer understanding of the present
applications. In order to minimise this parameter instrumentallimitations. The following section
 Quality and calibration techniques

T'-><etJcaI 2 HarmorllC'
In>,runenl
Respons<>
Furocbon
NV\

iU
VVVVV

f\N\
WlII't!tenglh
~

'\IV\
~ A.f\.A
'3.
~ .~ .. I
Sloo; ~E~S I
Wavelenqlh

.
ModoH",'l

Expenmenl

Figure 200. Route towards absolute standards for measurement of surface texture. [(ourtesy of R.K. Leach & M.C Hutley/NPL.]
briefly describes how, and in what manner, the influ- has a measurable output and operating under a
ence of process capability on the production of stable set of conditions causing the output to vary
components by industry is affected by calibration about a central value in a predictable manner. Such
and recalibration of inspection equipment. Measure- an output is depicted in Figure 201, where the process
ment uncertainty will then be described and how it capability (Cp ) index is initially estimated at the
can be determined, enabling instruments, arte facts beginning of a batch run to have a value of <1.0.
and their users to establish the probable causes of Process capability is an extremely important meas-
metrological uncertainties. ure of production performance and can be simply
expressed in the following manner:

Process capability factors

The output from a production process will differ
according to the stochastic nature of the operation. where: Cp is the process capability, Tu is the upper
A stochastic process can be defined as: a process that specification limit (tolerance), TL is the lower speci-

Gaussian distribution

Lower Higher
specification specification
limit limit

\\ \
\\
\\ \ \
Time
\\
\\ \
\
\
I ,

I ~>U I
~-----~
Figure 201. The process capability (9 index may change with time.
 (a) Manufacturing process mean may change with a systematic error present

TIME
-<:::JI-----
Drift in the process (i.e., systematic error present) .A-"""".."....,~

Scrap

Recalibration
required

Drawing tolerance

TIME
4
(b) Variable manufacturing output Variable output
resulting from changes in the
process capability index
Scrap Scrap
_=::s:s5~~~~~~~~~w.::;~~:o::.- Cp =0.75

Recalibration
required

...::::::J-*",=i~"'>u''>'''>'~''':'':~'":'-'~~-::::-H~ Cp = 1.33
Drawing tolerance

Figure 202. Illustration showing how (al the manufacturing process mean changes with time; (bl process capability index changes with time;
or how both can change with time.
fication limit (tolerance) and 0" is the standard has the output situated at the me an, but with time
deviation for the process. the variability has increased from an initial level of
In this case the Gaussian (normal) distribution Cp 1.33 to an intermediate value of Cp 1.0, where cali-
from the output at the beginning of the pro duc- bration is necessary. Finally, toward the end of the
tion run is centred on the me an value (x bar) and run, an undesirable situation occurs at a Cp value of
fills the complete statistical band (60").As the process 0.75, where scrap will be present.
continues to produce components the output is still If some form of in-process gauging and associ-
centred on the mean, but the spread (variability) has ated traceable calibration procedure had been
increased considerably with respect to time. Here is introduced to limit the "process drift" or variability
an undesirable situation as the process capability in output (or indeed both), then scrap parts are
has decreased to a value of Cp 0.8; this will inevitably unlikely to be present in a batch production run.
introduce scrap into the process. Some reworking What has been said here for the control of produc-
may be possible for certain parts outside one limit, tion processes is equally true for the calibration of
depending upon whether this is a non-critical im- general metrology equipment, through acceptance
posed limit in the process. However, the process of the fact that uncertainty will arise in any produc-
capability has highlighted the fact that in this case tion or metrological process requiring appropriate
there is an unacceptable variability occurring that action to minimise these effects.
needs drastic action if most of the production run
is not to be lost. What was unusual about the pro-
duction output in Figure 201 was the fact that the
mean value had not changed, but the variability had 6.4 Measurement uncertainty
increased. In the case of Figure 202(a) the mean
value has changed with respect to time - it has intro-
duced some "drift" as a systematic error in the Geometrical product specification (GPSj
production output has proportionally increased.
This drift has caused the process capability to The GPS procedure enables the designer to define
change from an initial value of Cp 1.33, being well in the geometrie shape, dimensions and surface char-
control, to an intermediate value of Cp 1.1, indicating acteristics of a component in a way that ensures
that calibration was required before this point had optimum functioning of that workpiece. The proce-
been reached in the production run, as some scrap dure normally adopted includes adefinition of the
is now inevitably being produced. The situation is optimum value and its dispersion for its intended
exacerbated with time until a Cp 0.90 production function, enabling it still to be satisfactory. The
output occurs, indicating the whole process is out production process will in many circumstances
of control, because of this tendency to drift. In the produce parts that are not perfeet, in that they may
case of Figure 202(b) once again the production run demonstrate some form of deviation from a defined

FUNCTION

COMPARISON

[Designer)

Specification Measured
(Drawing) Values

[Industrial
Engineer)

Extracted
Data/Signal

Figure 203. The basic geometrical product specification (GPS) framework: a new approach.
 Quality and calibration techniques

Table 19. The geometri(al produ(t specifi(ation (GPS) matrix model

Matrices: (hains of standards:

General GPS matrix Complementary GPS matrix

General GPS A Pro<ess specifi(, 8 Machine element,

tolerance standards geometry standards:

1. Size (hain A1 Machining (hain 81 Screwthread chain

2. Distan(e (hain A2 Casting <hain 82 Gears <hain
3. Radius (hain A3 Welding <hain 83 Spline <hain
4. Angle <hain A4 Thermal <utting <hain
5. Form of line (independent of a datum) (hain A5 Plastics moulding (hain
6. Form of line (dependent on a datum) chain A6 Metallic and inorganic coating chain
7. Form of a surfa<e (independent of a datum) <hain
8. Form of a surface (dependent on a datum chain
9. Orientation chain
10. Lo<ation (hain
11. Circular run out chain
12. Total runout chain
13. Datums <hain
14. Roughness profile chain
15. Waviness profile (hain
16. Primary profile (hain
17. Surfa<e defects chain
18. Edgeschain

50urce: B5 8888:2000 /150 TR 14638.

optimum. If a eomparison is undertaken between and the associated geometrie toleranees are defined.
the eomponent and its specifieation (Figure 203), it In box 1 the eodifieation of the geometrie toleranee
requires the following to be eonsidered: that the symbol for straightness with its associated toleranee
workpieee (BS 8888: 2000): is depieted. Box 2 indieates the definition of this
straightness toleranee and box 3 is the specifieation
• is eoneeived by the designer; operator, indieating the manner in whieh this
• is as manufaetured; straightness deviation may appear within the toler-
• is as measured. anee. Box 4 illustrates the anticipated uneertainty
associated with the upper and lower toleranee limits;
The standards introdueed in the GPS field more will be said on this subject later when
provide the fundamental rules for geometrieal speei- diseussing Figure 206. Boxes 5, 6 and 7 illustrate the
fieation, such as the basic definition, symbolic repre- metrologieal assessment of the real surface and the
sentation and measurement principles (see ISO metrology instruments with their antieipated eali-
TR 14638 for an overview of these eoneepts). Several bration requirements. In this visual manner these
eategories of the standard are eoneerned with boxes depiet the eomplete geometrie produet speci-
eoneepts, dealing with fundamental rules of specifi- fieation ehain, this being presented in a logieal and
eation, others provide global principles and defini- praetieal approach to the topie.
tions, or various geometrie eharaeteristies, the In the GPS philosophy the specifieation and
latter including workpieee eharaeteristies relating verifieation operators are slightly more complex than
to differing production processes, together with first appearanees would seem to indieate, as illus-
specifie maehine elements, as shown in Table 19. GPS trated in the fiow diagram in Figure 205. The linking
prineiples are neeessary throughout a produet's of design intent and metrology is the objeetive here
development: design, manufaeturing, metrology and - the duality principle, with geometrie features
quality assurance. oeeurring aeross three disciplines. These are:
In Figure 204 one of the "ehains of standards" is
depicted, illustrating a typical eomposition of a specijication - this is where the designer eon-
funetional performance standard, such as the eeives several representations of the workpieee;
geometrie toleranee for straightness. In boxes 1, 2 2 workpiece - as it would appear in the physieal
and 3 are shown the manner in whieh straightness world;
 3 4 5 6 7
1 2

Comparison Measurement
Codification Definition Specification Verification Calibration
with tolerance equipment
on a drawing of tolerance Operator Operator requirements
limits requirements

I-I I ~ t------>- m m
0.1 ~ I
UU
1

I I I
UU
I

I I
es:- )L ~
Figure 204. The (hain of standards, e.g. straightness.

3 inspection - where a given workpiece repre- width of the margin - its interval - the second is the
sentation is created by sampling of the part by confidence level; this latter value states how sure one
measuring instruments. is that the actual value oceurs within this margin.
Typically, a stylus length might be 100 mm plus, or
To comprehend the nature of the relationship minus 0.5 mm at the 95% confidence level. This
between these three disciplines is of great impor- uncertainty could be expressed as follows:
tance and Figure 205 depicts the parallel or
"mirror" activities that connect those between the Stylus length = 100 mm ± 0.5 mm,
design intent and metrology. The design intent will
result in specification characteristics - the concep- at a level of confidence of 95%.
tual perfect operator - with the metrology creating In reality what this statement is implying to a
the evaluation of characteristics - the actual oper- metrologist is that they are 95% sure that the stylus
ator. These "operators" can then be compared length williie between 99.5 mm and 100.5 mm in
with each other for comformity, as indicted in length. In this discussion it is essential not to confuse
Figure 205. the term uncertainty with error. They can be
Some of the key questions raised today by the distinetly dassified thus:
manufacture of parts to tighter tolerances are:
• errar - the difference between the measured value
• "What is the best manner to interpret metrology and the "true value" of the item being inspeeted;
data, whenever there is a measurement uncer- • uncertainty - the quantification of the doubt
tainty factor in a tolerance band?" existing with regard to the result of this measure-
• "How to standardise on tolerances across the ment.
design, production and inspection functions, so
that measurement uncertainty is accounted for In most situations one tries to correct for any
throughout the creative procedure in the prod- known errors; this is attempted by the application of
uct's design and build life eyde?" corrections from certificates of calibration, although
any error occurring that is not known will be a
The term uncertainty has been mentioned here source of uncertainty. The estimation of measure-
on numerous occasions and this will be the main me nt uncertainty is important because precise and
theme from now on in this section. accurate quality measurements enable the metrolo-
gist to comprehend the results without ambiguity.
However, the expression of the measurement uncer-
Uncertainty issues tainty is significant when taking any form of
measurements, because they may be part of:
The question often asked in calibration-related
tasks, is: "What is measurement uneertainty?" • calibration - where measurement uneertainty
Uncertainty of measurement refers to the doubt must be reported on appropriate certification;
that exists about any measurement; there occurs a • test - if measurement uncertainty is required to
margin of doubt for every measurement. This establish either a "pass" or "fail" condition.
expression of measurement uncertainty raises other
questions: "How large is the margin?" and "How bad Alternatively, information on measurement uncer-
is the doubt?" Henee, in order to quantify uncer- tainty is vital to hold a:
tainty two numbers are required, one being the
 SPECIFICATION VERIFICATION
OPERATOR: OPERATOR:

SKIN MODEL REALSURFACE

Geometrical representation Set of features which
(infinite set of points) physically exist

OPERATION OPERATION

- Partition - Physical partition

- Extraction - Physical extraction

Filtration Difference Filtration

contributes to
Association Uncertainty - Association

Collection - Collection

- Construction - Construction

Measurand Measured value

Characteristics Characteristics
spccification evaluation

Comparison ~

for conformance

Figure 205. The "duality principle": the link between design intent and metrology [BS 8888: 2000.]

• tolerance - where the uncertainty had been control has been established, statistical measures are
previously established, prior to deciding whether adopted. If the metrological process is not inftu-
the tolerance is met (see Figure 206). enced by systematic or random errors, then the
process is said to be behaving "normally" and any
Otherwise one might need to read and under- process output data is valid. Two statistically derived
stand the calibration certificate or written specifica- mathematical expressions are needed to define
tion for a particular testlmeasurement. It is worth whether a process is behaving correctly: these are
noting, that a measurement is not traceable unless the arithmetic mean and its accompanying standard
quoted wiith an uncertainty. deviation. Normally, the arithmetic mean is conve-
niently shortened to mean, this being denoted by the
symbol x (termed "x-bar"). This x value is the mean
Statistical measures of all the values of x and can be derived from the
following expression:
In order to evaluate the test data from an inspection
procedure, to determine whether adequate process
 Lower Upper
Specification Specification
Limit: Limit:
I
Specification Zone
[OUTof [OUTof
[IN specification] specification]
specification]

DESIGN/SPECIFICATION
PHASE

VERIFICATlON
PHASE

U U U U
Non-conformance Non-conformance
Uncertainty Conformance Zone Uncertainty Zone
Zone

Figure 206. Comparison of the measurement value with its associated tolerance.

1 The spread of these readings will give an indication

:x = -
n
2:n

i~l
Xi of the uncertainty of these measurements. Moreover,
knowing how large this spread is enables one to form
where X = arithmetic mean, ~X= sum of X and n = an opinion as to the quality of the inspection pro ce-
number of readings. dure or g.roup of measured data. The customary way
NB Sometimes several readings are taken, which to quanhfy the spread is by utilising the standard
modifies the calculation to deviation - often described by the Greek lower case
letter (J" (sigma). Usually only a moderate number of
measured data is obtained and, in this case, an esti-
mate of the standard deviation can be derived which
is denoted by the letter s, meaning the estimated
where ~fx = the sum of the frequencies of x. standard deviation. In principle the calculation of s
For example in roundness measurement, if 10 for the previous series of LSC roundness polar plots
LSC roundness polar plots were obtained in the would be
course of an inspection procedure - 17, 18,22, 18, 18,
17 + 18 + 22 + 18 + 18 + 19 + 21 + 19 + 18 + 20
19,21,19,18 and 20 /-Lm respectively, then the value x (/-Lm) = 19 (mean)
would be
The next step from the calculated mean is to find the
~x = 17 + 18, + 22 + 18 + 18 + 19 + 21 + 19 + 18 difference between each successive readingj this is
+ 20 (/-Lm) obtained in the following manner (by making the
= 190/-Lm mean LSC value of 19 /-Lm equate to zero):

Therefore -2 - 1 + 3 - 1 - 10 + 20 - 1 + 1

Then each of these values is squared as follows:

190110/-Lm
19/-Lm 4191104011
Next, sum the total, then divide by n -1; here, in the Therefore
case of n this is 10, so n - 1 is 9; hence:

(4 + 1 + 9 + 1 + 1 + 0 + 4 + 0 + 1 + 1)/9 = 22/9 NB The tolerance represents the specification limits

=2.44 J-Lm 2 for the production process.
A term to denote the relationship between the
The estimated standard deviation (s) can be ob- tolerances and their associated process capability is
tained by taking the square root of the previous the relative precision index' (RPI; see Figure 202).
answer; therefore The relationship between the tolerance and 6s (that
is, (3s), gives rise to three levels of precision for the
process; these are:
s '112.44 J-Lm 2
1.6 J-Lm (correct to one decimal place) low relative precision - where the tolerance band
is less than or equal to 6s (representing a value of
The overall process of calculating the estimated <1);
standard deviation for aseries of n measurements 2 medium relative precision - where the tolerance
can be mathematically expressed in the following band is greater than 6s (value of 1), but less than
manner: a value of 1.33 (a value of >1 but <1.33);
3 high relative precision - where the tolerance band
is greater than 1.33 (representing a value of > 1.33).
s =~_1 ~(X-X)2
n-14 z A distribution that equates to a high RPI that
i=l
is "centred on the mean" will be unlikely to pro-
where Xi is the result of the ith measurement and duce scrap parts, as any process drift introduced
X is the arithmetic me an of n results considered. by systematic or random errors are unlikely to pro-
To ensure that the production process was cor- duce components out-of-tolerance. Today, with
rect1y operating, this estimated standard deviation many companies artificially introducing RPIs that
s can show whether a process was in control or not. are close to or exceed values of 2, with these being
If plus or minus three s is utilised in conjunction "centred", there is quite a low probability that any
with x (mean), then if the process is normally dis- components from such a production process will
tributed all the LSC polar plot readings should fall produce scrap - unless this process has become
within the upper and lower values, because this completely out of control through a fundamental
represents 99.7% of the overall population. Hence, change in its output.
the lower and upper limits for these LSC polar
plots are Origins of uncertainties
x ± 3s = 19 ± 3 . 1.6 J-Lm There are numerous aspects that can contribute
to poor measurements, many are easily established
= 19 ± 4.8 J-Lm
while others remain more difficult uncertainties to
14.2 (lower) and 23.8 (upper) J-Lm determine. Due to the fact that actual measurements
are never taken under ideal conditions, any errors or
Therefore, with the lowest reading obtained being uncertainties can be the product of a wide range of
17 J-Lm and the highest reading 22 J-Lm, these read- contributing factors (see Figure 207), such as:
ings would fall within, say, some pre-selected toler-
an ce set by the designer which might be an LSC of • metrological equipment - the instrumental prob-
14 J-Lm and 24 J-Lm, respectively. lems can accrue due to drift, wear, ageing and
One further way of expressing these values is to many more problems;
employ its associated process capability index, • component inspection - which may not be stable,
usually denoted by the Cp, which indicates the spread an often quoted analogy of this problem being to
of the process. This can be derived in the following attempt to measure the size of an ice cube in a
manner: warm room;
• measurement process - the practice of measuring
overall tolerance a component may be difficult to achieve, this
Process capability Cp being particularly true if the subject of the
± 3s (that is 6s or 60-)
measurement is in dynamic motion as readings
24 - 14/9.6 are taken;
 MATERIAL MACHINE MEASUREMENT ENVIRONMENT
(Componenl) (Exlernal Noises)
(Hard/Software)
11
Geometrie behaviour Operator interference
Mounling Improper algorithm Thermal effeet (On inslrument)
Loading
Dynamie behaviour Ground-borne vibration

Imperfect programming Thermal effeet (Componenl)

Form and surface
Kinematic behaviour Air supply (As applieable)

Dala reduclion Temperalure variation (Ambienl)

Malerial properlies Thermal properties Electricity supply

Thermal properlies Truncated error Relalive humidity
StyluslProbe -
Sub systems

Harmonics/Filtering
Definition of measurement Reconstruction - component geometry
Probe/Stylus type
StyluslProbe error correction
Reference Artefacts
Location and plane
Probe/Stylus lenglh
Deformation compensation

Sampling rale
CUI-off/window
Temperature compensation
Datum sclection Correction - geometrie imperfection
Sampling algorilhm
Calibralion proeedure
Skid/Skidless operalion

MEASUREMENT MECHANISM MEASURING STRATEGY

(Operalion & Calibralion) (For eorreetion)

----- --------
Figure 207. "Cause-and-effect" (6M) categorisation for measurement uncertainty - generalised causation - in surface texture and roundness instruments. [Adapted fram Stout, et al.l
 Quality and calibration techniques

• imported uncertainties - any metrological instru- Table 20. Determination of measurement uncertainty
ment has its own uncertainty and this affects the
subsequent measurements taken; Step Procedure

NB If an instrument has not been previously cali- Decide what you need to find out from your
brated, then in this condition it will exacerbate any measurements. Decide wh at actual measurements and
future measurements taken. calculations are needed to produce the final result
2 Carry out the measurement needed
• operator skill - occasionally certain measure-
ments rely on the operator's skill and judgement, 3 Estimate the uncertainty of each input quantity that feeds
into the final result. Express all uncertainties in similar
with some operators being more efficient and
terms (see following text)
reliable than others; for example, when timing a
trial with a stop watch the level of uncertainty 4 Decide whether the errors of the input quantities are
depends upon the operator re action time; independent of each other, but if not then additional
calculations/information is required
• sampling concerns - any measurements obtained
must be representative of the process being 5 Caiculate the result of your measurement and include
assessed. known corrections from previous calibration, ete.
6' Find the combined standard uncertaintyfrom each
NB As examples of these sampling issues, any individual aspect
temperature measurement should be assessed dose
to the item being monitored and not away from its 7' Expressing uncertainty in terms of coverage factor,
together with a size of uncertainty interval and, state
local thermal environment. When selecting sampies confidence level
from the production line, ensure that they do not
originate in the initial "start-up" batch run, nor that S' Write down the measurement result and its uncertainty,
they are obtained from a mixed stillage from two stating how both were obtained
separate production processes.
a To be discussed later in more detail in this section.

• Metrological environment - the measuring instru- Source: Beginner's G.U.M. Bell/NPL, 2000.
ment can be significantly influenced by either the
component being measured or variations in tem-
perature, humidity, air pressure, possible vibra-
tional effects and many more conditions.
• Type B evaluations - uncertainty estimates
NB The overall measurement uncertainty will be obtained from any other relevant information,
affected by the individual uncertainties, and where such as from calibration certificates, manufac-
the error is known - that is, from a calibration turers' specifications, published information, past
certificate - a correction can be applied to the result experience or just common sense.
of this measurement.
NB It is tempting to think that Type A uncertainties
occur by random processes and Type B uncertain-
Calculation of measurement uncertainty ties result from systematic factors, but this logic is
somewhat questionable.
Prior to the calculation of measurement uncertainty, Table 20 attempts to illustrate how to use infor-
initially the sources of uncertainties need to be mation from both Type A and Type B evaluations for
identified. Once this task has been performed then an overall estimate of uncertainty.
an estimate of the size of the uncertainty from Any contributions to uncertainty must be ex-
each source must be established. Lastly, each of the pressed in the same units before they are combined,
individual and contributory uncertainties is then furthermore having matching levels of confidence.
combined to produce an overall figure for the whole The contributing uncertainties and their respective
uncertainty. There are two methods used to estimate confidence levels need to be uniform and this allows
uncertainties: these are termed either Type A or Type them to then be converted into standard uncertain-
B evaluations, and in many instances uncertainty ties. Plus or minus one standard deviation is the
evaluations of both types are required, their identi- margin for a standard uncertainty; moreover it also
fication being according to the following: gives information on the uncertainty of an average
and not just the spread of these values. It is normal
• Type A evaluations - estimates of uncertainty by to quote standard uncertainty using the symbol u,
statistical techniques, normally from repeated or as u(y), this latter term being the standard uncer-
readings; tainty in y.
"Type Aevaluation": standard Addition and subtraction: summation in
uncertainty calculation quadrature
Once aseries of measurement readings has been The most elementary case for this technique is
x
obtained for a "Type A evaluation", the (mean) where the sum of aseries of measured values - by
value and s (estimated standard deviation) can be either addition and subtraction - is calculated. By
calculated. From this set of readings the estimated way of illustration of the summation by quadrature
standard uncertainty u of the mean can be calcu- method, if it is required to determine the total height
lated from the following: of a group of "wrung-together" gauge blocks of
differing individual sizes, then the standard uncer-
u = s/--Jn tainty of each gauge block could be represented by
the notation a, b, c, etc. The known (measured)
where n = number of measurements taken. values are then found by the combined standard
uncertainty technique, by squaring the individual
uncertainties, adding them together, then taking the
square root of the total, as follows:
"Type 8 evaluation": standard
uncertainty calculation Combined standard uncertainty

If information is somewhat scarce, typically for esti-

mates of some Type B values, it might be possible to
establish the upper and lower limits of uncertainty.
If this action is taken, then it could be shown that
these limits might equally fall anywhere between
positions along the uniform distribution, or for that
Multiplication: summation in quadrature
matter in a rectangular distribution. In this latter
In cases of more complex uncertainty problems, an
case the standard uncertainty for a rectangular
approach adopted to simplify the calculations is to
distribution will be work in terms of either the relative or fractional
uncertainties. For example, in the simplest form of
u = a/--J(3) calculation, if a large rectangular surface area A has
to be established, then it is simply determined by
where a = the semi-range (half-width) between the
multiplying its length L by the width W, that is:
upper and lower limits.
A = L x W. The fractional uncertainty in the area is
In most cases, for a calibration certificate of a
established from the individual fractional uncer-
measuring instrument it can usually be assumed tainties for the length and width. Therefore in the
that any uncertainty indicated here will have been
case of its length L with an uncertainty of u(L), the
derived from a normally distributed set of statistical
relative uncertainty will be u(L)/L, likewise for the
data, although it is not uncommon to obtain data width W the relative uncertainty is u(W)/w. Hence,
from either a uniform or rectangular distribution.
in the representative area the relative uncertainty
u(A)/A is given by

Combining standard uncertainties u(A)

Whether the individual standard uncertainties were A

calculated by either Type A or Type B evaluations is
not an issue, as they can be validly combined by the
mathematical techniques known as summation by
quadrature, commonly termed either the root sum of Complicated functions: summation in
squares or root mean square. The result is known as quadrature
the combined standard uncertainty and is denoted
by the notation U c or uJy). In many instances in the calculation of the final
result, the value will be squared (Z2)j here the rela-
tive uncertainty due to this squared component
takes the form 2u(Z)/Z.
In some instances measurements taken are calcu-
lated using formulae that employ combinations of
 111
addition, subtraction, multiplication and division, can be achieved by using a coverage factor (k).
etc. For example, in some "uncertainty budgets" Multiplying the combined standard uncertainty (u c )
(more to be said on this topic shortly) it might be by the coverage factor (k) results in what is termed
necessary to measure the electrical resistance Rand the expanded uncertainty - normally denoted by U
voltage V and then calculate the resultant power P, in the following manner:
utilising the relationship:

V2
P=-
R NB A specific value of the coverage factor (k) will
introduce a particular confidence level for the
expanded uncertainty.
In this case the relative uncertainty u(P)IP in the In particular, if the overall uncertainty is scaled
value of power would be given by by utilising a coverage factor k = 2 to obtain an
approximate level of confidence of 95%, the k = 2
u(P) value is acceptable when the combined standard
P uncertainty tends to be normally distributed. Some
typical coverage factors for normal distributions are
In general, in any multi-step calculations the
procedure for combining standard uncertainties
k = 1 for a confidence level of around 68%;
k = 2.58 for a confidence level of around 99%;
in quadrature can be achieved in multiple steps,
utilising the relevant form for addition, multiplica-
k = 3 for a confidence level of around 99.7%.
tion, ete., at each step. For complicated formulae the Other less popular distribution shapes have a
combination of standard uncertainties is more com- different range of coverage factors. Alternatively,
prehensively discussed elsewhere (e.g., the UKAS whenever an expanded uncertainty is quoted, the
Publication M 3003). standard uncertainty can be found by reversing the
process, namely dividing this uncertainty by an
appropriate coverage factor. On calibration certifi-
Correlation cates the quoted expanded uncertainty, if expressed
properly, can be "interrogated" to obtain the
The equations required to calculate the combined required standard uncertainties.
standard uncertainties above are only valid if the When expressing the measurement uncertainty,
input values for standard uncertainties are not inter- it should be exacting in its definition, allowing no
related or correlated. Correlation refers to the ques- misinterpretation. To achieve accurate uncertainty
tion of whether all the uncertainty contributions are definitions, several important factors need to be
independent. Namely, could a large error in one mentioned:
input measurement cause a similar sizeable error
in another? Further, might some outside environ- • measurement result and uncertainty figure - an
mental inftuence such as temperature create a arte fact was 300 mm ± 1 mm;
similar effect simultaneously on several aspects of • coverage factor and confidence level - recom-
uncertainty, which are either easily established or mended wording to minimise confusion is: "The
require further investigation by to the metrologist? reported uncertainty is based on a standard
In many instances these individual errors are uncertainty multiplied by a coverage factor k = 2,
independent, but where they are not additional providing a level of confidence of approximately
calculations are required. 95%";
• description of how uncertainty was estimated -
reference to the appropriate documentation.
Coverage factor k
In the previous discussion on uncertainty the
components were invariably scaled to obtain the Analysis of uncertainty: Huncertainty
combined standard uncertainty. This result of the budgets"
calculations for combined standard uncertainty may
be thought of as equivalent to "one standard devia- To facilitate the process of calculating uncertainty, it
tion", but it may be necessary to obtain the overall is often helpful to summarise the uncertainty
uncertainty expressed in terms of a different level of analysis - uncertainty budget - in a spreadsheet, as
confidence, typically the 95% limit. Such rescaling indicated in Table 21.
 InClustrial metrology . .

Another approach, which is similar in concept to measurement uncertainty is small when compared to,
that depicted in Table 21, is to establish the applica- say, the inspected diametral tolerance, then the
tion of uncertainty budgets using spreadsheets, as metrology equipment can be used with confidence for
depicted in Table 22. Assuming that the input quan- the stated inspection procedure. In the case cited
tities have been set, the next procedure is to calculate above for the uncertainty budget (Table 22) for the
the variation limits (Xi estimates), uncertainty con- metrology instrument, the analysis illustrates that
tributions, and then to set up an uncertainty budget. around 50% of this "budget" is accounted for by tem-
This can be achieved in several steps, as follows: perature factors alone. Because of the large impact
that temperature influences have on the measurement
• calculation of estimates - based, in this case, on uncertainty for any measurements on the shop floor,
ambient conditions, materiallproduct specifica- they must be "corrected back" to 20°e. However, cor-
tions and possible variation limits that are calcu- recting for temperature within the production shop
lated for each input quantity; may lead to unforeseen practical problems being
• application of distribution factor - type of distri- encountered. The temperature on the shop floor could
bution is selected (depending on the probability vary during the working day as dimensional mea-
that the result lies within the prescribed limits) surements occur. The recorded thermal history may
for a rectangular distribution (100%) or normal also not be correct during this time, because any ther-
distribution (99.73%). mometer readings may not be accurate, or even some
NB As an alternative to complex calculations, the uncertainty may exist concerning the actual value for
following correction factors can be used: the coefficient of linear expansion for the workpiece
• rectangular distribution: multiply the limit by material, which leads to dubious inspection pro-
0.7; cedures. These uncontrolled factors of inadequate
• normal distribution (3s): multiply the limit by temperature correction mean that more generous
0.5. limits are necessary to take into account the variabil-
Owing to the fact that in Table 22 there is no corre- ity introduced because of such temperature influ-
lation between the uncertainty components, the com- ences. Here, the most obvious way to mitigate the lack
bined standard uncertainty is expressed in the form of temperature control and its adverse effect on the
of a standard deviation. In this case, the combined measurement uncertainty is to undertake final
uncertainty (UJ, which was established at 2.29 j.Lm, inspection in a temperature-controlled environment
will need to be multiplied by a coverage factor of k = that is as dose as practicable to that of 20°e.
2, to ensure a 95% probabilitythat the measured result Thermal-induced measurement uncertainties
willlie within the uncertainty band. The metrology during metrological inspection, notably the coeffi-
instrument in this case, which had a manufacturer's cient of linear expansion workpiece issues, become
stated specified accuracy of 2 j.Lm, here had a mea- critical factors at the highest levels of accuracy. In
surement uncertainty of 4.58 j.Lm when used to calibration laboratories where thermal influences
inspect shafts under workshop conditions. In any must be adequately controlled in order to minimise
uncertainty budget calculations the objective is to temperature-induced errors into measurement, the
ascertain whether the measured result will be ade- three greatest single contributors in uncertainty
quate for its intended purpose. For example, if the components are:

Table 21. "Uncertainty budget", in the form of a spreadsheet

Source of uncertainty Value ± distribution Probability Divisor uncertainty Standard

Calibration uncertainty 5.0mm Normal 2 2.5mm

Resolution (size of division) 0.5 mm' Rectangular \/3 0.3mm
Artefact not Iying perfectly straight 10.0mma Rectangular \'3 5.8mm
Standard uncertainty of mean 10
repeated readings 0.7mm Normal 0.7mm
Combined standard uncertainty Assumed normal 6.4mm
Expanded uncertainty Assumed normal (k = 2) 12.8mm

'Here the (±) half-width divided by.v3 is used.

Source: Beginner's G.U.M. BeIlINPL. 2000.

 Quality and calibration techniques

Table 22. A typical "uncertainty budget" for a metrology instrument

Quantity (xJ Type Estimate (x) Distribution Uncertainty

factor contribution (fJ.m)

Repeatability A 1.00
Variation of zero A 1.00
Indication error B 1.00 0.6 0.60
Flatness B 0.60 0.5 0.30
Straightness B 0.60 0.5 0.30
Parallelism B 1.00 0.5 0.50
Ambient tempo B 0.30 0.7 0.21
Temperature diff. B 1.40 0.7 0.98
Comp. form error B 2.00 0.6 1.20
Combined uncertainty U, 2.29
Expanded uncertainty U= 2x U, 4.58
Where

Sourees: Shankar (April 1999)/Beginner's G.U.M. BeIlINPL, 2000.

calibration of material thermometers - ensuring demands are for this specification. By way of illustra-
that these instruments (various types of temper- tion, sometimes a specification may incorporate
ature measurement devices) are thermally error- factors such as "attributes" (non-measurable enti-
mapped; ties) - visible appearance, interchangeability, electri-
2 actual workpiece temperature is established - as cal connectivity, etc. - which have no actual bearing
this will have a large influence on the coefficient on the measurements previously taken, but can
of linear expansion (which must also be known); impinge on the whole uncertainty problem.
3 air temperature reading errors must be known -
due to air convection currents (lamellar air flowl
air turbulence/hot -, or cold-spots within the Reducing measurement uncertainty
room, etc).
It is essential to remember that it is just as important
Other environmental factors such as humidity and to attempt to minimise uncertainties as it is to quan-
air pressure can also have a significant influence on tify them. Good working practices will help to
the measurement uncertainty, particularly if the reduce these measurement uncertainties, which
light path of laser-based equipment is employed in might include the following:
the calibration/measurement procedure.
In the previous tabulation and discussion of mea- • calibration of instruments (or have them cali-
sure me nt uncertainty, it was important that the brated bya third party) - then applythose calibra-
correct conclusions were drawn from these results, tion corrections which appear on the certificate;
which is of great significance when deciding if the • compensating for any corrections - make adjust-
values fall within, on or outside the specification ments for any other significantly known errors;
limit. If both the result and its accompanying meas- • ensuring measurements are traceable to national
urement uncertainty fall inside the specification standards - employing calibrations that offer
limits, then the process can be deemed to be com- traceability via an unbroken chain of measure-
pliant, in that the operation will be capable of sus- ments to the relevant standard (see Figure 199);
taining efficient measurement and process control. • selecting the optimum metrology equipment -
For non-compliance to occur, neither the result nor normally they would encompass a calibration
any part of the uncertainty band will fall within the facility with minimum uncertainty of measure-
specified limits. In cases where neither the result nor ment;
part of the uncertainty band is either completely • repeating and checking measurements - occa-
inside or outside the limits, then a dilemma occurs, sionally allow someone other than the usual
as no firm conclusion about the compliance state can inspector to repeat these readings, or utilise a
be made. Prior to stating the compliance with a spec- different method of assessment to ensure validity
ification, it is always advisable to check what the of the results;
• checking calculations and transcription of results artefacts. Due to the fact that most surface texture
- ensuring that results are substantiated and are measuring instruments have a wide range of operat-
correctly written; ing conditions and usage, the net result is that more
• utilising uncertainty budgets - the "budget" will than one type of artefact has been developed to cater
identify where the most significant uncertainties for these demands. The current standard relating to
occur, so that they can be minimised; surface texture calibration (ISO 5436-1: 2000) con-
• awareness of problems in "calibration chains" - at siders five general cases of artefact having several
every step in the chain, uncertainty can increase. versions of each type - the first four initially being
discussed below. A general consideration for any
NB Surface texture measuring instrument un- instrument calibration is that its profile should cor-
certainties covered in the Measurement Good res pond to the surface that is to be measured. In more
Practice Guide no. 37 by the NPL (Chapter 8) - see general terms, these arte facts fulfil the following
References. assessment criteria, namely, Type A calibration stan-
Utilise good working practices in measurements, dards will assess the instrument's vertical magnifica-
such as by following equipment manufacturers' tion factor, but do not give information with regard
instructions operated by using experienced and rele- to instrument calibration in the scanning axis; this
vantly trained personnei; checking and validating latter point of scanning is dealt with by the Type C
software (if employed) to ensure it operates satisfac- artefact. Furthermore, it is important to check the
torily; ensuring that any calculations in the "round- instrument's overall calibration and that the Type D
ing" of measurement values is correct, ensuring that artefact has been produced to assess the ability to
good records ofboth measurements and calculations measure and calculate a surface texture parameter.
are produced at the time of metrological inspection. Finally, the Type Bartefact checks and ensures that
Lastly, keep a written account of any other relevant the selected stylus performs to specification. In order
additional information, as this may be useful if at a to determine whether any surface texture instrument
later stage these results are ever called into doubt. performs adequately, at least four previously cali-
Extra factors need to be considered in any deal- brated and traceable types of artefact need to be
ings with uncertainty issues, particularly if they have available for this task. Hence, these Type A-to-E sec-
any one or a combination of those listed below: ondary calibration standards need to have the rele-
vant current calibration certification to a known
• when using statistics for very small data sets - standard. Such secondary artefact calibration can be
normally considered to be less than about 10; achieved by several means, either using interferome-
• if one uncertainty component is considerably try or by an alternative stylus instrument, which
larger than all the others involved in the "uncer- itself has been calibrated to a traceable standard in
tainty budget"; an "unbroken and documented chain of calibration"
• if just some inputs to the uncertainty calculation to the primary standard (Figure 199), this traceabil-
are correlated; ity being a major requirement of ISO 9000: 2000.
• when the distribution or spread has an unusual These surface texture calibration artefacts have a
shape namely, that it is not "normally distributed"; restricted range, relating to their own individual
• if the uncertainty has been obtained for other characteristics and to those of the instrument to be
than a single result, that is, by fitting a curve or calibrated, as listed in Table 23.
line to a number of points to obtain the required Each type of surface texture artefact will now be
information. briefly reviewed.

This explanation is by no means an exhaustive

account on the subject of measurement uncertainty,
but attempts to discuss some of the relevant princi-
6.5.1 Surface texture artefacts
pIes and practices behind its theoretical application (ISO 5436-1: 2000)
to dimensional measurement.
Depth measurement standard
(Figure 20Ba-c)
6.5 Calibration: surface
Type Al: wide grooves with flat bottoms (Figure
texture 208a-b)

The calibration of surface texture instruments prin- The first artefact illustrated in this subgroup is Type
cipally relies on the use of secondary calibration Al (Figure 208a), featuring a wide calibrated groove
 Quality and calibration techniques

Table 23. Surface texture artefact types, description and be assessed perpendiculady from the upper mean
applications line to the mid-point of the lower mean line. A
significant number, not less than five, of evenly
Type Artefact description Application: distributed traces shall be taken.
A Depth measurement Calibration of the vertical
profile - having a known Type A2: wide grooves with rounded bottoms (Figure
depth with wide grooves
208c)
B Tip condition Ca libration of the stylus tip
measurement condition - having various
Here (Figure 208c) the surface texture depth stan-
depths and widths and
narrow grooves
dard is comparable to Type Al, with the obvious
difference in their visual appearance being the
C Spacing measurement Calibration of vertical profile
rounded groove, rather than flat bottomed. The
- under certain conditions
artefact can also be utilised radius must be sufficiently large enough to be insen-
calibrating horizontal profiles sitive to either the shape or condition of the stylus
tip, or both. Once again, a significant number of
D Roughness Overall calibration of these
measurement instruments surface traces is demanded - at least five - which are
taken across the arte fact and these are distributed
E Profile coordinate Calibration of instrument - for
measurement its profile coordinate system
evenly over the measuring window.

Source: ISO 5436-1: 2000.

Tip condition measurement standards

(Figures 208d, e and 209)
and a flat bottom, incorporating a ridge with a flat
top. Alternatively, another approach is to have a Type Bi: narrow grooves (not illustrated)
number of separated features of equal or increasing
depth/height, similar to that illustrated in Figure The tip condition artefact (Type BI) can have either
213(1). On the arte fact (Figure 208a), each calibrated narrow grooves or a number of separated grooves
feature has enough width to accommodate the stylus designed to be progressively more sensitive to the
tip's shape or its condition, with at least five evenly stylus tip dimensions. The profile of these grooves is
distributed traces would be required over the that they have rounded bottoms, the radii of each
window of measurement. being sensitive to the stylus tip.
In a previous related standard (ISO 5436-1985,
seetion 8) mention was made describing how to Type B2 (Figure 208)
measure the depth of a Type Al calibrated artefact
groove with an appropriate stylus; the extract These calibration standards normally have two or
(slightly edited) has been reproduced below. The more groove patterns on a common base. In this
technique explains how and in what mann er cali- calibration procedure a significant number of traces
bration can be achieved; this operational procedure are demanded, the minimum being 18, which should
for surface trace assessment is as follows (Figure be evenly distributed over the measurement win-
208b): dow. Normally it is recommended that the stylus tip
is sharp ((2 fLm nominal radius) and has a filter with
A continuous straight mean line equal in length a AC cut-off according to the appropriate certifi-
to three times the width of the groove is drawn cation. In Figure 208(d) the Type B2 calibration
over the groove to represent the upper level of the artefact is illustrated, with the sensitive groove
surface and another to represent the lower level, patterns being formed by the isoseeles triangular
both lines extending symmetrically ab out the groove geometry, having both sharp peaks and
centre of the groove (see Figure 208b). To avoid valleys. The recorded Ra from this calibration arte-
the influence of any rounding of the corners, the fact is dependent on the stylus tip's size.
upper surface on each side of the groove is to be Alternatively, the Type B2 groove patterns that are
ignored for a length equal to one-third of the insensitive to the stylus tip are shown in Figure
width of the groove. The surface at the bottom of 208(e), which occur by forming grooves on the arte-
the groove is assessed only over the central third fact that are either sinusoidal (Figure 208e, left) or
of its width. The portions to be used for assess- arcuate (Figure 208e, right). These types of cali-
ment purposes are therefore those at A, Band C brated groove profiles allow the value of Ra to be
in Figure 208(b). The depth d of the groove shall assessed independent of the stylus tip.
 (a) Type Al calibration artefact (b) Assessment procedure of Type Al artefacts

3w

(c) Type A2 calibration artefact

(d) Type B2 calibration artefact

(e) Type B2 calibration artefact

Figure 208. Depth measurement standards (Type Al and tip condition measurement standards (Type B). [Source: ISO: 5436-1: 2000.1
 \ 3 r =stylus path
2

Key:

The output profile essentially represents the 1 Stylus

stylus tip shape if the radius and apex angle 2 Razor blade
of the razor blade are much liner. __- - 2 3 Recorded profile

___ ~ Stylus motion

Schematic diagram of a razor blade trace for

profiling the shape of a stylus tip to determine
its radius.

Figure 209. Calibration artefact Type 83. [Source: ISO 5436-1: 2000.]

Type B3 (Figure 209) instruments. The Type B3 arte fact has a sharp razor
bl ade, but even here a finite radius appears at the
The artefact shown here represents a knife edge artefact's edge, as shown by the magnified view in
having a fine protruding edge, the radius and apex Figure 209.
angle of which is shown, being both smaller and
narrower than that of the stylus under test. The
condition of the stylus can be established by
traversing the stylus over the arte fact and recording Spacing measurement artefacts
the surface profile, as schematically depicted in the (Figure 21 Oa-d)
main diagram. This type of stylus calibration tech-
nique can only be employed with low traversing Artefacts of this type of design are mainly used to
speeds and for direct surface texture profiling calibrate vertical profile components; however, an
 (a) Type Cl calibration artefact

(b) Type C2 calibration artefact

(c) Type C3 calibration artefact

(d) Type C4 calibration artefact

r
Psm

l
-~.
Figure 210. Calibration artefacts for spacing measurement. [Source: ISO 5436-1: 2000.]

alternative application for them is to calibrate these parameters were selected to ensure that the
the horizontal features based on the condition attenuation by either the stylus or filter is negligible.
that the groove spacing will be maintained within As previously mentioned, a significant number of
acceptable limits of tolerance for this purpose. traces are advisable, the minimum number being 12,
With spacing measurement artefacts, the grooves which are evenly distributed across the measure-
must be both consistent and repetitive and would ment window. This assessment strategy allows the
normally come in a variety of profile shapes (Figure parameters to be calculated according to the appro-
210a-d). priate ISO standard.

Type Cl (Figure 210a) Type C2 (Figure 210b)

The Type Cl artefact has sine wave profile grooves This Type C2 arte fact has regularly spaced grooves
that are characterised by Rsm and Ra. The values for based on the profile of an isosceles triangle, enabling
both Rsm and Ra parameters to be assessed. The
(a) Type Dl calibration artefact, undirectional irregular
groove profile geometry values selected here are profile
such that negligible attenuation occurs by either the
stylus or filter. Once again, a significant number of
traces are required, the minimum being 12, these
being evenly dispersed over the measurement
window. The surface texture parameters are calcu-
lated according to the relevant ISO standard.

Type C3 (Figure 21Dc)

With this Type C3 arte fact (Figure 21Dc), simulated

sine wave grooves that feature either tri angular
profiles with rounded (right-hand side diagram), or
(b) Type D2 calibration artefact, circular irregular profile
truncated (left-hand side diagram) peaks and
valleys may be present. At least 12 traces are required
that are evenly distributed over the measuring
window, having the parameters calculated according
to the appropriate ISO standard.

Type C4 (Figure 21Od)

In the case of this standard Type C4 artefact the

grooves have an arcuate profile, being characterised
by Psm and Pa. The values selected here should
guarantee that negligible attenuation occurs by
either the stylus or filter. As with the previously
discussed assessment criteria, at least 12 traces are
necessary that are evenly distributed across the
measuring window. The parameters should be calcu- \
lated according to the relevant ISO standard.
/ \
/ \
Roughness measurement artefact /
/ \
(Figures 2 77a, b) / \
Artefacts such as these are possibly the most
/ ~
commonly utilised for the overall calibration of sur- }. 1. A 1A A A
face texture instruments. To obtain the fuB benefit 'V I,
11\
r-v
7ft
nr
JiT
I'VIR
1'/
.,. 1I IW 1111
IN. 7\
I
from such Type D standard artefacts it is normally
necessary to average a statistically determined I

number of appropriately placed traces on the arte-

fact. L SN: SN:
J
Type D1 - undirectional ir regular profile (Figure
211a) Figure 211. Roughness measurement artefaets_ [Source: ISO
5436-1: 2000_]
This Type D1 artefact has an ir regular profile in the
direction of the stylus travel and is repeated for a
given percentage of AC in the longitudinal direction,
as shown schematically in the diagram. The profile standard is characterised by the parameters Ra and
shape is constant normal to the direction of meas- Rz. As in most of the previous standards, the
urement across the standard. The standard simulates minimum number of traces required is 12, which are
workpiece surfaces that contain a wide range of evenly distributed across the measuring window.
spacing of crests, providing reassurance for an The parameters should be calculated according to
overall check on the instrument's calibration. This the appropriate ISO standard.
 InClustrial metrology . .

Type D2 - circular ir regular profile (Figure 21lb) Profile coordinate measurement artefact
The Type D2 artefact (Figure 211 b) is characterised (Figures 272 and 2 73)
by Ra and Rz having irregular profiles repeated
every SAc in the radial direction. The profile shape Type Ei (Figure 212)
is constant normal to the measuring direction of the
artefact, in the circumferential direction. At least 12 This Type EI artefact measurement standard is char-
traces are required that are evenly distributed over acterised by radius and Pt. The radius of the sphere,
the measuring window and, as before, the parame- or hemisphere as depicted in this instance, should
ters should be calculated according to the appro- be sufficient to allow the stylus tip to remain in
priate ISO standard. contact with the convex surface profile and not foul
on the stylus's stern during its measurement travel.
Prior to calibration assessment the stylus tip should
be positioned symmetrically either side of the
highest point of the intended trace.

Hemisphere

Range

r/
of Stylus Length
gauge

eosin, OIT.,
I -Lcorrected
-r- - - - - ---
------
----------- --
---- - ~

---[:::::>. Stylus traverse

Known hemisphere
(e.g.,UKAS)

Radius change
of stylus - if
damaged

NB Checks for gauge linearity and for any stylus damage.

Figure 212. Type E1 profile coordinate measurement artefact for gauge and stylus calibration using a "qualified" hemisphere. ((ourtesy of
Taylor Hobson/ISO 5436-1: 2000)
 Quality and calibration techniques

Type E2 (Figure 213) window as shown in the measurement location

pattern.
The Type E2 artefact is aprecision prism charac-
terised by angles between the surfaces and Pt on Type F standards (not illustrated)
each surface. The size and shape of the standard
should be such that, as in the case of the previous ISO 5436: 2000 Part 2 describes the use of reference
arte fact (Type EI), the stern does not foul during software and its associated reference data sets -
stylus travel across it and the stylus should be termed "softgauges" - in facilitating calibration tasks
symmetrically set for ramping up, over and down for surface texture measuring instruments. At
the adjacent prism faces. The length of the top plane present these software gauges are not yet commer-
should be long enough to enable the standard to be cially available and, as such, are outside the current
levelled in a stable mann er. discussion. However, they will no doubt be featured
For any calibration procedure, the surface texture in future calibration applications by industry at
instrument should be calibrated at and in its oper- large.
ating environment and across the anticipated
ambient conditions that could influence the in-
service use of the equipment. Generally, adoption of 6.5.2 Stylus damage
the following calibration procedure will minimise
unexpected problems: Because of the potential and possibly fragile nature
of most surface texture styli, calibration is an import-
• the arte fact should be aligned to within 10% of ant diagnostic/condition monitoring aid for the
the measuring range of the instrument; current state of the tip. In particular, the conical
• selection of the appropriate measuring condi- diamond styli are particularly susceptible to unin-
tions, such as tentional damage as the forces on the point are high,
• sampling length, because of the very small contact area on the work-
• evaluation length, piece under test. Normally, wear is not the main con-
• cut-off wavelength; tributor to stylus rejection via calibration, but the
• undertake the specified measurements on each more likely scenario is as a result of tip fracture. The
arte fact distributed across the measurement tip condition ofboth a perfect and unacceptable pair

Prism

Figure 213. The profile eoordinate measurement artefact.

 InClustrial metrology . .

Figure 214. Photomicrograph taken on an electron microscope of a Figure 21 S. SEM photomicrograph illustrating catastrophic damage
conical diamond stylus with a radius tip of 5 fLm. [(ourtesy of to the tip of a stylus. [(ourtesy of Hommelwerke, GmbH.]
Hommelwerke GmbH.]

of styli is depicted in Figures 214 and 215, respec- avoid the potential "worst case" situation that is
tively. In Figure 214 the tip is in excellent condition shown in Figure 216, where contact could be made
and the 5 f.Lm tip radius is intact. Conversely, in along the stylus flank, rather than at the tip. The tip
Figure 215 the tip has suffered catastrophic damage, damage as shown in the lower diagram in Figure 216
through misuse or perhaps because of inappropriate will not fully enter the valley and will foreshorten the
protection -left unprotected without the skid and its peak height as it traverses across the surface, leading
accompanying cover - resulting in a potential condi- to a lower recorded Ra reading under this condition.
tion for accidental damage. The uninformed user This unintentional error in the required surface
may not have noticed this damage, which may not be texture parameter could have major repercussions in
visually apparent until high optical magnification is service, or at best will allow substandard/scrap parts
employed. Therefore with the stylus tip in this poor to enter the supply chain. Clearly, there is a require-
state, the operator is currently unaware that false me nt to prevent these unanticipated and undesirable
readings are being introduced into the measurement measurement/quality problems from occurring;
cyde. This stylus tip condition (Figure 215) is dearly with systematic and regular calibration their likeli-
unacceptable, which is why it is important to hood is significantly diminished.
periodically check - via calibration artefacts - that
the tip is in perfeet condition. This calibration pro-
cedure should, as aminimum, be undertaken at least
once per day. Ideally, calibration needs to be part of 6.6 Calibration: roundness
the measurement strategy and its frequency adjusted
according to:
Roundness-testing instruments are generally of two
• conditions of inspection frequency - many read- basic configurations, as depicted in Figure 112.
ings and components to be assessed, necessitates Instrument calibration consists of some general
more frequent calibration routines to be under- testing procedures and the use of several artefacts to
taken; assess the equipment's capabilitity. The types of
• critical nature of these components - high accu- testing regime to be undertaken might normally
racy, more calibration and vice versa; indude the following:
• other unusual metrological conditions - which
might degrade the stylus performance during the • spindie - checking with a previously calibrated
measurement routine. glass sphere or hemisphere;
The ramifications of a damaged stylus on the • stylus deflection - using an appropriate set of cali-
recorded surface texture result can be gleaned from brated gauge blocks, or utilising the so-called
the schematic diagrams shown in Figure 216. Here, "flick-standard";
both aperfect and damaged stylus is illustrated • stylus-to-component alignment - using the "crest-
traversing across a calibrated surface. In reality, ing standard";
this standard should have a wider induded angle • column straightness - employing a calibrated
between the peaks (see Type B2, Figure 208d) to cylindrical square;
 90° conical
stylus Stylus displacement Recorded Ra

Point of
motion
ofstylus

Recorded Ra
with fractured
stylus

90° calibrated
Standard
--~
,....---- ---

Ra - Standard->!_ _ __

Fractured 90° conical stylus

Figure 216. Effeet that a damaged/worn stylus plays in influeneing the Ra value from the ealibrated standard.

• column squareness - utilising the same calibrated • variability in the user's interpolation of trace
cylindrical square. position on polar graph (for manual "fitting"
techniques only).
NB In the case of the latter two column checks, these
are only undertaken if the instrument has an appro- NB This latter uncertainty is a human-induced
priate alignment column. error and is not strictly speaking an instrument cali-
Some of the known sources of uncertainty in bration error.
roundness measurement are the result of the
following factors:

• horizontal and vertical fluctuations of the instru- Spheres and hemispheres: spin die
ment's spindle path from one trace to the next; assessment
• variability due to stylus type and its length;
• general misalignments in the stylus column to A hemisphere can be employed to check the round-
that of the spindle axis; ness measuring instrument in one of three ways,
• non-uniformity of recorded displayed profile; these are:
 InClustrial metrology , .

(i) Verification - a previously calibrated hemisphere made to coincide with a similar mark on the machine,
is measured on the instrument being tested. If the equating to both their datum (zero) positions. On
differences fall within the limits, then the instrument sophisticated instruments, this known error at specific
is accepatable for further use; angular orientations means that a software correction
NB Prior to such calibration work, the instrument can be applied to minimise such errors. Spheres and
must have previously been calibrated for gain by hemispheres used to assess the performance of round-
using a "flick-standard" - see following pages. ness measuring instruments, are usually round to bet-
ter than 250 nm and have a calibration unccertainty of
(ii) Secondary calibration - a previously calibrated ±10 nm. To achieve sphere calibration, a special preci-
hemisphere is measured on the instrument being sion indexing fixture has to be manufactured (Figure
tested. The result from the instrument is compared 217; see the appendix for a schematic view).A displayed
to the known result. These differences are stored in profile is made with the spherical artefact in this posi-
the instrument and are used to correct subsequent tion averaged over four complete revolutions), then via
measurements (software correction); the indexing fixture it is indexed clockwise by 360/n
NB The instrument must have previously been cali- (where n in this case is 10). Once again, four artefact
brated for gain by the "flick-standard". revolutions are undertaken and recorded, with a total
(iii) Primary calibration - of spindie errors. of 11 (36° interval) indexings being carried out to
return to the original angular datum fixture position
Regardless of the equipment configuration prior (Reeve method). It is normal practice to obtain three
to any instrument calibration, the artefact should be orthogonal planes on the calibration sphere at 90°.
centred; on some instruments they utilise automatie Rather than a glass sphere artefact being supplied with
procedures to align the work axis to the table spindie, the instrument (Figures 217 and 218), the more usual
while others require manual centring and levelling. In item is a glass hemisphere, which has its roundness cal-
order to ensure consistency and repeatability in the cal- ibrated at a 3 mm offset from the equator. In any cali-
ibrated data, it is necessary to orient the artefact to that bration assessment it is important to record the angle
of a known angular position on the instrument. This (8) between the stylus arm and the tangent to the arte-
artefact orientation can be undertaken by its "witness fact at the point of contact. This angle is normally set
mark" which denotes the zero angular position; this is at 0° for a sphere measured at its equator but for
.. .

Figure 217. Calibrating a "test sphere" with a special-purpose Figure 218. Calibrating a roundness-testing machine using a glass
indexing fixture on aprecision roundness-testing instrument. sphere, to 28 nm departures from roundness. (Courtesy of Taylor
((ourtesy ofTaylor Hobson.) Hobson.)
 Quality and calibration techniques

hemispheres it can be both angled and offset (3 mm) Gauge blocksl"fIick-standard": stylus
with respect to the equator (see Appendix C, Figure
242). (Figure 218). deflection
This offset correction for 0 is a useful tip to note
when assessing aprecision workpiece on a round- Stylus deflection calibration, for accuracy and sensi-
ness instrument that has not been recently cali- tivity of the probe and its associated lever arm
brated. If a significant "error spike" appears on the length, can be established by utilising either a set of
polar trace, its origin can be established by simply calibrated gauge blocks and their accompanying
rotating it through 90°; if it does not disappear then optical flat (Figure 219, inset photograph), or by
the error is in the instrument's spin dIe. using a "Flick-standard" (Figure 220). In the former

Gauge
blocks
and
optical
flat

Table
Gauge blocks

Optical flat

I I I I
I II I over gauge
blocks
I I I I

Three gauge blocks:

2.5mm 2.8mm

1 I
Two gauge blocks:
3mm

Figure 219. Calibration of roundness instrument employing optical flat with gauge blocks. [Photograph courtesy ofTaylor Hobson.]
 InClustrial metrology , .

case the optical flat is positioned on the instrument's 0.25 . , - - - - - - - - - - - ,

table, then initially centred and accurately levelled,
enabling the calibrated precision gauge blocks to be
§. 0.20
"wrung" onto the flat's surface, eliminating the "air
gap" which would otherwise occur. The stylus tip is ..
-.
."

brought down to contact these gauge blocks at a ,; 0.15

Co
radius offset and the table is rotated, as shown in e
..;
Figure 219. If three gauge blocks are used, then a O.10 .L.-.--"'.JL.LL......
two-step height can be accurately established, giving 10 100 1000
a recorded series of step heights which can then be Undulations per revolution (upr)
compared to the gauge block calibration certifica- (a) Flick standard amplitude spectrum [Courtesy of PTB, Brannschweig,
tion for establishing the stylus deflection/sensitivity. Germany.]
The correction figures can be entered into the
instrument's software and such correction factors
play an essential role in minimising uncertainty in
the measurement (traceability) chain. For example,
the calibration factor for a two-gauge block step
height could be found in the following mann er:

Measured step = X - Y = Z
Gauge block step = C
Calibration factor = C/Z = F
The general artefact configuration of gauge
blocks with their associated optical flat arrangement
for the calibration of stylus deflection can be
enhanced by employing the reduction lever principle (b) Flick standard and centring iris on instrument
are illustrated in Appendix C, Figure 241. For ultra-
high precision instruments the conventional gauge Figure 220. A "flick standard" and its amplitude speetrum, being
blocks and optical flat arrangement have to be used to assess a roundness instrument.
mechanically magnified by areduction lever
(Appendix C), typically obtaining magnifications of
20:1. This arrangement can produce a total uncer- wave numbers less than 75 upr contribute to the
tainty budget on gauge block steps to <0.016%, this signal, while the individual wave number amplitudes
being equivalent to <3.2 nm on a step height of are quite low if compared to the 12 /-Lm form devia-
2/-Lm. tion. Any calibration work has to be undertaken in
For calibration of stylus deflection on either a suitable measurement range for the full form devi-
longer stylus arms or with earlier roundness testing ation, although the actual calibration is dependent
equipment featuring analogue-based instrumenta- upon the individual wave amplitudes, hence the
tion, the flick-standard can be successfully utilised signal-to-noise ratio (SNR) tends to be quite low for
(Figure 220). This artefact is designed around a such flicks.
precision cylinder with a small flat on the peripheral It has been suggested that a new form of cali-
face (Figure 220b). The standard artefact has a cali- bration standard will improve data capture and
brated flat on its periphery, which when rotated in response, based on a spatial embodiment of the
contact with the stylus tip (Figure 220b), gives a superposition of several sinusoidal waves, termed a
notable "flick" on the displayed profile (Figure 220a), multi-wave standard (MWS). This waveform artefact
which equates to the calibration value - if instru- can be reproduced around the cylindrical circumfer-
ment calibrated is within its range. It has been ence, whereas the generatrices should be straight
reported that such flick-standards have a major lines without form deviations as these act to lower
dis advantage such as a small dynamical conte nt for the influence of potential variance of differing axial
larger wave numbers. For example, the amplitude of sections on the prospective result (Figure 221). Such
the signal of a typical "flick" having a <1>20 mm arte fact geometry has been termed "R-type MWS"-
cylinder with a flatness depth of 12 /-Lm has no the "R" denoting roundness, with the waves being
significant contributions from larger wave numbers reproducible both for internal and external cylindri-
other than at 75 upr (undulations per revolution) cal artefacts. Hence, these internal "R-MWS" arte-
(see Figure 220a). For this calibrated artefact all the facts are the first internal calibration standards.
 (a) Displayed profile of peripheral form profile of multi-wave standard arte fact and its integrating fre-
MWS artefact quency-space analysis, when compared to the con-
Scale:
ventional flick-standard, exhibits a much higher
stability under test, inferring improved transfer sta-
bility of calibration to the end-user.

Cresting standard
Yet another artefact that is utilised for assessment of
angular offsets (arcuate uncertainty) produced by
spherical stylus tips on roundness instruments is
the so-called "cresting standard". "Cresting" is align-
ment of the stylus to the axis of rotation of the
instrument (Figure 222a-c). An incorrect stylusl
component contact causes a certain amount of noise,
which is included with the measurement data.
However, this stylus and component geometrie
relationship will not normally significantly influence
the filtered results.
(b) Corresponding MWS amplitude spectrum The results of the measurement taken can be
influenced in the following three ways:
4
Errors in peak-to-valley measurements (Figure
222d) - this is probably the least significant error
introduced by poor "cresting" and its value is of
the order of: 1 - cos a (where a is the angular
offset) . Typically, with diameters of components
of approximately 10 mm and the instrument
"crested" within 0.5 mm, a is 5.7° and this will
. relate to (1 - cos a) = 0.005, that is, introducing

o - 1 a 0.5% error in peak-to-valley values. If the diam-

eters are larger than 10 mm, the cresting problem
becomes even less significant, although for very
1 10 100 1000 small diameters the use of the bar-type stylus is
Undulations per revolution (upr) recommended. Normal cresting accuracy of a
roundness instrument is typically ab out 0.1 mm,
Figure 221. Form profile and spectrum of a multi-wave standard
being set at installation. If customer-manufac-
(MWS). [Source: Jusko and Lüdicke, 2000/PTß.]
tured styli are produced, then any cresting may
be adjusted to suit.
2 Errors in absolute radii (Figure 222e) - when cali-
Another variant of these waves machined along the brating at a larger diameter than the measured
generatrices of a cylinder and their nearly perfeet diameter, the condition can arise where Erad = Ez
circles as roundness profiles are termed "S-type - EI' For example, if E eres ! = 0.5 mm, R I = 50 mm
MWS" - the "s" in this case denoting straightness. and R z = 25 mm, then Erad = (5 f.1m - 2.5 f.1m) =
Additionally, such a waveform could be machined 2.5 f.1m. This intro duces a 2.5 f.1m error in
into a nominally flat surface, which then can perform absolute R.
its task as a calibration artefact for surface roughness 3 Errors in centring and levelling - as the absolute
instruments. The 80 mm approximate diameter of radius information is used in order to undertake
the R-MWS artefact has its wave profile machined automatie centring and levelling operations (on
into a chemically applied nickel layer on an alu- some sophisticated instruments), then the previ-
minium substrate, by diamond turning coupled to ous error criteria apply. Such factors will not
a fast tool servo-mechanism. A reference mark prevent accurate cent ring and levelling from
denotes the 0° (datum) position or witness mark. occurring, but will have the effect of increasing
The embodied wavelengths were selected to match the number of operations required, particularly
the limit wavenumbers of filters defined in ISO/FDIS: as the component approaches its theoretical
12181,namely, 5,15,50,150 and 500 upr. This evolved centre.
lI:f.!lII InClustrial metrology , .
~

(a) Stylus leading (incorrect) (b) Stylus lagging (incorrect)

I
_ ill:onent/Artefact

~StylUS

I Rotation

(c) Stylus crested (correct) (d) Errors in peak-to-valley measurements

I

-$-
(e) Errors in absolute radii

Cresting standard

Calibrated.
diameter

Figure 222. Stylus fine crown (cresting). ((ourtesy ofTaylor Hobson.)

Column straightness and squareness One major problem when assessing any straight-
ness/squareness alignment is termed "coning
Many of the more sophisticated roundness instru- error". This coning error is principally the result of
ments are equipped with a straightness column (see the initial set-up.
Figures 113, 114 and 122 for typical machine con-
figurations). In order to calibrate the alignment
(straightness and squareness) of this column with
respect to the spindie axis, most notably for rotating 6.7 Probing uncertainty:
table instruments, a cylindrical square is employed.
These "squares" can be purchased in a number of
roundness and form
lengths, typically 500 mm, 1000 mm and 2000 mm.
The smallest of these artefacts has an LSC round- One method to obtain more discrimination in
ness of better than 0.25 fLm and the largest less than roundness assessment and at the same time reduce
0.5 fLm. In Figure 223 is depicted one of the shortest measuring uncertainty is to be able to calculate an
cylindrical squares. estimate of how much a measured result may vary
 Quality and calibration techniques

of the result (see Figure 224b). Both conventional

roundness instruments and scanning instrumenta-
tion comes into their own, as component accuracy
and precision increase. Higher density of data points
means improved process control. Gauge repeatability
and reproducibility (GR&R) tests become signifi-
candy enhanced when sampling over several
hundred data points, by eliminating dependency on
where the actual data was taken, from one trial to the
next. In any GR&R work component alignment is
critical. If an operator removes and then replaces the
same part, its initial set-up datum will be lost, result-
ing in different measurement results. Scanning
instruments can successfully mitigate this set-up
problem if the part has to be disturbed for whatever
reason, as repeatability and consistency in data
capture can be more readily controlled, via the soft-
ware algorithms and the manner in which geometric
data is extracted from the component.

6.8 Nanotechnology
instrumentation: now and
in the future
Figure 223. Aprecision artefact - "cylindrical square" - being
employed to calibrate column straightness/squareness on rotating-
table roundness instrument. ((ourtesy of Taylor Hobson.) The term nanotechnology was first introduced in
1974 by Professor Taniguchi to describe production
tolerances and finishes within the nanometric range.
from the true value. This uncertainty may be the His graphical representation of achievable machin-
consequence of a number of factors, not least of ing accuracies (Figure 198) was extrapolated into the
which are operator inftuence and environmental future and has been shown to be a remarkably accu-
conditions. For example, if collecting a greater data rate prediction of what came to pass (Table 18).
density - perhaps acquiring 800 instead of eight data Taniguchi quite correctly concluded that by the mid-
points on a roundness profile - an industrial 1990s some metrological instrumentation would
company can reduce its measuring uncertainty, as be consistently operating in the region of between
illustrated in Figure 224a. 100 um and 1 um. This has meant that highly sophis-
At one extreme, today's scanning instruments ticated instruments would need to be developed and
can read hundreds or thousands of points in the time that further exploitation of surface and roundness
it takes a co ordinate measuring machine (CMM) technologies would be even more exacting in the
with its associated touch-trigger probe to check future. Nanotechnology encompasses more than
around six data points. By way of illustration, a high- simply highly accurate and precise metrology equip-
speed scanning probe can measure a <1>150 mm bore ment, as many micro-miniaturised components and
in around 10 seconds, taking over 800 measuring assemblies today are fabricated as integral nano-
points in the process. To press the point still further, metric devices. Physics, chemistry, biology, engi-
if a 30-second scan of a <1>50 mm boss is taken this neering, materials science and other disciplines
results in approximately 3000 measuring points on meet at the atomic scale.
the component feature. However, if a CMM is
equipped with a scanning head, it can achieve equiv-
alent levels of operational performance, to that of Instrument performance
high-speed scanning probes. If the measurement
involves the assessment of a near-perfect compo- Instrument characterisation falls into a number of
nent, the number of probing points and their loca- categories, although it is possible to list them by
tions become somewhat irrelevant to the calculation three variables:
 (a) Uncertainty depending on the number of points measured

Typical range for

single point probing

Typical range for

Roundness/Scanning m/e's

10 >200 >2000
Number of points
measured (log seale)

(b) Significant form measurement errors occur with just four recorded points taken, but true
contour assessed when continuous probing occurs

Geometrie uneertainty
of position - produet of Maximum
few points taken Inseribed
Minimum CircIe (MIC)
Cireumseribed
CircIe (MCC)

Single
points

Sequenee of Roundnessl
Seanned points (eontinuous
probing)

Figure 224. Probing uncertainty arises if tao few point are taken on roundness parts. (After Knebel!American Machinist, 1999.)
 (a) Feature and scale of instrument techniques
Macro

ATOMICTYPE
INSTRUMENTS
Glossmeter
Milli
SEM (1970)
STYLUS CONTACT
INSTRUMENTS OPTICAL
INSTRUMENTS

Diffraction

SEM (1990)
SPATIAL
Micro
Nano
HEIGHT

Nanosurf
AFM / STM (1990)

(b) The scale of size offeature/object

Macro

Conventional
Milli engineering

QUANTIFY
% INFORMATION

VISUALISE ~ ~ Metrology
% INFORMATION ,./ ~ ~ ~ / ~ calibration
ACTUAL
TREND Jt~/ DESIRED gap
TREND

Figure 225. Future instrument trends. [After Whitehouse 1991]

• spatial bandwidth - by lateral rangelresolution; ware for a range of operational functions, such as
• vertical bandwidth - by height range/resolution; precise and accurate specimen positioning and
• frequency response - or response time. surface manipulation and image enhancements. In
the early 1990s it was shown at the nano-scale that
Any of the above instrumental characteristics is both friction and surface topography could be suc-
important, because they directly affect spatial and cessfully measured simultaneously. Moreover,
vertical bandwidths. They can also indirectly affect scaling theory becomes a critical factor due to the
the instrument in the case of frequency response, manner in which physical phenomena change with
being related to measurement fidelity. The relative their accompanying scale. By way of illustration, the
weighting of each of these instrument character- influence of system inertia dramatically reduces in
isation categories will change from that of conven- importance in relation to damping factors and stiff-
tional instrumentation to those at the nanometric ness, as equipment is employed when approaching
scale. Significant progress has been made recently in nano-resolution levels. Furthermore, sensor technol-
vertical scale calibration in the atomic region; this ogy dictates that in future metrological applications
was previously difficult to achieve because there are single atomic particles might be utilised to explore
many natural spacings that emerge on the surfaces surface detail at the atomic level, perhaps using
of a boundary. By utilising X-rays in the form of deflection techniques. In Chapter 3, concerning the
an interferometer, it is possible to achieve height uses and applications of atomic force/tunnelling
calibration to better than 10- 2 nm. However, as microscopes and their derivatives, it was mentioned
instrumentation applications are spread wider from that it is now standard operational procedure to reg-
basic technological work to that of cross-disciplines ularly manipulate the atomic structure by moving
where physics, chemistry and biology merge, it is atoms to specific and preferential sites on a speci-
spatial information which becomes paramount. This men's surface. The manipulation and quantification
emphasis toward spatial recognition characterisa- of surfaces within the atomic range is set to continue
tion explains the great emphasis being made on and diversify, with a new range of instrumentation
nanomechanistic movements, the general trends in being developed for the twenty-first century, offering
such development being addressed in Figure 225; the exciting times ahead for the metrologist.
graphical relationship shown in Figure 225(b) indi-
cates how requirements change with discipline as a
function of the scale of size. Nanometric instrument: design and
operation
The instrument illustrated in Figure 226 is one of the
Instrumentation: the future latest research developments from the National
Physical Laboratory in Britain. This "NanoSurf IV"
The previous comments concerning nanotechnol- surface texture measuring instrument offers trace-
ogy, and to some extent their associated enabling able measurement based on sound metrological and
technologies, will in the future be expanded in kinematic principles, by measuring displacements in
their operational capabilities to embrace flexible two orthogonal axes, having a measurement uncer-
and multi-purpose applications. Such instruments tainty of approximately 1 nm. This "NanoSurf IV" is
will be fast in operational and data capture and in a stylus-based instrument that has been primarily
this way will minimise the effects of environmental manufactured from ZerodurTM, comprising three
changes, most notably negating the influences result- basic elements:
ing from thermal drift and vibrational effects. More
will be said shortly concerning minimisation of • base plate - to support the entire structure of the
environmental effects when discussing the newly instrument;
developed instrument depicted in Figure 226. One • slideway - to translate the specimen for measure-
approach to increasing instrumentation versatility is ment;
to enable the equipment to perform many different • bridge - which carries a probing system and a
functions with the same set-up and at this time, but vertical stage to position the probe over the spec-
across a wider range of resolution. For example, it lmen.
should be possible to assess various critical aspects
of the surface topography, including its associated The base plate is lapped and polished flat. The
frictional forces and coefficients, together with bridge section of the instrument forms part of a
surface micro-/nano-hardnesses. Such assessment metrology frame, being kinematically located on the
could be undertaken in a single vacuum chamber, base plate via ball contacts and three associated
which would have the ability to utilise the same soft- vee-grooves. The bridge carries the probing system,
 Quality and calibration techniques

The light source for the interferometer is a 633 nm

frequency-stabilised He-Ne laser, with an output
power of approximately 1 m W. The beam from the
laser is split into two components to be utilised by
the X- and Z-interferometers. A modified version of
the method developed by Heydemann is used to
correct the quadrature signals from the interfero-
meters for DC offsets and differential gains.
Computer simulations by Birch were used to obtain
a correction method to predict a fringe-fractioning
accuracy of 0.1 nm.
The uncertainty analysis of the "NanoSurf IV" has
been calculated according to the ISO G.D.M. and gives
a combined standard uncertainty of measurements in
the X- and Z-axes at 95% confidence of ±1.3 nm. The
traceability to the metre is via the calibrated fre-
quency of its laser source. An instrument such as the
"NanoSurf IV" can be used to measure ultra-preci-
sion parts, such as sinusoidal gratings, depth stan-
dards and random roughness sampIes; conversely, the
instrument can be employed in characterising other
surface texture measuring equipment.
Progress will continue to refine and develop new
instruments for present and future applications of a
Figure 226. The ultra-precision "NanoSurf IV" instrument, having an metrological nature, with yet wider scope and range
uncertainty of measurement of 1 nm. ((ourtesy of NPL.) being presented for surface and roundness exploita-
tion in combination with work of a fundamental and
applied scientific nature.
a Z-axis slideway and a mount for the Z-axis inter-
ferometry.
The "NanoSurf IV" slideways are of the dry- References
bearing prismatic design, similar to those currently
available on the "Nanostep" instrument. The bearing Journals and conference papers
surfaces for the X-slideway are the top surface of the
base plate and a straight edge that is cemented to Birch, K.P. Optical fringe subdivision with nanometric accuracy.
Precision Engineering 12, 1990, 195-198.
this base plate. Two magnets are fixed to the under- Bosse, H., Krüger, R., Lhr, W. and Lüdicke, F. Oberfiächensimulator
side of the carriage, being attracted to a steel rod to für die Dynamische Kalibrierung von Längenmeßtastern.
pull two bearing pads into contact with the straight Technisches Messen 2, 1994, 82.
edge. Three further pads contact the base plate, Donaidson, R. How good are your measurements? Quality Today
ensuring the full complement of five constraints, July 1993, 16-20.
Downs, M.)., Birch, K.P., Cox, M.G. and Nunn, ).W. Verification of
allowing linear motion in one axis only. a polarization -insensitive optical interferometer system with
The probe on the "NanoSurf IV" has its origin in subnanometric capability. Precision Engineering 17, 1995,
that developed by Taylor Hobson for their "Nano- 1-6.
step" instrument (Figures 58-60). So that the laser Downs, M.). and Nunn, ).W. Verification of the subnanometric
capability of an NPL differential plane mirror interferometer
interferometer could be incorporated as the sensing with a capacitance probe. Measurement Seience and Technology
element, the Taylor Hobson "Nanostep" probe was 9,1998,1437-1440.
modified to allow the polished and aluminised end Eigler, D.M. and Schweiger E.K. Positioning single atoms with a
of a Zerodur™ rod to act as the plane mirror of a scanning tunnelling microscope. Nature 344, April 1990, 524.
Evans, C.)., Hocken, R.J. and Estler, W.T. Self-calibration: reversal,
cat's-eye retro-reßector for the Z-interferometer. The
redundancy, error separation and absolute testing.Annals of the
diamond-tipped stylus is kinematically mounted CIRP 45,1996,617-623.
onto the other end of this Zerodur™ rod. Franks,A. Nanometric surface metrology at the National Physical
The two interferometers on the "NanoSurf IV" Laboratory. Nanotechnology 2,1991,11-18.
are based on an NPL-designed retro-reßecting Frennberg M and Sacconi A. International comparison of high-
accuracy roundness measurements. Metrologia 33, 1996,
Twyman-Green type, which does not have polarisa- 539-544.
tion optics and therefore avoids the errors caused Garratt ).D. and Bottomley, S.c. Technology transfer in the devel-
by polarisation leakage. Such interferometers have opment of a nanotopographic instrument. Nanotechnology 1,
been shown to produce sub-nanometre accuracy. 1990,28-43.
Haitjema, H. Uncertainty analysis of roughness standard calibra- Schneider, U. and Hübner, G. Dynamic calibration and testing of
tion using stylus instruments. Precision Engineering 22, 1998, roundness measuring devices by me ans of a waviness standard.
110-119. Proceedings of the Fourth International Symposium on
Haitjema, H. and Kotte, G.J. Dynamic probe calibration up to 10 Dimensional Metrology in Production and Quality Control,
kHz using laser interferometry. Measurement 21(3), 1997, Tampere, Finland, 1992,394-414.
107-111. Schobinger, J.P. Investigation into the production of roughness
Haitjema, H., Bosse, H., Frennberg, M., Sacconi, A. and Thalmann, standards. PhD thesis, Federal Technical College, Zürich, 1959.
R. International comparison of roundness profiles with nano- Shankar, N. Reducing uncertainty in measurement. Quality Today
metric accuracy. Metrologia 33, 1996, 67-73. April 1999, 36-40.
Hall, C. and Griffiths, B. Comparing 2D and 3D stylus surface mea- Sharman, H.B. Calibration of surface texture measuring instru-
surement. Quality Today July 1999,30-34. ments. Proceedings of the Institute of Mechanical Engineers 182,
Hawthorn, J. The practical use of SPC charts. Quality Today 1967,319-326.
September 1994,38-40. Sheer, B.W. and Stover, J.C. Development of a smooth-surface
Heydemann, P.1.M. Determination and correction of quadrature microroughness standard. Proceedings of SPIE No. 3141,1998,
fringe measurement errors in interferometers. Applied Optics 78-87.
20,1981,3382-3384. Shen, Y.1. and Duffie, N.A. An uncertainty analysis method for
Hillman, LW. Calibration of contact stylus instruments within the coordinate referencing in manufacturing engineering.
Deutscher Kalibrierdienst (DKD = German Calibration Transactions of the ASME: Journal of Engineering for Industry
Service). EUROMET: Workshop Traceable Measurement of 117(1), 1995,42-48.
Surface Texture 289,1993,23-30. Smith, G.T. Why the need for calibration? Metalworking
Hundry,B.A briefhistory of quality.Manufacture and Engineering Production September 1998, Q5-Q8.
September 1991,49-52. Stout, K.J. and Johnson, A. Nano-tronics: the role of the engineer
J0rgensen, J.F., Garnaes, J., Thulstrup, D. and Rasmussen, S.E., in nanotechnology. Korean Journal of Precision Engineering
Calibrated surface characterization, problems and solutions. October 1998.
Seventh International Conference on Metrology and Properties Taniguchi, N. The state of the art of nanotechnology for process-
of Engineering Surfaces 1997,358-365. ing of ultraprecision and ultrafine products. Precision
Engineering 16(1), 1994,5-24.
Jusko, O. and Lüdicke, F. Novel multi-wave standards for the cali-
bration of form measuring instruments. Physikalisch- Taniguchi, N. Current status in and future trends of ultra precision
machining and ultrafine materials processing. Annals of the
Technische Bundesanstalt, Braunschweig, Germany.
CIRP 32(2),1983,573.
Kellock, B. Getting the measure of calibration. Machinery and
Taniguchi, N. On the basic concept of nanotechnology. Proceedings
Production Engineering December 1997,47-48.
ofICPE, Tokyo, 1974.
Knebel, R. Scanning for better results. American Machinist
Teague, E.C. Uncertainties in calibrating a stylus type surface
November 1999,76-78.
texture measuring instrument with an interferometrically mea-
Leach, R., Haycocks, J., Jackson, K., Lewis, A., Oldfield, S. and
sured step. Metrologia 14, 1978,39-44.
Yacoot, A. Advances in traceable nanometrology at the
Tsukada, T. and Kanada, T. Evaluation of two- and three-dimen-
National Physical Laboratory. Nanotechnology 12, 2001,
sional surface roughness profiles and their confidence. Wear
RI-R6. 109,1986,69-78.
Leach, R.K. Traceable measurement of surface texture in the optics Tyler Estler, W., Evans, c.J. and Shao, 1.Z. Uncertainty estimation
industry. Large Lenses and Prisms Conference, University for multi position form error metrology. Precision Engineering
College London, March 2001. 21,1997,72-82.
Leach, R.K. Traceable measurement of surface texture at the Vasseur, H., Kurfess, T.R. and Cagan, J. Use of the quality loss func-
National Physical Laboratory using NanoSurf IV. Measurement tion to select statistical tolerances. Transactions of the ASME
Science Technology 11,2000,1162-1172. 119, August 1997,410-416.
Leach, R.K. Telling the rough from the smooth. Materials World Whitehouse, D.J. Nanotechnology instrumentation. Measurement
8(2),2000,18-19. and Control24, March 1991,37-46.
Lee, G.1., Mou, J. and Shen, Y. An analytical assessment of mea- Whitehouse, D.J. Nano-calibration of stylus-based surface mea-
surement uncertainty in precision and machine calibration. surements.Journal ofPhysics E: Scientific Instruments 21, 1988,
International Journal of Machine Tools and Manufacture 37(3), 46-51.
1997,263-276.
Lee, G.1. and Mou, J. A method to enhance the quality and
reliability of inspection data for precision manufacturing
applications. Proceedings of ASPE (lOth Annual Meeting) 1995,
Books, booklets and guides
96-99. Guide to Uncertainty Measuurement. BIPM, IEC, IFCC, ISO,
Litsikas, M. The do's & don'ts of roundness measurement. Quality IUPAC, IUPAP annd OlML, 1995.
Magazine November 1996, 1-6. Bell, S. A Beginner's Guide to Uncertainty of Measurement.
Locke, J.W. Quality standards for laboratories. Quality Progress National Physical Laboratory (NPL), 2000.
July 1993, 91-94. Birch, K., Estimating Uncertainties in Testing. Measurement
Nakajima, N. et al. Study on microengines. Sensors and Actuators Good Practice Guide no 36, MTA, March 2001.
20, 1989,75-82. Bergman, B. and Klefsjö, B. Quality. McGraw-Hill, New York, 1994.
Petzing, J. and Joy, M. Metrology, measurement and calibration, Dietrich, C.F. Uncertainty, Calibration and Probability (2nd edn).
assessment and training. Engineering Technology December Adam Hilger, 1991.
1996,21-23. EAL- R2-S 1. Supplement 1 to EAL-R2 Expression of the Uncertainty
Pope, C. Made to measure. Professional Engineering March 2001, of Measurement in Calibration. European Co-operation for
29. Accreditation of Laboratories, Rotterdam, November 1997.
Reeve, C.P. The Calibration of a Roundness Standard. US EAL- R2. Expression of the Uncertainty of Measurement in Calibra-
Department of Commerce, National Bureau of Standards - tion. European Co-operation for Accreditation of Laboratories,
NB SIR 79-1758, June 1979. Rotterdam,April,1997.
Saundry,1. Calibration and the user. Quality Today March 1999, Farago, F.T. Handbook of Dimensional Measurement. Industrial
27-30. Press, New York, 1968.
ISO 5436-1985. Calibration of Stylus Instruments. Ozeki, K. and Asaka, T. Handbook of QuaIity Tools. Productivity
Jusko, O. and Lüdicke, F. Novel Multiwave Standards for the Press, 1990.
CaIibration of Form Measuring Instruments, Physikalisch- Price, F. Right First Time. Gower, 1984.
Technische Bundesanstalt (PTß), Braunschweig, Germany. Reeve, C.P. The Calibration of a Roundness Standard. National
Leach, R.K. CaIibration, Traceability and Uncertainty Issues in Bureau of Standards (USA), Report No 79-1758, June
Surface Texture Metrology. NPL Report CLM-7, June 1999. 1979.
Leach, R.K. The Measurement of Surface Texture using Stylus Smith, S.T. and Chetwynd, D.G. Foundations of Ultraprecision
Instruments. Measurement Good Practice Guide no 37, NPL, Mechanism Design. Gordon & Breach, London, 1993.
July 2001. Stout, K.J. From Cubit to Nanometre. University of Huddersfield,
Moore, W.R. Foundations of Mechanical Accuracy. Moore Special December 1998.
Tool Co., 1970. Stout, K.J. Quality Control in Automation. Kogan Page, 1985.
Munro-Faure, L. and Munroe-Faure, M. Implementing Total Swanson, R.C. The Quality Improvement Handbook. Kogan Page,
Quality Management. Pitman, 1994. 1995.
Manual of British Standards in Metrology. Hutchinson, 1984. The National Measurement System Programme for Length.
Nakazawa, H. Principles of Precision Engineering. Oxford National Physical Laboratory (NPL), October 1999.
University Press, 1994. Taniguchi, N. Fundamentals and AppIications of Nanotechnology.
Engineering Dimensional Metrology. National Physical Labora- Kogyo Chosakai (in Japanese), 1988.
tory, 1996. UKAS pub. no. M 3003. The Expression of Uncertainty and
Oakland,J.S. Statistical Process Control. William Heinemann, 1986. Confidence in Measurement. December 1997.
Appendices

"Begin at the beginning," the King

said gravely, "and go on till you
come to the end: then stop."
(Alice's Adventures in Wonderland, Lewis Carroll, 1832-1898)

297
AppendixA Ra
1
= -
Ir
f Ir
Z(X) I dx (units ofm)
Previous and some current surface texture parameters 0

In Chapter 1 the current parameters were discussed The Ra parameter would be primarily used in app1i-
relating to ISO 4287: 1997; here their immediate cations to monitor a production process, where a
predecessors and some current ones are mentioned gradual change in the surface finish was anticipated,
at some length, as many companies still require for example, due to too1 wear in the cutting process.
information relating to these previously valid para- There are a number of conditions that must be
meters for surface texture assessment. met for Ra to be utilised satisfactorily:
Prior to a discussion on the previous methods of
obtaining roughness data from a surface texture • the Ra va1ue over one sampling length represents
trace, a brief review of these surface texture para- the average roughness. The effect of a single
meters will allow a greater understanding of how, spurious non-typical peak or valley within the
where and why they are applied to gain an insight profile's trace will be averaged out and, as such,
into methods of classifying surface conditions. In has on1y a small inftuence on the Ra value;
essence, surfaces that are manufactured can be • it is usual to take assessments over several
produced in a myriad of ways, with surface rough- consecutive sampling lengths and then to accept
ness parameters being historically developed to fulfil the average as the Ra value. The evaluation length
specific commercial and industrial needs. This has contains several sampling lengths (Figure 18a);
meant that a considerable number of surface this ensures that the Ra value is typical of the
descriptors, as they are sometimes known, have surface under test;
occurred, resulting in a frequently quoted term, • an Ra value alone is meaningless, unless quoted
"parameter rash", coined some years ago by with the metre cut-off (AC), or if this can be
Professor David Whitehouse. Many of these para- assumed. Repeatability of the Ra value will only
meters are utilised only by particular specialist occur at an identical cut-off length;
industries. Generally, they can be broken down into • where a dominant surface texture pattern occurs
distinct groups: (lay), then Ra measurement is undertaken at 90°
to this direction;
• amplitude parameters - which are measures of • that Ra does not provide information as to the
the vertical characteristics of the surface devia- shape of either the profile or its surface irregu-
tions; larities. Different manufacturing processes will
• spacing parameters - measures of the horizontal produce divergent surface finishes; therefore it is
characteristics of the surface deviations; normal procedure to quote the actua1 production
• hybrid parameters - which are some combination process along with its expected Ra numerical
of amplitude and spacing parameters; value(s);
• statistical parameters - amplitude heights treated • Ra offers no distinction between peaks and
as statistical data sets. valleys on the surface trace;
• assuming that the Ra numerical value has not
been computed, requiring determination from a
graph previously made of the surface, this can be
Amplitude parameters simply undertaken using suitable software, or
previously in a more basic form by a planimeter.
Ro: arithmetical mean roughness Figure 227 graphically depicts this Ra problem
and illustrates the fact that the workpiece's Ra
The classification of the relative roughness of numerical value alone (as has already been men-
surfaces was developed in England and was termed tioned) is not only meaningless but can, under
centre line average (CLA), while in the USA its equiv-
certain conditions of incorrect interpretation, cause
alent term was the arithmetic average (AA). The catastrophic consequences. This identical Ra value
derivation of Ra is graphically illustrated in Figure shows three potential in-service conditions (Figure
18(b) (see Chapter 1). Mathematically, Ra is the 227: (a) bearing surface; (b) intermediate surface;
arithmetic average value of the departure of the and (c) locking surface; all having an Ra of 4.2 f.lm
profile from the centre line throughout the samp1ing in these examples. In case (a) this surface might
1ength. Determination of Ra is normally computed prove ideal for a "light" bearing application, perhaps
by the software but can be derived using the where some local oil retention was necessary to
following formu1a: reduce its frictional characteristics; whereas case (c)
 (a) (b) (c)

Figure 227. Differing profiles producing identical Ra values.

would prove somewhat useful if surfaces were 3,4,52,4,61,4,7

needed to be locked together for either a permanent
or semi-permanent state. Case (b) might be more In each case, the arithmetic average is 4, the succes-
appropriate if the in-service application required sive increase of 1 in the higher value being balanced
neither a locking nor a bearing condition. by a decrease of 1 in the lower value. The rms values
What can be conclusively stated is that if a are "";16.6, "";18.6 and "";22, respectively, illustrating
designer required a critical and highly stressed part that the increase in the highest number in the group
to have bearing condition (a), but the metrologist outweighs the decrease in the lowest, or, for a more
had incorrectly assessed the surface just using mathematical treatment of the rms value:
Ra alone - thus giving a false interpretation of the

fT, f
lr
surface condition and leading to the selection of a
surface with locking state (c) - this would drastically Rq = - Z2(X) dx (m)
Ir 0
shorten the component's life under highly stressed
states. Therefore, using such a parameter and only The reason why both Ra and Rq roughness average
quoting its numerical value in isolation, without parameters were adopted was, in the main, his-
further specifying the other mutually related torical. The Ra surface descriptor is simpler to esti-
descriptors to quantify the surface condition, can mate from the profile recording and as a result
later lead to disastrous consequences. Incorrect use was initially adopted before roughness-measuring
or interpretation by a designer or metrologist of any instruments became more popular. Instrumentally,
surface descriptor could result in, at best, the part Rq has the advantage of simplicity, as phase effects
being totally inappropriate for its current applica- in the electrical filters become of comparatively less
tion - causing poor reliability and, at worst, leading importance and can be disregarded, although they
to catastrophic failure. inftuence the arithmetic average somewhat and
cannot be ignored.
What can be seen in many of the amplitude para-
Rq: root mean square meters is that there is no fixed relationship allowing
conversion from Ra to Rq, as it would depend solely
An alternative method of calculating the average on the shape of the profile. For many production
roughness of a surface is to use statistical tech- processes the Rq value of the surface would be at
niques, such as the root mean square (rms). The rms least 10% greater than its equivalent Ra. In general,
can be obtained by squaring each value, then taking a reasonable approximation of the ratio Ra :Rq =
the square root of the mean. For example, the arith-
1:1.1, but care must be taken with this conversion
metic average of four values, a, b, c and d, would be:
factor as some surface topographies can introduce
an error. As an example of the misleading ratio
(m) effect, the case of a lapped surface has an asymmet-
4 rical profile (i.e., with deep valleys and rounded
crests) - the Ra:Rq ratio can increase to 1:1.5.
When compared to the arithmetic average, rms has
the effect of giving extra weight to higher values, as
shown by the following three groups of values:
Ht, Htm, Hv, Hz, Hmox, Hpand Hpm: peak • Rt is directly affected by the presence of dirt and
scratches, because there is no "averaging effect";
and valley heights • if the slope of the profile is ignored, then there is
no ambiguity as to which are the highest and
Occasionally, it is useful to specify the maximum lowest slopes.
roughness (peak-to-valley) height, rather than Ra,
which gives the me an height. Typical of one of the The symbol Rmax (Figure 228) is the vertical
maximum roughness descriptors is Rt; this para- distance between the highest and lowest points of a
meter correlates to a tactile assessment of the sur- profile within the sampling length, but because of its
face when handled. Rt equates to the German term sensitivity to scratches, ete., it is more usual to take
for roughness depth - Rauhtiefe - being the highest the mean - Rtm - of the five individual (consecutive)
and lowest points of the profile within the evalua- sampling lengths (Figure 228). In Germany, Rtm
tion length, measured perpendicular to the ordinate equates to RZ(D1N)' whereas elsewhere around the
lines on the profile chart (see Figure 228). Some world RZ(ISO) is a different parameter, as described
points worth mentioning relating to the Rt para- later in this section.
meter are that Rp is illustrated in Figure 228, which is another
height term, being the maximum height of the
• with the definition of Rt the centre line does not profile above the centre line within the sampling
occur; length. As the Rp value tends to be extremely sensi-
• the value of Rt is determined over the evaluation tive to large profile peaks within the sampling
length; length, the information is not too reliable and as a

Rmax2
Rmax4 RmaxS
_ RmaxI!

Rp Rt

L L L

L 1--------------1
Evaluation Length J
Rmax= maximum peak-to-valley height within the sampling length L.
Rt = vertical height between the highest and lowest points of the profile within the
evaluation length
Rtm =Mean value of the Rmax of five consecutive sampling lengths.
Rp = Height of the highest point of the profile above the centre line within
sampling length L.
Rpm =Mean value of the Rp of five consecutive sampling lenghs.
Figure 228. Derivation of some peak parameters.
result should be used with care. Due to this uncer- RZrJIS): 1O-point height
tainty introduced by "stray" dominant peaks using
Rp, the "me an levelling depth" parameter Rpm offers Although the Rz 01S ) or RZ(ISO) parameter has now
a more consistent alternative. This latter surface been deleted from the latest ISO standard (ISO 4287:
roughness descriptor is the mean value of the level- 1997), it has been included in this discussion as there
ling depths of five consecutive sampling lengths, are still many engineering drawings in existence that
quantified by the following equation: utilise this surface texture descriptor. The Rz 01S )
parameter averaged the peak-to-valley values, which
had the effect of reducing the influence of either a
spurious irregularity or a scratch. As its name
suggested, it was derived from 10 points within the
or sampling length (i.e., being a German-based stan-
1 5
dard the letter "z" was derived from the German;
Rpm = - 2: Pi Zehn Punkt Höhe which equates to "lO-point
5 i=1 height") and is the mean distance between the five
highest and lowest peaks and valleys, respectively,
This Rpm term gives reliable information on the measured perpendicularly to a base line (see Figure
profile's shape and can indicate whether the part 229). RZ(JIS) was determined using the following
in question might be suitable for wear-resistant equation:
applications. Due to Rpm being used to determine
the functionality of a surface, it is often used in Rz 01S ) = (ZPI + ZP2 + ZP3 + ZP4 + ZPs)
tribological applications, for example, bearings,
for establishing interference/shrink fit component - (Zv 1 + ZV 2 + ZV3+ ZV 4 + ZV s) + 5
behaviour, sliding surfaces, or surface treatment
analysis - prior to coatings. Small values of Rpm
are characterised by surface topography having Rz01S) =.51 [ ~Pi
5
+ ~ Vi
5 1
wide peaks and narrow valleys, while large Rpm
values indicate a sharp-pointed profile (i.e., "spiky The five peaks and valleys can be located
surfaces"), this latter surface obviously having anywhere across the sampling length, unlike R(DIN)'
considerably worse wear resistance than the former allowing it to be utilised for short surface lengths,
topography. whereas the alternative R(DIN) would be inappro-
priate because significant peaks or valleys per cut-
off length are not present. RZ(JIS) suffers from the

I SAMPLING/CUT-OFF LENGTH I
~f---------~
Figure 229. The 10-point Rz parameter derivation.
limitation that if a number of peaks and valleys Hence by this criterion local peaks e and f (Figure
occurred condensed within the cut-off length they 230a) would be accepted; conversely, the smaller
would be unrepresentative of the whole surface peaks a, b, c and d would be ignored. This decision
and will have influenced the average condition to ignore sm aller local peaks is concurrent with the
over the profile. The cent re line (i.e., Figure 229) is practical significance of these minor profile irregu-
only used as a reference direction from which peak larities, although it does ignore the fact that the crest
and valley measurements are made, although the of a profile peak should be retained in the surface
same result can be achieved if measured from any assessment. What has been said for local peaks is
line inside or outside the profile, but parallel to also true for profile valleys and their associated
the centre line. Rz01S ) lent itself to assessment from local valleys.The roughness amplitude is important
the graph, needing only 10 direct linear measure- for surface assessment for many applications. The
ments. Further, there was no ambiguity in defining spacing of these roughness peaks has importance
the maximum peak, or valley (Zp\ or Zv\); however, too, particularly on rolled sheet steels, etc., where the
prior to determining which were the peaks, ZP2 to control of surface texture is essential to obtain
ZPs' and valleys, ZV2 to ZV s' it was necessary to define uniform lubrication as the sheet is either pressed or
what constituted a peak or valley. A slight digression drawn to minimise scoring and its influence on the
here might be in order to establish how and in what final paint or plating finishes. The value of the Ra
manner peaks and valleys can be established. parameter alone is not sufficient to specify the
differing types of texture obtainable from the rolling
process. Peak spacing control gives improved bond-
Peaks and peak counting ing of finish treatments with better uniformity while
reducing the tendency to cracking during drawing
From a profile graph, if an attempt is made to or forming operations. With our eyes being very
measure a parameter such as the example shown in sensitive to differences in appearance, some plastic
Figure 229, or to count the number of peaks and parts are plated to enhance their cosmetic appear-
valleys within the sampling length, the question ance. When identical parts have been produced at
arises: "What constitutes a peak or valley?" Figure two factories, if their average peak spacings (peak
230(a) illustrates the dilemma associated with peaks; counts) are not similar they may not be a visual
in this example two areas of material project above match. Moreover, peak spacing is often a significant
the mean line, area A between points C and D, factor in the frictional performance of surfaces for
plus area B between points E and F - these can be automobile brake drums.
termed profile peaks. In this example within area A
are superimposed a number of minor peaks a-f,
termed "local peaks". Are these to be regarded as Spacing parameters
separate peaks and included within the assessment?
It might be said that such local peaks, being of low To the eye the visual presentation of a surface is a
amplitude, can be discounted because they are non- combination of its profile depth and peak spacing.
repeatable and are not really part of the surface Even when a surface has the same Ra value, the
topography, resulting from factors such as instru- appearance can be markedly different. The number
ment instability, vibration or debris on the surface. of peaks within the sampling length becomes a
Whatever the cause for these minor peaks, once the controlling factor where visual appearance is crit-
surface is in service and a tribological action occurs, icaI. Establishing the type of peak from the surface
then various smaller local peaks will be worn away is important. Peaks can be established in a number
before any significant effect occurs to the surface's ofways:
performance. As a result of the last statement, the
question raised is: "Should local peaks be profile peaks are normally established if their
neglected?" Taking the case of local peak f in Figure profile crosses the centre line twice; thus the
230(a), it is almost of comparable height to the actual peaks are half these numerical values. An
highest peak and, as such, should not be ignored. example of counting peaks is shown in Figure
Guidelines based on this criterion for inclusion of 230(b), where a, b, c and d represent two peaks;
local peaks have been proposed: when a peak's 2 establishment of "local peaks" provided they
height measured from the smallest adjacent local exceed a certain height; in this example there
valley is not less than 1% of the maximum height of were seven, namely A, B, C, D, E, Fand G;
the profile peak to the valley, measured from the 3 the number of peaks similar to profile peaks pro-
mean line (i.e., height h in Figure 230a must not be jecting above a reference level at height Y can be
less than 1% of Rp + Rv). determined by halving the number of tim es the
 (a) Local and profile peaks

d abc e f

Rp

Rv

(b) Peak counting

Upper reference line

F

Figure 230. The influence of local and profile peaks when peak counting. ((ourtesy of Taylor Hobson.)

profile crosses this reference line; in this case In the first case, the counting of these profile
crossings at e, f, g, h, i, j, k and I give a count of four. peaks is termed high spot count (HSC; see Figure
4 The number of peaks which project through a 231a), whereas the peaks counted by the second,
selectable band centred around the mean line can third and fourth cases are known as peak counts
be established; typically, when the profile projects (Pc), but to minimise any ambiguity it should be
below the reference level at m, then above the stated what represents a peak's definition in our case.
upper level at e, crossing this level twice at fand Pc is normally determined over the greatest assess-
h, this represents just one peak. ment length possible; that is:
 (a) High spot count (HSC)
Selectable of slice level Depth of slice level
1 Count 2 Count 3 Count 4 Count 5 Count 6 Count 7 Count

, I

If.=f--------~~~-~--------,p.j
Samplinglcut-off length I

(b) Mean spacing between profile peaks (RSm)

L --~-~--c:=-j
Samplinglcut-off length I
Figure 231. Spacing parameter derivation: high spot count and mean spacing.

No. of counts up rather than down. As a predetermined evaluation

Pe = - - - - - - - - - length is utilised the count is determined over this
Assessment length (m)
distance, but in order to ensure uniformity it can be
converted to count per unit length, this being inde-
= Peaks/rn pendent of the evaluation length.
Another factor that should be noted is that at the
beginning and end of the evaluation length, if peaks RSm: mean peak spacing
occur in these regions do we count them? This needs
to be defined as the count could otherwise vary by The RSm parameter is concerned with the mean
±2. When there are an odd number of crossings of peak spacing of the profile and is the average value
the reference line within the evaluation length, then of the length of the cent re line section containing a
do we halve this count or round it up? This paradox profile peak and an adjacent valley, or in other words
can be resolved, since there is every chance that at between adjacent crossings - in the same direction
the beginning of the evaluation length a peak prob- - of the centre line (see Figure 231b; i.e., SI' S2' S3
ably existed; therefore it might be prudent to count ... Sn)' It can be calculated as follows:
 Hybrid parameters
A further refinement from the amplitude and
RSm is particularly useful as it can be used to repre- spacing parameters are those that can be classified
sent (normally) the feedrate of a machining pass as neither, hence the term "hybrid" parameters,
for an anisotropic surface - that is, one having derived from differentials in the surface texture
predominant machined cusps; conversely, RSm can profile. This hybrid group of parameters tended to
be used to determine the feed associated with be somewhat unreliable initially, because of the rela-
surfaces having directionallay. tive inability to measure the differentials correctly
The more sophisticated instruments can utilise with analogue-based instrumentation, as they are
additional geometrical parameter terms, such as naturally "noisy". Digital techniques have enabled
WSm and PSm; these are the corresponding para- them to be utilised more readily, but care must be
meters established from the waviness and primary taken in their selection and usage. Hybrid parame-
profiles (form errors), respectively. These para- ters have been particularly favoured by users in the
meters are defined in the relevant standard - Clause tribological and optical industries, and abrief
4 (ISO 4287: 1997) - and can be calculated from the mention of how and why they are favoured might
profile. well be illuminating to the reader.
In fact, these P-, R- and W-parameters are derived
as follows:
Pmr, Rmrand Wmr. material
• P-parameter is calculated from the primary ratio/Abbott-Firestone curve (ISO 4287:
profile;
• R-parameter from the roughness profile; 1997)
• W-parameter from the waviness profile.
The interaction of engineering surfaces in contact
NB The first capitalletter in any of these parameter with each other determines their ability to perform
symbols designates the particular type of profile satisfactorily in service. If enhanced surface contact
being evaluated. For example, Pt is calculated from is desirable, such as in the case of bearing applica-
the primary profile and likewise Ra from the rough- tions, adhesive situations, sealing properties, etc.,
ness profile. then the surface topography must be established to
suit such conditions. Considering the case of the
bearing, the interface between adjacent surfaces
would normally be separated via some form oflubri-
cating film. Hence any surface asperity - peak - on

Material ratio
curve
100 (bi + bÜ.fu + b 41
L

Sampling length
o 100
---co-
Mr%

Figure 232. Representation ofthe material ratio/Abbott-Firestone curve. ((ourtesy ofTaylor Hobson.)
either of the contact surfaces that is higher than the fraction and p, the profile depth. This Tp parameter,
film of lubrication will penetrate through this superseded by Pmr (although it simulates wear
medium, resulting in physical contact with its asso- effects) is only an approximate substitute for true
ciated mating surface. Such contact will introduce running-in tests, this is because:
friction, thermal increases and premature bearing
wear. Thus any surface having too many inappro- • the bearing ratio is only a fraction of the actual
priate peaks will not be as effective as a bearing, length and not representative of a surface area;
when compared to one having low peaks and deep • estimation of the bearing area is obviously estab-
valleys. lished from a predetermined and short cut -off
The material ratio curve, namely Pmr, Rmr and length - ignoring gaps that could result from
Wmr, for primary, roughness and waviness profiles, waviness and form errors;
respectively, or the Abbott-Firestone curve (as it was • Pmr, being a theoretical value, by its very nature
previously known and denoted as Tp ' but often relates to an unloaded surface, whereas in reality
written as tp )' can be expressed as a percentage of the actual surface may undergo elastic deforma-
the evaluation length at a predefined depth below tion;
the highest peak. It can be determined by the • practical in-service conditions mean that two
following expression (see Figure 232): contacting surfaces are involved and each specific
surface's features contribute to overall wear;
• induced wear is often accompanied by bulk or
Pmr (%) plastic flow of bearing material, with the result
that the geometrical concept of crests being
neatly truncated by an arbitrary line drawn
through them is impractical and unrealistic.

Despite these reservations, Pmr, Rmr and Wmr

Bearing applications enable the kinematic move- are very useful surface descriptors, finding num-
ment of mating component surfaces, either as they erous applications which have been shown to be
slide, or produce rotational motions, or a combina- successfully correlated with their in-service per-
tion of both relative to each other. Through tribo- formance. The converse of these parameters can also
logical studies (from the Greek tribos meaning "to be applied, namely the depth at which a certain
rub") and the associated wear, this being a natural bearing ratio is obtained (see Figure 233). For
consequence of bearing surface contact. Is signi- example, if experience has shown that a bearing
ficantly reduced when a lubricating film is present, ratio of 65% is necessary for an application, the
thus the bearing ratio, which simulates wear effects, depth p at which the crests must be truncated to
has progressively been shown to be of increasing reach this value can be easily established.
use by industry. In the early days bearing ratio was The material/bearing ratio can be determined
manually established from a profile graph; today from the graph by establishing a line at the selected
many surface texture measuring instruments in- depth, parallel to the centre line, then computing the
corporate this parameter. Practical workshop exam- lengths of the intercepted profile. Examples of pre-
pIes of artificially induced wear are relevant here, selected depths on the profile's trace are indicated in
such as when scraping slideways, or the bedding-in the material ratio curve given in Figure 232.
of large bore of white metal bearings. In the latter For differing slices or depths through the profile
example, the use of"engineer's blue" in combination a specific ratio of air-to-material occurs (see Figure
with a "master diameter reference surface" - this 233a), its calculation being dependent upon the pre-
being coated in the substance and manipulated in a selected depth being expressed as a unique and
controlled fashion by the fitter over the surface to be continuous curve for that profile. Figure 233 depicts
scraped - allows telltale "witness marks" to be left three such profiles and their associated curves. This
by the blued "master surface" and these high spots Abbott-Firestone curve (named after the two
are then scraped off by the fitter. Then the "master" Americans credited with its development), alterna-
is reapplied and further scraping occurs until the tively known as the material ratio curve - the latter
desired amount of "high spots" are uniformly term being the preferred name - gives meaningful
distributed over the newly created surface, giving the information on the profile's shape.
estimated fractional bearing area - being the crite- The top of the highest peak within the profile,
rion for the quality of the bearing surface. have been evaluated, establishes the reference or
The material/bearing ratio Tl? designation (now zero percentage line for the material ratio curve.
Pmr, Rmr and Wmr) was originally derived from the Calculation of this curve is influenced by the highest
German word Tranganteil, which equates to bearing peak's height in relation to the others, although in
 (a) Derivation of the Abbot-Firestone curve: Mr =45%
100%
Abbot-Firestone curve
or bearing ratio

(b) Good bearing surface: Mr = 75%

75%

(c) Good locking surface: Mr = 5%

Figure 233. Hybrid surface texture parameters: spacing and depth.

reality the effect of a single peak on a surface's in- pIe, a 5% reference line (Figure 233c) testifies that
service function has litde significance. In order to the top 5% of the profile is not included as part of
minimise the effect of a single peak on the material the calculation for the material ratio. Selection of the
ratio curve, an artificially induced reference line is zero line beneath the highest measurable peak will
chosen to shift the line below this highest peak as be dependent on the topography of the associated
depicted in Figure 233(a and b), this being expressed peaks in the profile, but industrial practice suggests
as a material ratio percentage. Specifying, for exam- the reference level is set between 2% and 5%.
 The inftuence of filtering on the preselected From the idealised profile traces and their accom-
depths through the profile will be an important panying material ratio curves illustrated in Figure
factor in modifying the material ratio curves, as 233, it is possible to visualise whether a surface is
will the measuring length selected. Terms such as either a "bearing type" or "locking type" of surface.
micra-material and macra-material ratias relate to Furthermore, pre-selecting a zero reference line at
the filtered or unfiltered (total) roughness profiles, some predefined depth below the highest peak will
respectively. Normally the filtered parameter is show the anticipated wear pattern or likely asperity
preferred, because it eliminates the effects of wavi- (peak) collapse under predefined in-service condi-
ness in the measured profile. Filter distortion of the tions. This level of peak attrition by tribological
measured profile can introduce erroneous results; action or through plastic deformation may be an
this is particularly a problem when plateau-type important factor relating to its functional operation,
surfaces occur. Typical of these are sintered, honed, anticipated usefullife and subsequent reliability.
plasma-coated, ceramic and chrome finishes. Some
apparent confusion has arisen when measuring
these plateau-type surfaces, as to whether a mechan- Parameters derived (rom
ical form of filtering is preferred, via the use of a skid
to minimise the effect that waviness can play in Abbott-Firestone/material ratio curve
inftuencing the shape of the material ratio curve. (ISO 13565: 1996 - Parts 1 to 3)
Some of the literat ure categorically states that skids
should be utilised for these plateau-type surfaces, The standard denotes several surface parameters
while others are equally adamant that a skidless relating to the functional behaviour of highly
mode of operation is preferable, particularly when stressed surfaces, namely lubricating, sealing and
quantifying secondary machined powder metal- rolling faces. Being derived from the material ratio
lurgy surfaces. Powder metallurgy and many scored curve (Figure 234), these parameters not only char-
surfaces are affected by filtering, particularly via acterise the profile's roughness but also the shape of
"overmodulation", necessitating the use of a "phase- the profile. These derived parameters include:
corrected" filter (ISO 11562: 1996). Regardless of
which skid technique is employed, topographical • Rk - core roughness depth;
filtering of plateau-type surfaces to remove waviness • Rpk - reduced peak height;
effects is important; further, as has been mentioned • Rvk - reduced valley depth;
earlier, it is also strongly inftuenced by the selected • Mr1 - peak material ratio;
measuring length for the surface under test. These • Mr2 - valley material ratio;
conditions of surface determination should appear • Al - material filled profile peak area;
with the surface texture symbol (at the appropriate • A2 - lubricant filled profile valley area;
position - see Figure 3). • Va - oil retention volume.
Previously, no standardised measurement condi-
tions for the assessment of Tp existed; this parameter The material ratio curve is broken down into a
has of late become of less importance with the series of straight lines comprising three parts to
advent of newer parameters such as; Rk, Mr1 and calculate the "Rk family" of parameters - the k repre-
Mr2 - this will be discussed in more detail shortly. sents the word "kerne!", meaning core:

o Mr120 40 60 80
Mr
2100 (%)

Figure 234. Derivation of the kernal roughness depth, Le., depth of roughness core profile, Rk. ((ourtesy of Taylor Hobson.)
• profile peak area - Rpk; Plateau-like surfaces, when measured, can be repre-
• co re roughness area of the profile - Rk; sented by peak height, whereas the valleys' depths
• profile depth area - Rvk. are only a means of establishing the character of its
respective roughness profile. The functional behav-
Such parameters are exdusively utilised for ob- iour of a component is not only decided by its rela-
taining the roughness profile of plateau-like (asym- tive "global" roughness, as indicated by an equivalent
metrical) surfaces with the digital valley suppression Ra value, but also by the profile's surface structure.
filter (ISO 11562: 1996). The derivation of the Rk
parameter is depicted in Figure 234. It is necessary to
find the tangency points for the minimum slope (the The average (RJ.a) and RMS profile
turning point) from the "S-shaped" material ratio
curve, delineated here by the verticallines at Mr1 and (RJ.q) slopes (ISO 4287: 1997)
Mr2. These tangency points relate to the 0% and
100% material ratio lines, being equatable to heights Figure 235 illustrates how the parameters RLla and
Rpk (for peaks) and Rvk (for valleys), respectively. RLlq are derived. Expressly, the parameter RLla is
The Rk zone represents the depth of core roughness, derived from the average slope of a filtered profile
as it is in the vicinity where the profile has the densest relative to the mean line of the trace. Its derivation
bearing region. The locations of both Mr1 and Mr2 can be established by dividing the profile peak in
relate to the material ratios at the top and bottom of question into smaller portions, then numerically
the roughness core, respectively. These values can be averaging the slope for each portion, as indicated in
used to determine still other parameters, such as the enlarged detail of the peak shown in Figure 235.
the "oil retention volume", derived from the Rvk Calculation of RLla can be found from the following
and Mr2 relationship (Va, which relates to the shaded formula:
triangular area A2 in Figure 234), principally used by
the automotive industry. The material ratio curve Va
can be described in terms of the area below the co re RLla 1
= -. 2: I-'
n
n i~l
Lly-I
Llx i
roughness (Mr2) and the demarcation of the 100%
material line. The numerical value of Va can be
where n represents the number of ordinates.
expressed by
Rather than taking individual numerical average
Rvk(lOO - Mr2) values for profile slopes within the sampling length
Va = ---'-------'- (as in the case of RLla), the root mean square "RMS
200 profile slope" (RLlq) can be calculated. A bonus of
utilising the RMS - RLlq - value is that it offers
Rpk - reduced peak height - is illustrated in
increased sensitivity to extreme values, unlike that
Figure 234, being the height of the top portion of the
surface profile positioned above the core profile. Jf a of the numerical method of determination using
small value of Rpk occurs, this characteristic would RLla, which tends to minimise their influence.
represent good wear resistance or its "running-in"
behaviour. The magnitude of Rpk is indicated by
triangular area Al' whose height is equal to Rpk and RLlq = JUr) fr [e(x) - eF dx
with a base of Mrl. By utilising such areas the effect
of a spurious peak in the profile trace becomes where
minimised. At the lower portion of the material ratio
curve the height Rvk represents the depth of the
lowest part of the profile, with valleys extending
below the core profile. This parameter has an oil-
e= Ur) fr e(x) dx

retaining capacity for lubricated faces, such as cross-

honing of cylinder liners, hence the need under this and eis the slope of the profile at any given point.
condition for a high numerical Rvk value. An advantage gained from using these hybrid
Porous components produced via powder metal- parameters, namely RLla and RLlq, is that they can
lurgy processing cannot be accurately measured, due detect modifications in the profile's geometry, both
to the fact that the pores te nd to be interconnected prior to and after processing. Typically, if peaks on
internally (in most cases), so their exact depths are the surface topography become somewhat rounded
not readily apparent, due to the stylus tip's size and through tribological action - wear - this will have
geometry not being able to measure them with suffi- the effect of diminishing the RLla and RLlq values,
cient accuracy. Therefore, when considering valleys, which could be used as a means of quality control
they are normally discussed separately from peaks. for a particular set of processing conditions.
 Im

Figure 235. Derivation of the parameters average (~a) and RMS (~q) profile slopes. (Courtesy of Taylor Hobson.)

The average (AO) and RMS profile (Aq) Aa 27TRa

= .. _ -
wavelengths RL1a

Figure 236(a) indicates how Aa is derived, wh ich is Whenever a complex profile occurs - this being
the average wavelength of the surface profile. The an amalgamation of several different wavelengths -
parameter Aa should not be confused with the then the average wavelength of these "weighted
"spacing parameter" Sm - this being the me an peak- components", according to their amplitude, can be
to-peak spacing within the profile's length when represented by Aa. In Figure 236(a) the roughness
peaks exceed a certain threshold, whereas Aa con- profile is illustrated for three surfaces. If a large
siders all the "components" of the surface profile, average roughness (Ra) is present, but having a sm all
namely, the wavelength in conjunction with the slope (Rßa) - as in the case of the top trace - this
amplitude (its peak-to-valley height) -although if a indicates a wavy type of surface having a large Aa
profile has a significant sinusoidal profile of pre- value. Conversely, if the magnitude of roughness
dictable periodicity, under these conditions Aa will decreases but the slope increases, the surface be-
approach that of the peak-to-valley spacing repre- comes rougher but less wavy, as indicated by the
sented by Sm. middle trace, or as indicated in the bottom trace
The average wavelength (Aa) surface descriptor where the roughness has been superimposed onto
can be determined as follows: waviness. The visual appearance of a surface may
 (a) Average wavelength (Aa)
Aa= 8

Aa= I

Aa=4

e e
e

e
(b) RMS profile wavelength (Aq)

Figure 236. Average (Aa) and RMS (Aq) profile wavelengths. ((ourtesy ofTaylor Hobson.)

vary from that of a highly polished and reflective one of surfaces - an important factor for automobile
to matt surfaces, but they can be compared using the bodies. '\q can be utilised in tribological situations
'\a parameter. where very closely spaced roughness irregularities
The RMS profile wavelength ('\q) can be visu- of the surface te nd to be of relatively small ampli-
alised in Figure 236(b) and is a measure of the spac- tude and rapidly wear away in service - such as for
ings between local peaks and valleys, taking into a bearing application. Since these irregularity ampli-
account their relative amplitudes and individual tudes are quite small, any changes in the overall Ra
spatial frequencies. Its derivation can be established value during running-in are also small, with the
from shorter wavelengths disappearing and resulting in
more pronounced changes in the average wave-
27TRq length. Of more practical importance is the use of
'\q=~~
RLlq the average wavelength as a means of directly moni-
toring the manufacturing process. For example, in a
As in the case of the slope parameter RLlq, this continuous outside-diameter turning operation of
equivalent wavelength parameter ('\q) is more sensi- high-quality finishing, the average wavelength
tive to extreme values than either the RLla or '\a, these directly relates to the feedmarks promoted by the
being for average slope and wavelengths, respec- residual effects of the partial tool geometry as it
tively. passes along the workpiece. If the machine tool
The '\a parameter is particularly useful in sheet settings are incorrect, the values of both '\a and '\q
steel applications where the average wavelength is a will dramatically change, even though the Ra or
measure of the openness or closeness of the texture similar height parameter may not alter significantly.
and correlates weIl with the cosmetic assessment Similarly, for most grinding operations the para-
 . AppendixA

Table 24. Comparison of aetual lca and Ra values from measure- cutting process. The resultant surface profile for the
ments made on aseries of speeimens bored hole would depict the interactions from the
boring bar deformations and any harmonie oscilla-
Production proeess lca(/J.m) Ra (/J.m) tions. This boring bar motion reflected in the profile
trace would exhibit low average profile height, but
Planning 400 17.8 with a large range of height values.
Fine cut planing 440 5.1 Surface texture data can be statistically manipu-
Multiple-tooth milling 1300 5.1 lated, beginning with the profile trace's height or
Fly-eut milling 300 2.2
amplitude distribution curve - this being a graph-
00 turning 100 1.7
Oiamond turning 37 0.3
ical representation of the distribution of height ordi-
Eleetro-erosion 340 10.1 nates over the total depth of the profile.
Eleetro-sinking 57 1.8 The characteristics of amplitude distribution
Surfaee grinding 77 0.95 curves can be defined mathematically by several
Cireumferential grinding 33 0.55 terms called "moments"; these are:
Plunge grinding 43 0.17
Honing 19 0.07 • arithmetic average 01 the profile - the first
moment (mI);
• profile height variance - the second moment
meter Aa or Aq has a direct relationship to the (m2);
average grit size, hence by monitoring the average • skewness - the third moment (m3);
wavelength this will provide relevant information • kurtosis - the fourth moment (m4).
relating to when the wheel must be redressed to
maintain its surface quality.
In Table 24 are shown the Aa values taken using Arithmetic average of the profile
sampies from a range of production processes. This
table illustrates that surfaces generated by a variety The first moment (mI) can be found from the
of production methods having similar Ra values expression
(fine cut planing and multiple tooth milling, or alter-

(~)~Yi
natively turning and electro-sinking) can exhibit
considerably differing values of Aq. This enables the ml = y=
n I~l
metrologist or manufacturing engineer to assess a
surface under test with considerably more dis er im-
ination than might otherwise be the case if only the
Yi
where is the height ordinate and n is the number
of ordinates.
Ra value alone had been used. NB By definition, the mean height of a filtered
profile equates to zero.
Statistica/ parameters
The amplitude distribution curve (Figure 237) Profile height variance
represents the comparative total heights over which
the trace achieves any selected range of heights The formula for the second moment (m2) is
above or below the centre line. One can deal with
this amplitude or height data statistically, in the
same mann er as one might physically measure
m2 = R q2 = (~)~ (yy
n Fl
anthropometrie data such as a person's average
stature, within a specific population range. As with NB The variance gives an indication of the profile
the statistical dispersion of population stature, engi- range heights, its square root being the standard
neering surfaces can exhibit a broad range of profile deviation of the amplitude distribution curve or its
heights. roughness parameter Rq.
For example, a boring operation with a relatively
long length-to-diameter ratio may cause deflection
(elastic deformation of the boring bar) and occa- 5kewness (Rsk) -1504287: 7997
sion the cutting insert to deflect, producing large
peak-to-valley undulations along the bore (wavi- Skewness is a measure of the asymmetry of the
ness). Superimposed onto these longer wavelengths amplitude distribution curve; alternatively, this can
are small-amplitude cyclical peaks (periodie oscilla- be expressed another way, as the symmetry about
tions) indicating vibrations resulting from the the mean line (see Figure 237a). If symmetry occurs
then Rsk equates to zero, but an unsymmetrical iron surface topography, having comparatively large
profile gives rise to a skewed curve (see Figure 237b numerical values of negative skewness. The surface
and c). The bias of Rsk depends upon whether the is sensitive to extreme ordinate values within the
bulk of the material is above or below the mean line. profile under test; this is due to Rsk being a function
In the case of Figure 237(b), the bulk of material of the cube of the ordinate height. As a result of this
occurs below the mean line - promoting a "locking" peak sensitivity, it is a hindrance when attempting
type of surface. Conversely, in the case of Figure to inspect plateau-type surfaces.
237(c), the majority of material is present above the Skewness is the third moment (m3) of the ampli-
mean line - representing a "bearing" or plateau tude distribution curve and can be expressed in the
surface. Such skewness bias enables a metrologist to following manner:
distinguish between two profiles having identical Ra
values but divergent shapes and hence considerably
different in-service applications. A numerical value m3 = Rsk = (~(
1 3)
Rq)
(!) !
n i=l
(yy
can be given to Rsk and in the case of Figure 237(b)
the "positive skewness" may eventually obtain an The shape or "spikeness" of the amplitude distri-
adequate bearing surface, although it is unlikely to bution curve can also relay useful information about
have oil-retaining abilities. This type of surface can the dispersion or "randomness" of the surface profile
typically be exploited for adhesive bonding applica- which can be quantified by means of a parameter
tions. The surface characterised by Figure 237(c) known as kurtosis (Rku).
might occur in the cases of porous, sintered or cast

(a) Skewness bias

Negative skewness (Rsk) Positive skewness (Rsk)

(b) "Locking" surface texture

Amplitude
distribution
curve
(c) "Bearing" surface temperature

Figure 237. How surface texture topography influences the amplitude distribution curve.
 · AppendixA

(a) Kurtosis is influenced by the distribution shape

____ Kurtosis (Rkll»3

Kurtosis (Rku)<3

(b) Material distributed evenly across the whole

of the surface topography

Amplitude
distributioll

(c) Material distributed aboul the centre line

Figure 238. Variation in surface topography influences the shape and height of the amplitude distribution curve.

Kurtosis (Rku) -ISO 4287: 1997 By definition, Rku is the fourth moment of the
amplitude distribution curve (m4), being the relative
"sharpness" of the curve. The amplitude distribution
The surface parameter skewness (Rsk) is not sensi- curve shown in Figure 238(b) indicates smaller and
tive if the profile's "spikes" are distributed evenly less pronounced condensed peaks and valleys than
both above and below the mean line. Another term, that of the larger, more widely spaced "spiky" peaks
kurtosis (Rku), has this ability (see Figure 238a). and valleys illustrated in Figure 238( c). As a result of
Kurtosis provides a means of measuring the sharp- these topographical profile trace differences in peak
ness of the profile, with a "spiky" surface exhibiting and valley height and their density distribution, the
a high numerical value of Rku (Figure 238c); alter- Rku shape changes, as will their respective numer-
natively, a "bumpy" surface topography will have a ical values, as indicated in Figure 238(a).
low Rku value (Figure 238b). As a consequence of Due to the fact that Rku is a function of the fourth
this ability to distinguish variations in the actual power of its profile ordinates, it has extreme sensi-
surface topography, Rku is a useful parameter in the tivity to variations in peaks and valleys within the
prediction of in-service component performance profile trace under test. Due to Rku sensitivity, in
with respect to lubricant retention and subsequent practice it can be difficult to apply to the analysis of
wear behaviour. surface texture. As has already been mentioned,
kurtosis - being the fourth moment of the ampli- This completes the review of just some of the
tude distribution curve (m4) - can be calculated in previous and current surface texture parameters, but
the following manner: it is by no means an exhaustive list of those previ-
ously utilised, as many industries have developed
m4 = Rku = (~)4)
(Rq
(~)
n
~ (yy
,=!
their own parameters for specific industrial or
research-based requirements.

NB Higher levels of moments for the amplitude

distribution curve can be calculated, but they are of
litde practical relevance for analysis of surface
texture.
 length (AW) analysis, which determines charts of
Appendix B instrument performance in AW space (see Figures
Amplitude-wavelength (AW) analysis: "Stedman diagrams" 239 and 240).

AWspace: the concept

Introduction
In essence, the concept provides a topographical
A myriad of different types and models of pro- characterisation of surface texture that can be
filometers and associated instrumentation exists for modelIed in two-dimensional space, by combina-
use by the surface metrologist, ranging from the tions of surface wavelengths and amplitudes of
basic (portable) stylus contact instruments, to sinusoidal profiles enabling mapping in AW space.
sophisticated non-contact equipment. Each type of Individual points in this AW space correspond to a
instrument has its own distinct characteristics and unique amalgamation of amplitude (A) and wave-
capabilities; these instruments can be compared by length (W) for a specific topographical character.
predicting their performance from information For example, the sine wave y = A sin(21Tx/W) has
supplied by the manufacturer in terms of their stylus a maximum slope (Smax) given by 21TA/W and a
radius, ranges and resolutions, slide quality, etc. minimum radius of curvature (R min ) given by
Techniques have been developed that are based on W 2/41T 2A. The particular values of Smax and Rmin can
how such instruments or probes will respond to be plotted as straight lines of slopes 1 and 2 respec-
measuring sinusoidal waves of varying amplitudes. tively in a logarithmic AW space (see Figure 239a).
The method employed is termed amplitude-wave- This technique allows for four topographical

(a) Lines of constant slope (arctan Smax) and Iines of constant radius (Rmin) in
amplitude-wavelength space

yL
~ lA 10
E
x E

w 100
"
"0 E 10
Z "-
Ci
E
«
100
10
E
c:

0.1
0.01
0.1 10 100 I1 10 100 I1 10 100 1000
nm mm
Su rface wavelength

(b) Measurement limits of a stylus profilometer in log AW space

Log Rs
Amplitude.........--I----f._--,.4---,

r /2 -+-f-l---,.L,+-I--+-

I
2rh Rh
Log Surface Wavelength

Figure 239. Basis for the Stedman diagrams. [Sourees: Stedman, 1987, 1990; Franks, 1991.]
 (a) Stedman diagram: range and resolution
Cmax :sharpest curvature; Zmin : smallest height difference;
8max :steepest slope; Amin :shortest wavelength;
Rp :Iargest height difference; Amax :longest wavelength

I /Cmax //8max
Rp/2 I / ///
- ---I------~VQ~~"T7::~

I
Log A I
I

i~~88&~
1/X)<X)<x~<~<~,(';

Amin I Log A Amax

(b) Stedman diagram: stylus instrument [After Thomas, 1999]

8max:45
Im Rp:5mm

1 mm

l~m

rmax:5~m Zmin:5~m

1m

Amin: 2~m Amax:l00mm

Figure 240. Stedman diagrams for instrument assessment of comparative performance. [Source: Stedman (bl. 1987.]

features of the particular instrument in question to this a110wing acceptable measurements to be made.
be found, namely height, surface wavelength, slope In a similar manner the vertical range (Rv ) and reso-
and curvature, which can be represented at any lution (r) restrict amplitudes, with the horizontal
potential scales within an individual uncomplicated range (Rh) and resolution (rh) restricting the wave-
chart of AW space. lengths that can be measured. For other classes of
If one considers the instrument or probe interac- profilometers typified by those based on optical or
tion with the sinusoidal profile, it fo11ows that most other non-contact probes, analogous limit lines can
instrument parameters enforce limits on the sine be derived. For example, the numerical aperture of
wave ranges that can be reliably measured. By way an objective through which a beam is focused to
of illustration, the tip radius (R s ) of a stylus must not form an optical probe will limit the maximum slope
exceed Rmin of a sine wave that is to be drawn. Rs can that can be sensed.
be plotted as a limit line of slope 2 in low AW space, In a co11ective sense, these limit lines will define
with only sine waves to the right of the limit line a polygonal zone, within which a11 the conditions for
being measurable. The slope errors (Sd) of a datum acceptable measurement are satisfied (see Figure
or reference slideway should idea11y not be greater 239b). This bounded profile can be taken as the
than Smax of any sine wave to be measured. The Sd working zone for the instrument in question, being
plots as a limit line of unit slope, with values above referred to as its AW map.
Comparison of instruments in AWspace that links both specification and function to its
manufacture and metrology.
It is possible to compare a diverse range of instru-
ments in AW mapping space. In a single scan it is
possible to show the preset magnification and scan AWanalysis: high er-order assessment
length for a selected probe or stylus that is con-
strained within the area of AW space, being only The AW analysis for instrument performance previ-
sensed to a particular location of the overall AW ously described can suggest limit lines that delineate
map. Moreover, a more in-depth AW mapping appli- sharp boundary attributes between a "go" and "no
cation might show the effects of setting different go" situations. Often in practice, such boundaries
magnifications and filters, together with other oper- tend to be "fuzzy" and this anomaly can be resolved
ating conditions for a single instrument. Such by higher-order analysis that can quantify this
detailed maps can be an important insight to the b~undary fuzziness as error contours. This strategy
user and can be incorporated in the instrument's wIll enable AW maps to be realised, having closed
manual. Further, it can display the appropriate contours for different levels of uncertainty. Thus, AW
conditions currendy set up for the instrument in the maps can provide a unifying link between the work-
computer's software. piece specification, its surface function, production
processes and subsequent metrological assessment,
hence playing a vital role in quality assurance.
AWspace: surface function and quality Moreover, these AW analysis tools are a vehicle in
presenting complex information in a succinct
assurance manner.
Yet another facet of this important technique of AW
mapping is that surface texture requirements can be
expressed by appropriate analysis of the function of References
a surface. For example, a mirror's optical specifica-
tion could be converted to AW limits by considera- Franks, A. Nanometric surface metrology at the National Physical
Laboratory.Nanotechnology 2,1991,11-18.
tion of diffraction theory and geometrical optics, Stedman, M. How to compare different surface texture measuring
plus the dimensions of the mirror would define lllstruments. Quality Today, April 1990, 35-36.
AW space relevant to the function of the mirror. Stedman, M. Basis for comparing the performance of surface
The application of surface metrology within the AW measuring machines. Precision Engineering 9(3), 1987a, 149-
window can be a route to quality assurance. The 152.
Stedman, M. Mapping the performance of surface-measuring
capabilities of the production processes - grinding, instruments. Proceedings of SPIE, 803, 1987b, 138-142.
lapping and polishing - can also be mapped, with Thomas, T.R. Rough Surfaces (2nd Edn.). Imperial College Press,
AW analysis being utilised as the unifying concept 1999.
AppendixC
Surface texture and roundness: calibration diagrams and photographs

* Major lever length can be accurately set and rneasured to ±0.05 rnrn. This could produce a 0.04% uncertainty in step at the
stylus
**Minor lever length can be set using gauge blocks to an accuracy of ± the uncertainty in the blocks (typically <±0.0001 rnrn),
producing a possible error of 0.001 %.
NB: Total uncertainty budget on actual step generated is ±O.16% (±0.16% (±3.2 nrn on a 2rnrn step height on gauge blocks).

B C

I
I
I
I
I
A/20=B+C/2
I I

127 mm*
~--------------------------------------------------------~~
~ l 16.35 mm*
[;:0.1

Figure 241. The traceable magnification and calibration on surface texture/roundness instruments using the "reduction lever principle".
((ourtesy of Taylor Hobson.)

(a) ,-----_--,
Indicator Indicator outputs

11 8 = R(8) + B(8)
Workpiece

Rotary
table 12 8 = -R(8) + B(8)

Figure 242. The Donaldson Ball Reversal Technique.

 AppendixC &I
Reference sphere

Toroidal shaped
stylus

GJass hemisphere

Rotary tahle

Kinematic location
spheres

72 °, or 36°
indexing

Geneva catch
plate

Figure 243. Calibration of a glass reference sphere. (Courtesy ofTaylor Hobson.)

 (a) Mechanical reversal technique for checking pure component error
using a cylindrical standard

-t:::=-- - - - --<::::r--
Gauge Gauge
direction direction

I
--,---
'"

Position 2: Invert
Position 1: Normal -
Gauge: Reversed
Gauge: Normal
-I + C
1+ C
Air Material

Algebriac addition Bisector Twin-trace bisector: C

of traces : 2C : component error

(b) Mechanical reversal technique for checking pure instrument error

using a cylindrical standard

~­ --=:::::I-
Gauge Gauge
direction direction

Position 1: Invert Position 2: Invert

Gauge: Normal Gauge: Reversed

-I - C -I + C
Material Air
Twin-trace bisector: -I
Algebriac addition Bisector
= -instrument error
oftraces : -2 1
Note: The air / material
!"elationship is reversed

Figure 244. Calibration a roundness instrument for component or instrument uncertainty. (Courtesy of Taylor Hobson.)
 AppendixC

(i) PTB; Depth setting standard: (ii) PTB; Roughness standard:

Ra 0.2 0.5 1.5 Jlm
1.4 3.5 5.8 10.0 Jlm R z: 1.5 3 8.5 Jlm

Roughness
data
information

Figure 245. Typical ca libration standards for surface texture instruments.

Figure 246. A range of three calibration standards for checking electronic stylus surface texture instruments [Courtesy of Rubert & Co.].
 InClustrial metrology . .

Figure 247. Calibrating the 3-D surface scanning instrument to an

accredited artefact. (Courtesy ofTaylor Hobson.)

Figure 249. Calibrating the "Nanostep" instrument in conjundion

with an optical flat. (Courtesy ofTaylor Hobson.)

Figure 248. A non-contact laser diffractometer being calibrated with

a suitable series of roughness comparison blocks. (Courtesy of Taylor
Hobson.)
 Relationships between values determined on "ROCKWELL", "ROCKWELL" Superficial »
"'C
"'C
and TUKO\I Hardness Testers and values determined on other testers. tD
AllhOUQh conversion lables daaling wilh hardness cen only be test and Ihls in turn depends on the degree 01 previous cold working 01 Ihe
::::s
apprOXlmate and never mathematicallyexact. It isof considerable material. The B-scale retatlanships in Ihe labia are basBd largely on annealed t~~r;ijc~~~n~~~~v~I~~~wTe~~I~:v!~:eddln3t~e ~~~r;n agfal~~~~~es Q.
value to be able 10 compare different hardnass scales. Thls labia is melal. lor Ihe low values and cold worked melals lor Ihe higher values. Laboratory . ;:0
based on the assumption that the metal tested is homogeneous to a Therefore. annealed metals of high B~scale hardness such as· austenitic
deplh sevaral limes as greal as Ihe deplh 01 Ihe indanlallon. sl.lnless sieeis, nickels and high nickel alloys da not conlorm closely to these Note 1: A 10mm sleel ball is usBd for 450 BHN and below. A 10mm
general labia•. Nellherdocold-worked melals 01 low B-scale hardness such as carbide ball is used above 450 BHN. C
The indentation hardness values measured on the various sealas aluminium aod softer alloys. Special correlations are needed tor more exact NOte 2: The tenslle strenglh relation 10 hardness is inexacI, even for
depand on tha work hardenlng behaviour ofthe malerial during the relationships in these eases. All valu6s are consistent with ASTM E-140 Tables 1 8teel, unless it is determined for a specific material.

Hardened steel and hard al/oys. ::J:

Soft steeJ, grey and malleabJe cast iron and most non-ferrous metal. CI

C A 0 15-N 3Q-N 45-N VIc"I,. Knoop 800'" G B F G 15-T 3O-T 45-T E K A Kn_ BRINEll B F 15-T 3O-T 45-T E H K A
a..
:::l
Herd- Ha.rCl· HttO· ""- ~,.n"'l I'D
H.nt- H,rdnoSJ Maid-
II~
SOK 110 Kg 100 Kg 1S Kg 30 Kg 451(g nou M" no .. SOK T.nllh1l n_ VI
3000 IIlWIQII \00 Kg 60 Kg so K 151<g 30 Kg 451<g 100 kg 150Kg 60Kg .,." T~IJI. 100 k! eo KO 'SK; 30 .0 ...s I<g ooK ,501<. eo '0 ..... Ha"'· VI
r'~', K. :ooo!!~ Soo K ~ ~r::~1 eo '. $000< 500 ,..,
'0.. .. ~o':l~ &a... · K. o
N N 11'6" 1110' 1110'" 1/10'" 1/10" 1/10" "'... ,,,.. 1111- 1/1(1" 111&'" ,/8" , /8" U)"'''' :::l
. 10mm V~,. Onty 1/'0" 1/1 6" 1/ '6" 1/r
8r.-' 8 .... BI.1o SI,,,, S,." tl/'_" """ e.u &in Ball 8111 BlJI BIll 8t11 B.ll a,1I 8,." B.l1 10 Kg Bol! 8.11 6.n ""n B.U Boot a.1I 8,.,•. ., a.11 <
I'D
~ ~
""'I ..J ..J ..J ;:;:;
~ !i; 0
. < < < 0 .,
Ü Ü Ü 0< ~ ~ ~ ! 0 ~~! ü o'
~. ~ ü ü z
a: 0: ~ 0: :;r U:a: U:a: 0: 0: 0 ::l 'I
..
~ffi cru:o: ~ ce a: a: ~cr ~a: a: a: er j ~ l~ i 0: a: '"a:
0:", w cr 2 ,..,:::l
.... ~ .... ~ .... "'wU:a:.... g!
.. Ei w "'" w a::..., ffiw
~cr crw w w w
~:E w <<( w .... ffi~ ....w t- ....w '" Z' . ~
"' .,w....ffi ....w g e '"t; l ~ ~ ~ t;:f~ eLf;; ~t; ~ t; ... ~ ~ 0.1 l .... e>. ..
"'"' .... "VI
'"w.,....
0
'"t;w :0",
......
:0 .. :>w w CI
....
"'....'" ~~ ~~ ~~ fS! 0 - ....w ~ ~ ~ ~~ ~~ ~~ ~ ~ ~ ~ ~ ~~ • c ....tu .... ...... VI ... ., ... .
'".... j- .... 13.... ö'" ! :
~
'"...w ......... ~ ~
.'" ...
::ll:! "'VI ::l~ ::ll:! ::l~ ::l~
",
e: :I~ ::4' :j~ :j~:j~ ~ ~ ~. ~S ... '" ~'" ~
... '" "'VI ::ll:! ::l~ :j~ ::l~ ......., j-
°w a:0 ! r;
. ::!2 "'w l! 1oUt3 I.Uw I.uI,U
) :j~ :j~ iI4Iw
j~ l.uw
j~ '1j~
11..., ..... w "'W ..... III < « ..,Q '2 ;;:13 ~'" "'w "'w <
~Z
...<w . "'0:'"z
~~ ~~ ~~ ~~ ~~ ~Z .. 2 "Cl :l:z :l:z .. .. ~~ ~~ :l:z :l:z ~~ ~~ 0 o
><0 >cO "'0 "'0 "'0 ;;:" >cO >CO '<0
~~
"'~'" ~~
00: 00: Ocr ~i!i ~i5 ~i!i ~~ ~2i ~~ ~~ ~~ ~~ 8"~ .~~ "'0 ><0 "'0 "'0
00: "'0 "'0 z .::::
g~ g~ 00:
"'" "' .... 0< Oa:; Oa: (,)0: 0
°~ :lcc 0", "0: 00: 00: 00: 00: -~g
0< -> g.:r (J~ (Jer gct 8a: ga: 2I ~' :
8z9Z ", CiS g~ 0< 0< 0<
0:::<: ~! ~~ 0::" CO: ~~ a::<: > .. a:" 0< 0< g~ 0< 0:% 0'" a:x
::+
o:x
t
"'"
,
co'" J ~~ Q:~ ~~ ~~ ~:r ~~ cz:~ Q:~ Q:~ ~m ~ ~1i: t- 0:% a:% 0:% 0::" 0:" 0:: X
"'" "" I'D
VI
80 92.0 86.5 96.5 92.0 87.0 1865 64.5 35d 107 .
'<
100 82.5 93.1 83. I 72.9 61.5 251 201 240 116 50 854 76.9 49.1 22.7 87.0 83
79 91.5 85.5 96.3 91 .5 86.5 1787 - 99 81 .0 92.8 82.5 71 .9 609 246 195 234 114 49 84.8 766 49.0 21 .7 86.5 83.5 34.6 106 . 82 ....o
78 91.0 :845 960 91.0 85.5 1710 98 79.0 92.S 81 .8 70.9 60,2 241 189 1!28 109 48 84.3 76.2 48.3 207 85.5 62.5 34 I 105 81
~
77 90.5 84.0 95.8 90.5 84.5 1633 ;:;;-
97 77.5 92.1 81 I 69.9 595 236 184 222 104 47 83.7 75.9 47.7 19.7 850 61.S 33.7 104 80 o
76 90.0 83.0 95.5 90.0 83.5 1556 96 76.0 91 .8 80.4 68.9 58.9 231 179 216 102 . 46 831 756 47.0 18.7 84.5 61 .0 133.3 103 80 :::l
75 89.5 ~25 95.3 89.0 825 1478 -- 95 74.0 91 .5 79.6 67.9 58.3 226 175 210 100 45 82.6 75.3 463 177 840 60.0 32.9 102 79 ;x,
~ o
74 89.0 '815 95.0 86.5 81 .5 1400
~
..
~ 94 72 .5 91.2 79.1 68.9 57.6 221 171 205 98 44 820 74.9 45.7 18.7 83.5 59.0 32.4 101 78
--,..,
73 88 .5' 81,6 94.8 86 .0 80.5 1323 93 71 .0 90.8 78.4 65.9 NA 57.0 216 167 200 94 43 614 74.6 45.0 157 82.5 58.0 32.0 100 77
0 0 ~
72 88.0 80.0 94 .5 87 .0 79.5 1245 Z 44.3 14.7 82.0 57.5 31.6 99 , 76
--
92 69.0 90.5 77.6 64.6 100 58.4 211 163 195 92 42 80.8 74.3 !!.
71 87.0 795 94.3 86.5 78.5 1160 -, 91 67.5 90.2 77.1 83.8 99.5 55.8 206 160 190 90 41 603 740 43.7 13.6 81 .5 56.5 ' 31.2 98 , 75
70 86.5 78.5 94.0 86.0 77.5 1076 972 90 68.0 89.9 76.4 62.8 98.5 55.2 201 157 185 89 40 797 73.6 43.0 12.6 81.0 55.5 ' 30.7 97 75
69 88.0 78.0 : 93.5 85.0 76.5 1004 946 ' 69 84.0 89.5 75.8 61.8 98.0 54.6 196 164 180 86 39 79 I 73.3 42.3 116 80.0 54.5 30.3 '961 . 74
68 85.6 76.9 : 93.2 &4.4 75.4 9<tO 920 88 62.5 89.2 75.1 60.8 97.0 54 0 192 151 176 86 38 78.6 73.0 41 .6 10.6 79.5 54.0 29.9 95 73
87 85.0 76.1 92.9 83.6 74.2 900 895 67 61 .0 86.9 744 59.8 96.5 534 188 146 172 64 37 78.0 72.7 41 .0 9-6 79.0 NA 53.0 29.5 94 ' 72
88 84.5 75A 92.5 82.8 73.3 865 870 NA 77.4 12.3 40.3 8.6 78.5 100 52.0 29.1 93 . 72
86 59.0 86.6 73.8 58.8 955 52.8 164 145 169 83 36
65 83.9 74.5 92.2 81.9 72.0 832 846 739 85 57.5 86.2 73.1 578 94.5 523 180 142 165 82 35 76.9 72.0 39.6 7.6 78.0 99.5 51.5 28.7 92 71
84 83.4 73.8 91 .8 81.1 71 .0 600 822 722 84 58.0 87.9 72.4 58.8 940 51 .7 176 140 162 81 34 76.3 71.7 39.0 66 77.0 99.0 50.5 28.2 91 · 76
63 82.8 73.0 ' 91 .4 80.1 69.9 772 799 706 83 54.0 87.6 718 55.8 93.0 51 1 173 137 159 80 33 75.7 71.4 36.3 5.6 76.5 98.8 495 27.8 90 69
62 82.3 722 91.1 79.3 68.8 746 776 688 82 52.5 87.3 71 . I 64.8 92.0 50.6 170 135 156 77 32 75.2 71 .0 37.6 4.6 76.0 98.5 48.5 27.4 89 , 69
61 61.8 71.5 90.7 78.4 67.7 720 754 · 670
81 51 .0 86.9 70.4 53.8 910 50.0 167 133 153 73 31 74.6 70.7 37.0 3.6 75.5 98.0 48.0 27.0 88 ' 88
60 81.2 707 90.2 77.5 66.6 697 732 ; 654 NA BO 49.0 86.6 69.7 52.8 90.5 49.5 164 130 150 72 30 740 70.4 38.3 2.6 75.0 97.8 47.0 266 87 67
59 80.7 .69.9 . 89.8 76.6 65 ..5 674 710 634 351 29 73.5 700 35.6 1.0 74.0 97.5 46.0 260 87 - 66 •
79 47.5 68.3 69.1 51.8 89.5 48.9 161 128 147 70
58 80.1 69.2 89.3 75.7 84.3 6S3 690 615 338 28 73.0 69.3 34.5 NA 73.5 97.0 45.0 25.5 86 : 66
78 46.0 86.0 68.4 50.8 68.5 48.4 158 126 144 69
57 79.6 88.5 86.9 74.8 83.2 633 670 , 595 325 77 44.0 85.6 67.7 49.8 86.0 47.9 155 124 141 88 27 72.5 69.5 34.0 730 96.5 44.5 25.0 85 65 :
56 ' 79.0 87:7 86.3 73.9 62.0 613 650 . 577 313 33.0 72.5 96.3 43.5 : 24.5 84 65 ~
76 NA 42.5 85.3 67.1 48 8 87.0 47.3 152 ' 122 139 67 26 72.0 69.0
55 78.5 68.9 87 .9 73.0 60.9 595 58<! 301 720 96.0 42.5 24.3 83 64 ,
! ~g 75 99.6 41 .0 85.0 68.4 47 .8 86.0 468 150 120 137 66 · 25 71.0 88.8 32.5
54 78.0 681 87 .4 72.0 59 .8 577 643 292 24 32.0 71.0 95.5 41 .5 24.0 82, 64
74 99. I 39.0 84.7 65.7 46.8 85.0 46.3 147 118 ':15 , 65 70.5 68.5
0-
11
 I
53 77.4 6s,;.I 86.9 71 .2 586 560 59. 525 283 82
S12 73 98,5 37.5 84.3 85.1 45.8 • 84.5 45.8 145 l 116 132 84: 23 700 88.0 31.0 10.5 95.3 41.0 23.5 . 63 ,
52 76.8 6Ü 86.4 70.2 57 .• 54. 576 273 63 ,
72 98.0 36.0 84.0 64.' 44.8 NA 83.5 45.3 143 114 130 22 695 67.8 30.5 70,0 95.0 40.0 23.0 , 81 ' 62 '
S1 763 63.8 85.9 69.• 561 528 558 496 264
71 97,4 345 83,7 63.7 43.8 '00 82.5 '4.8 141- 112 127 62j 21 69.0 87.5 29.5 69,5 945 390 22.5 81 62
SO 75.9 631 65.5 68.5 55.0 513 542 461 255
70 96.6 32.5 83.' 83. 1 42.8 99.5 815 443 139 110 125 61 20 68.5 67.3 29.0 68.5 943 380 220 60 61
49 75.2 621 65.0 67.6 53.8 49J1 . S26 469 246
69 96.2 31 .0 83.0 62.4 41,8 99.0 81 .0 43.8 137 109 ,23 60 19 68.0 67.0 28.5 88.0 94.0 375 , 21.5 79 61
46 74.7 614 84.5 86.7 52.5 464 510 , 455 238 78 60
86 956 29.5 82.7 61.7 408 98.0 60.0 433 135 '07 '21 59 18 67.0 66.5 27.5 675 935 36.5 ' 213
47 74,1 60,6 83.9 65.6 51 .4 471 495 ' 443 229
67 95.1 28.0 82.4 61 .0 39.8 91.5 79.0 428 133 106 119 58 17 88.5 66.3 27.0 61.0 93.0 35.5 ' 210 78 60
46 736 60.0 83.5 84.8 SO.3 458 460 432 n1 77 ' i 5~
66 84.5 26.5 82.1 60.4 38.7 97.0 78.0 42.3 13' 104 117 57 18 660 66.0 26,0 66.5 926 35.0 !20.5
4S 73.1 59,2 83.0 64.0 49.0 446 466 421 215 76 59
65 93.9 25.0 81 .8 59.7 37.7 96.0 715 416 129 102 116 56 15 65.5 65.5 25.5 65.5 92.5 34.0 20.0
44 725 58.5 82.5 63.1 47.8 434 452 409 208
'-
62,0 62.2 48.7 423 439 400 201 64 93.4 23.5 61 .4 59,0 36.7 95.5 765 41.4 . 127 101 114 NA 14 850 65.3 25.0 85.0 92.0 33.0 NA 15 59
43 720 57.7 ' ·75 56
81.5 61 .3 45.S .,2 184 63 92.8 22.0 81.1 58.4 35,7 95.0 755 40.9 ' 125 99 112 13 845 85.0 2.,0 84.5 91.6 32.0
42 71.5 56.9 426 390 ]4, 56
62 92.2 20.5 60.8 57.7 34.7 94.5 74.5 40.4 124 98 110 12 84.0 84.5 23.5 84.0 9'.5 31 .5
41 70.9 56.2 80.8 60.4 44,3 402 414 381 198
61 91 .7 19.0 605 57.0 33.7 935 74.0 400 122 96 108 11 63.5 843 23,0 63.5 91 .0 30.5 . 73 57
40 70.4 55.4 80.4 59.5 43.1 392 402 371 162
60 91.1 17.5 60.1 56.4 32.7 93.0 73.0 395 120 95 107 , 10 63.0 64.0 220 62.5 90.5 29.5 72 57
39 689 54.6 79.9 58.6 41.8 382 391 362 177 11 ; 57 ,
59 90.5 16.0 79.8 567 31 .7 925 720 390 116 84 106 9 62,0 83,8 21.5 62.0 90.3 29.0
38 694 53.8 79.4 57.7 40.8 372 380 353 171 I . ' 71
58 90.0 14.5 79.5 55,0 30.7 920 71.0 38.6 , '17 92 104 I 8 615 83.,5 20.5 6'5 900 28.0 56
37 68.9 531 78.8 56.8 39.6 363 370 344 186 56 ,
57 89.4 13.0 79.2 54,4 29.7 9 1.0 70.5 36.' : 115 91 103 7 61.0 63.0 20.0 61 .0 89.5 27.0 10
36 86.4 52.3 78,3 55.9 38.4 354 360 336 161
58 98.8 11 .5 78.8 53.1 28.7 905 695 37.7 114 90 101 8 60.5 62.8 195 60.5 69.3 26.0 . 69 55
....'" .
35 679 51,S 77.7 55,0 ~.2 345 351 327 156 ' 0
67.4 SO.8 77.2 54.2 .1 338 319 152 55 88.2 10.0 78.5 53.0 27.7 90.0 885 37,2 112 89 100 z 5 60.0 625 185 60.0 89.0 25.5 69 55
34 342
149 84 87.7 8.5 78.2 52.4 26,7 89.5 880 36.8 111 87 4 59.5 620 180 59.0 88.5 24.5 69 55
33 66.8 SOO 76.6 53.3 34.9 327 33. 311 ~A
53 87.1 7.0 77.9 51 .7 25.7 89.0 67.0 36.3 110 66 3 590 61.8 17.0 585 880 23.5 . 68 54
32 86.3 49.2 76.1 52.1 33.7 318 326 301 146
NA 52 66,5 5.5 77.5 51 .0 24.7 98.0 86.0 35.9 109 85 2 580 61 ,5 16,5 58.0 87.8 23.0 68 54
31 65.8 48.4 75.6 51 .3 32.5 310 318 294 141 68
SI 86.0 4.0 77.2 SO,3 237 87.5 65.0 35.5 108 84 1 57,5 6 1.0 16.0 57.5 87,5 220 54 '
30 85.3 47,7 75.0 SO.4 31 .3 302 311 286 92,0 138
SO 85,4 2,5 76.9 49.7 227 87.0 84.5 350 '07 83 t 0 57,0 60,S 15.0 570 87.0 210 67 53 1
29 84.8 47.0 74.5 49.5 30.1 294 304 279 91.0 135
28 64.3 46.1 73.9 46.6 28.9 286 297 271 90,0 131
27 83.9 452 73.3 47.7 27.8 279 290 264 99,0 128

26 83.3 44.6 72.8 46.8 26.7 272 284 258 88.0 125
25 628 438 72.2 45.9 25.5 266 278 253 87 ,0 123
24 624 43. 1 71.8 45.0 24.3 260 272 247 86.0 119

23 62.0 42.1 71.0 44.0 23.1 254 266 243 84.5 117
22 6'-5 41 .6 70.5 43.2 22.0 248 261 237 83.5 115
21 61.0 40.9 69.9 42.3 20.7 243 256 231 82.5 112

20 60.5' 40.1 69.4 41 .5 19.6 238 251 226 81.0 110

- - - --
Index

A artificial neural network (ANN) 94

Abbe error 43 alogorithms 94
Abbe principle 43,46 architecture 95
Abbott-Firestone curve 30,79,307,308 asperity 178
absorbency 85 aspheric form 46
abusive machining 204,205 assessment of part geometry 152-5
abusive regime 220 assessment of tapered component 178
accelerated machining test 196,197 Assizes of Measures 251
acceptable tolerance errors in products 255 asymmetrieal machining 193
acoustie mieroscopy 128 asymmetry (Ssk) 76
adaptive control constraint (ACC) 214 atomic force mieroscope (AFM) 104,119,128
addition and subtraction: summation in applieations 126
quadrature 270 operating principle 123
adhesion 85 probe geometry 124
ageing 85 atomie force/scanning probe mieroscope
air bearing 149 applieations 126
air temperature reading errors 273 atomie number contrast 107
altered materiallayers (AMLs) 209 atomie resolution 123
altered material zone (AMZ) 187,209 attributes 273
amplitude characterisation 76 sampling 8
amplitude distribution curve 26 Auger electron spectroscopy 104
amplitude parameters 299 austempered ductile iron (ADI) 223
amplitude parameters (average of ordinates) auto-correlation 77
24 auto-correlation function (ACF) 32
amplitude parameters (peak-to-valley) 23 automated surface profiling interferometer 55
amplitude spectrum 286 auto-scaling facility 109
amplitude-wavelength (AW) analysis 317-19 average (Rda) and RMS profile (Rdq) slopes
analogue filter 14 310
analogue transducers 40 average wavelength 311,312
analysis of gothic arch profiles 48 average wavelength analysis: higher-order
analysis of uncertainty: "uncertainty budgets" assessment 319
271-3 average wavelength map AW map) 318
analytieal mieroscopy 115, 116 average wavelength space
angle 48 concept 317
angle-resolved scatter (ARS) 59 surface function and quality assurance 319
anisotropie profile 39 axonometrie projection 48
anisotropie surface 5, 71
apparent roughness 8
archaie length measurement 251 B
areal analysis 112 backlash 233
areal combination 92 backscatter 107
types 92 backscattered electron detection 107
areal motifs 90 balanced turning 228
arithmetie average (AA) 299 ball bearings 148
of the profile 313 ball reversal technique 181,320
arithmetic mean 265 bandpass filter 19
deviation of the assessed profile 24 barrelling 228
summit curvature 78 bearing area fraction 30
arithmetic mean roughness 299 Beilby layer 191
arithmetie roughness (Ra) value 8 bell-mouthed hole 218
artificial intelligence (AI) 96 bell-mouthing 218-20
best fit reference cylinder 176 change tree charts 92
beta function 236,237 chatter marks 6
biology 108 chemical affected layer (CAL) 187
biotechnology structures 108 chemical resistance 195
Birmingham 14 - primary set 75,76 chemical state identification 131
blind holes 200,220 chisel-edge styli 12,39
blunt wedge-shaped indentor 218 chromatic aberration 50
boring 240 circular interpolation 233
boring operations 223,225 circular lay 6
boring process 223 circumferential surface texture 13 7
Bragg law 118,119 CITS (current imaging tunnelling spectroscopy)
Bragg reflection conditions 119 127
Brale indentor 200 clearance fit 139
brightfield 62 climb (or down-cut) milling 233
brightfield image 119 coaxiality 172
Brinell hardness testing 198 coefficient of linear expansion 272
Brown, Joseph R. (1852) 254 coherent convergant beam electron diffraction
bubble-raft patterns 118 115
burnishing 189,196,198 cold field emission (CFE) 116
burnishing-like manner 196 cold-spots 273
colour altitude co ding 80
colour representation 80
C column squareness 283,288
calculation column straightness 283,288
measurement uncertainty 269 combining standard uncertainties 270
calibration 50 comparison block surface assessment 8, 10
check 9 comparison blocks 8, 323
glass reference sphere 321 comparison of DFX files to contour 49
instruments 273 comparison of instruments in AW space 319
material thermometers 273 compensating for corrections 273
Nanostep instrument 324 complex symmetrical profile assessment 48
optical instruments 50 compliant 273
roundness 282 complicated functions: summation in
roundness instrument employing optical flat quadrature 270-1
with gauge blocks 285 component reliability 189
roundness instrument for component of concave circular arc 48
instrument uncertainty 322 concentricity 171
roundness-testing machine using glass sphere cone-shaped stylus 39,282
284 confocal microscope 63
surface texture 274 conical diamond stylus 282
"test sphere" with special-purpose indexing conical stylus tip 39
fixture 284 conical type of styli geometry 11
3-D surface scanning instrument 324 coning error 287
calibration artefacts connectability 90
for spacing measurement 278 constant current mode 123
calibration artifacts constant height mode 123
type B3 277 contact and non-contact operational aspects on
calibration factor 286 surfaces 50
calibration standards for surface texture contemporary surface engineering 244
instruments 323 contour diagram studies 80
candlestick effect 228 contour map 91
cantilever deflections 123 contours: profile curving, or irregular-shaped
cantilever force 125 figure 49
capillary force 125 conventional machining 188
cartographic characterisation 91 convergent-beam electron diffraction 115
cast irons 11 convex circular arc 48
cast replica 8 co ordinate measuring machine (CMM) 149,
catastrophic damage to tip of stylus 282 174,289
cause-and-effect (6M) categorisation for core fluid retention index 77
measurement uncertainty 268 correlation 271
cavities 85 corrosion effects 208
centre line average (CLA) 299 coverage factor 271
centring operation 120,150 cresting standard 287
chains of standards 263,264 crevice-like defects 189
cross-honing operation 6 drill out-of-balance forces 225
crystallographic structure 195 drill wandering action 225
cubit 251 drilled hole slope angle 218
curved datums 46 drilling 217
cusp 6,212,213 geometry 217
cut-off 11,19 drybearings 148,151
cut-offlength 14 dual displays 112
cut-off wavelength 281 dual magnification and digital zoom displays
cutter forces 223 112
cutting force generation 218 duality principle 263,265
cutting forces and tool geometry affects on dynamic AFM (non-contact and discrete contact
roundness 225 AFM) 126
cylindrical grinding 210 dynamic error compensation system 232
cylindrical square - being employed to calibrate dynamic loading 187
column straightnesslsquareness on
rotating-table roundness instrument 289
cylindricity 172 E
cylindricity measurement 173,176 eccentrici ty 171
effect of cutter length on resultant circular
interpolated profile 232
D effect of damaged/worn stylus 283
darkfield 62 effect of filters: roundness 163
darkfield image 119 effect of insert shape and its approach angle in
darkfield observation 61 influencing cutting forces 229
datum kits 46 effect of stylus 37
degraded tool 198 effect on machined surface integrity during
density of peaks 77 surface/cylindrical grinding 210
departures from roundness 137,226-32,241, eight neighbours method 76,77
243 elasticity 122
depth measurement standards 274-6 elasticity image 128
depth of field 103 electrical discharge machining (EDM) 59, 188,
derivation of Abbott-Firestone curve 308 206,207
derivation of kernal roughness depth 309 electrochemical hypothesis 195
derivation of parameters average and RMS electrolysis 208
profile slopes 311 electromechanical milling (ECM) 188
derivation of peak parameters 301 electron beam fabrication 189
determination of measurement uncertainty electron beam texturing 93
269 electron energy-loss spectrometry (EELS) 116,
developed (interfacial) area ratio 78 117
diagrammatic representation of TEM 116 electron holography 115
diametrical measurements 142 electron spectroscopy 131
diamond-turned component 60 electronic receiver gauge 144
differing profiles producing identical Ra values electronics: SEM observation 108
300 electropolishing 59
diffraction order numbers 59 elemental detection: by XPS 131
diffraction patterns 60,116,119 Eli Whitney 254
diffuse reflection 51 Elizabethan yard or ell 251
digital filter 14, 15 elliptica1!hyperbolic (conic) geometry:
digital transducers 41 elimination 46
digital zoom displays 112,113 elliptical shape 141
dimension: utilising form analysis software 49 emerging diameter 211,227,228
dimensional tolerancing 8 end standards 251
directionality: lay 6 energy-dispersive X-ray spectrometry (EDS)
discrimination level 21 112,114,117
dislocations 119 engineering surface 3
dispersion 76 enhanced dual profile 66
displayed profile 152 envelope and me an systems 74
distance-potential energy graph 123 error mapping techniques 232
distortion of peak shape 11 error separation techniques 181
distortion-free assessment for PM parts 11 errors
dominant peak 60, 302 in absolute radii 287
dominant wave 34 in centring and levelling 287
Donaldson Ball Reversal Technique 320 in peak-to-valley measurements 287
drill deflection 219 estimated standard deviation 266
Euclidean error 94 functional performance 8
Euclidean geometry 78 correlations 80
evaluation length 16, 18, 281, 301 for specific topographieal and surface features
exaggerated errors caused by incorrect drill 254
geometry 222 functional surface condition 4
expert systems 66 future instrument trends 291
exponential function 123 future surface engineering applications 246
extinction contours 119
extreme-value parameters 23
extremes: three-dimensional surfaces 76 G
gauge blocks/"flick-standard": stylus deflection
285
F gauge repeatability and reproducibility (GR&R)
facing-off operation 6 289
fast Fourier analysis 34 Gaussian beam 60
fast Fourier transforms (FFT) 34,163 Gaussian filters 89,167-8
fastest decay auto-correlation length 77 Gaussian surface 77,78
fatigue characteristies of surface region 204 gentle regime: machining 202
feedrate 213,225 geometrie corrections by boring operation 227
ferromagnetic material 195 geometrie element fit to an unknown contour
filter attenuation 19 49
filter edge effects 19 geometrie shape: cylindricity 173
filtered elements 19 geometrie tolerance 3
filtered signal 11 geometrieal product specification (GPS) matrix
filtering 13, 162-3 model 263
and harmonics 162 Gibb's phenomenon 14,15
components profile for differing harmonie glaze: grinding wheel 209
effects 167 gloss meter principle 51
filters 11, 166-7 grinding 59
finite stylus tip radius 38 grinding process 208
fixed steady 228
Fizeau surface texture interferometer 53,
54 H
flatness 172 hair size: dimensions 254
fliek-standard 285,286 hand: dimensional measurement 251
floating reamer 220-1 hard etching 203
fluid retention 75 hard reaming 223
fluorescence 62 hardness conversion chart 325-6
flying spot 63 hardness penetration 192
foot: dimensional measurement 251 hardness testing 107
footprinting technique 200 harmonie analysis 163-4, 170
force: AFM/STM 122 harmonie corrections by boring operation 227
force modulation/phase detection 126 harmonie departures from roundness see
form: 3-D topography 88 departures from roundness
form analysis 46 harmonie errors 137
form assessment 48 harmonie suppression 224
form error 5,48 harmonies 34
form-measuring instruments 48 Hatchet stylus 155
form removal 88 health check 9
Fourier analysis: profile 35,52 heat-affected layer (HAL) 187
Fourier transform 33,258 heat-affected zone (HAZ) 206
fractal analysis 91 helieal wandering 218,223
fractal characterisation 78 high-pass filters 14, 19
fractal techniques 79,90 high-sensitivityanalysis 115
Fraunhoffer diffraction pattern 59 high-speed milling (HSM) 110,111,213
free surface: of PM compacts 238 high spot count (HSC) 304,305
frequency response (response time) 292 high-temperature superconductors 114
friction force: AFM/STM 122 holding film 8
friction force mieroscopy (FFM) 119 hole accuracy 218
fringe spacing 54 hole calculation 87
fuel injector systems 173 holographie imaging instrument 98
fuH-film hydrodynamie lubrication 139 holographie interferometry 97
functional 3-D performance 85 honed surface 6,39
functional characterisation 77 horizontal surfaces 153
hybrid characterisation 78 Kirchner-Schultz formula 216
hybrid parameters 29,299,306 Knoop hardness number (KHN) 199
hybrid surface texture parameters: spacing and kurtosis 27,315
depth 308 of assessed profile 27
hydrodynamic bearing 148 of topography-height distribution 77
hydrostatic bearing
air 149
oil 149 L
laser pick-up 42,43
laser triangulation 50-1
lateral force microscope (LFM) 125
image capture and processing 63 lateral modulation LFM 127
image stitching 55 lay 5,21,214
imperial standard yard 251 lay condition 6
indirect measurement 109 lay effects 72
individual feature tolerancing 49 least squares best-fit radius 46
induced chatter 4 least squares circle (LSC) 159
induced radial vibration 137 least squares cylinder 175,176
Industrial Revolution 253 light-emitting diode (LED) source 59
influence of local and profile peaks when peak light section microscope 55
counting 304 line analysis 112
influence of spindle's relative squareness in Linnick and Mirau interferometers 54
concavity of workpiece surface 216 Lissajous figures 219
influence of subsurface features on function loading: surface grinding 209
211 lobing error 142
influence on dynamic strength 195 local dislocations 195
influence on static strength 194 local slope dZ/dX 21
infrared link 45 look-up table (LUT) 112
inhomogeneous workpiece 50 low-noise capacitance sensors 55
in-service conditions 8 low-pass filters 14,19,39
in-service performance 78 lubricant pockets 93
in-service production problems 139 lubricant trapping 78
instrument noise 19 lubrication 85
instrument performance 289-92 lubrication film 8
instrumental response function 258
instrumentation: the future 292
integrating sphere 51,52 M
inter-atomic force 123 M-system envelope 74
interchangeable styli 66 machinability testing 196
interference fringes 53, 54 machined cusp 212
interference instruments 52 machined surface topography 211
interference microscopy 52 macro-cracks 203
interference patterns on spherical surfaces 180 magnetic force microscope (MFM) 119,127
interferometer 55,96, 178-80 magnetisation 195
internal stresses 192 manufacturing envelopes 239
International Bureau of Weights and Measures manufacturing process envelopes 29,236
(BIPM) 251 manufacturing process me an 261
International Prototype Metre 251,252 master reference surface 307
internationally accepted filters 162 material area ratio 93
interpolation and its effect on harmonics 231 materiallength of profile at level "c" MI( c) 22
interrupted surfaces 155 material ratio 30
intervariant lamellar boundaries 119 material ratio curves 309
inverted image 8 material ratio of profile 29
isotrophy index or texture aspect of surface 77 material ratio/Abbott-Firestone curve 306
isotropic surface 5 materials science 108
Maudslay
Henry (1771-1831) 253
J maximum height of profile 23
jobber and split-point drilling 242 maximum inscribed circle (MIC) 160
maximum inscribed cylinder 175,176
maximum profile peak height 23
K maximum profile valley depth 23
kinematic errors/uncertainties 231 Maxwellian concepts 91
kinematic motions 137,181 Maxwellian/Motif landscaping technique 79
mean height of profile elements 23 nanometric instrument: design and operation
mean peak spacing 305 292
mean spacing between profile peaks 305 Nanostep instrument 11,63,64,65,324
me an surface 75 Nanostep stylus 64
me an width of profile elements 29 nanotechnology 289
measured size 142 instrumentation 254,289
measurement fidelity 292 nanotopographic instruments 63-6
measurement loop 46 National Bureau of Standards (USA) 199
measurement methods 150-5 National Physical Laboratory (NPL) 51,252,
measurement uncertainty 268,269 257,258,259
cause-and-effect (6M) categorisation for negative skewness 26,238,314
268 neural networks 66,71,94-6
measuring reflectivity 51 nickel electroforming 9
measurment macros 49 noise 287
mechanical filter 20, 45 non-compliance 273
mechanicallindustrial engineering 108 non-conductive specimens 121
mechanical-rheological model 93 non-contact gauges 49
mechanically affected layer (MAL) 187,198 non-contact laser diffractometer 324
metal removal production processes 190,191 non-contact measurement 96-8
methods of measurement 150-5 non-contact systems 50
M, tres des Archives 251 non-contact three-dimensional surface profiling
metrological environment 269 interferometer 56
Michelson laser interferometer 42 non-conventional machining 188
micro-area diffraction 115 non-destructive testing (NDT) 203,246
micro-crack sites 203 normal distribution 272
micro-hardness applications 200 normalised auto-covariance function (ACVF)
micro-hardness footprinting 199 34
micro-hardness testing 198 Normarski: inspection modes 61,62
microprobe analysis 107
micro-roughness 88
microscope applications 60-2 o
micro-structural changes 190 oblique incidence 54
milled surfaces 6, 214 on-machine inspection 8
milled topography 6 on-screen measurement 112
milling 213,231 operational characteristics of TEM 115
milling operations imparting isotropic optical and laser length measurement 252
machined surface topography 214 optical diffraction 59
minimum circumscribed cirele (MCC) 160 optical distortion 109
minimum circumscribed cylinder 175,176 optical microscopy 62,103,105,107
minimum zone algorithm 48 optical profiler 63
minimum zone cylinder 175 optical sectioning 57-9
minimum zone reference cirele (MZC) 160 optical techniques: reflection and scattering 51
moment of inertia 220 ordinate value 21
Monte Carlo simulation techniques 90 orientation of cutter rake angle geometry 225
morphological analysis 121 origins of uncertainties 267
motion-sensitive pick-ups 40 orthogonal mixed method 181,182
motional kinematics 231 over-tempered martensite (OTM) 204
moving average technique 14 over-travel: of stylus 19
moving-coil transducer pick-up 41
moving steady 228
moving step 211,227,228 p
multi-boring tools 223 paradigms oflearning 95
multidimensional modular receiver gauge 144 parameter rash 74,299
multidirectionallay 5 parameters derived from
multiple diameter assessment 145 Abbott-Firestone/material ratio curve
multiplication 309
summation in quadrature 270 particulate lay 6
multi-wave standard (MWS) 286 pattern recognition 80,85
peak and valley event 21
peak and valley heights 301
N peak angles 37
N-number 7,8 peak counts 304
N-roughness grade 8 peaks and peak counting 303
nanometre beam diffraction (NBD) 117 peaks and valleys 21
pen recorder 11 profile valley 21
penetration into valleys 11 profile valley depth 21
performance curves for Schmaltz-optical programmable-steadies: for workpiece support
sectioning-microscope 58 228
phase effects l3, 26, 44 pseudo-colour display 112
phase grating interferometric (PGI) pick-up pseudo-standards 258
42,43 pyramidal stylus 11
phase-damping behaviour 241
phase-damping effect 241
photo-diodes 59 Q
photo-electric effect l33 quad displays 112
photo-micrographs 62,108,109-11,113,116, qualitative analysis 112
118,254 quantitative analysis 112,133
photo-simulation with lighting effects 80
pick-up 40,42
piezo-electric pick-up 40,41,45 R
pin-cushion indentation: hardness assessment radial cutting force 230
199 radiallay 6
pinhole aperture 62-3 radius of curvature 48
plastic flow zones 202 radius of curvature: of stylus 50
plateau honing 15,85 rapid facing tests 198
ploughs: cutting operation 189 RC2 filter 89
polynomial approximation 89 real surface: definition 21
polynomial expression 46 reamer design 223
porous parts 11, 15 reaming technology 223
portable surface texture instruments 45 rectangular distribution 272
portable three-dimensional measuring reducing measurement uncertainty 273
instruments 89,324 reduction lever principle 286,320
position-sensitive photodiode (PSPD) 125 re-entrant features 11, 13
position-sensitive pick-ups 40 reference cylinders 175
positive skewness 26,314 reflectivity 51,85
powder metallurgy (PM) compacts 11,237 reflectivity testing 51
power spectrum analysis 234 relative material ratio 31
harmonics for machined surface 235 relative precision index (RPI) 267
of machined surfaces 234 relative size and scale of things 253
sub-harmonic and total spectrum 236 release of strain 137
precision reference specimens 8,323 replica kits 8
predictable accuracy: its evolution 253 replica surfaces 8
premature failure state 6 replicas: of surfaces 9
pressure tightness 137 representation of material
primary profile 20 ratio/ Abbott - Firestone curve 306
primary profile me an line 20 re-sampling operator 80, 86
probing uncertainty 290 residual stress 4, 194,206
roundness and form 288 in machined surfaces 192
process capability factors 260 residual stress and deformation by machining
process capability (p) index 260,261 194
product liability 3 residual stress deformations 192
production cost 8 resistance to galling 75
production processing 188 resistance to rotation 225
profile 3,5 retention of lubricant 78
profile co ordinate measurement artefact 280, rigidity: of rotating centres 227
281 rigidity rule 218,225,231
profile element 21 Rockwell hardness 200
profile element height 21 Rockwell hardness test indentor 198
profile element width 21 Rockwell hardness test indentor scales 200,325
profile extraction 85, 87 root mean square 34,166,300
profile filter 20 root mean square deviation 76
profile height amplitude curve 31 from assessed profile 26
profile height variance 313 root mean square profile wavelength 311,312
profile me an line 15 root me an square slope of assessed profile 29
profile motif method 92 roughness 3, 4, 88
profile peak 21 roughness filtering l3
profile peak height 21 roughness measurement arte facts 10,279,323
profile section height difference 31 roughness profile 20
roughness profile me an line 20 skid or piek-up operation 43
rounding effect: of cutting edge 196 skidless mode 13,44
roundness geometrie shapes 141 sliding datum 13,44
roundness measurement 141,157-60 slope analysis 169
roundness of bearing 139 slope and windows 16-9
roundness reference circles 158-60 smart parameters 39
roundness-related errors 141 smearing: machined topography 198
runout: geometric parameter 171 soft imaging 62
softgauges: reference software-calibration 281
software mapping capabilities 89
5 spacing irregularities 14,60
sampling length 16,17,22-5,33,281,301, spacing measurement artefacts 277-8
301-2,305-6 spacing parameters 29,299,303
sampling techniques 177 spall: surface delamination 211
saw-tooth effect: by drilling 220 spatial bandwidth by lateral range/resolution
scanning electron mieroscope (SEM) 62, 292
104-6,121 spatial characterisation 77
applications 108 specific measurement requirements 255
image processing 112 specification limits 260,266-7
images 108,110-11,113,213 specimens
operation 105 with conductivity 121
soft scanned image 109 with no conductivity 121
scanning near-field optics mieroscope (SNOM) speckle: use of 55
119 speckle patterns 55
scanning probe microscope 121,122 specular reflection 51
applications 126 specular-to-diffuse reflection 51
scanning probe/atomic force microscope 121 spheres and hemispheres: spindie assessment
scanning tunnelling microscope (STM) 119, 283
122,125 spherieal and roundness assessment by error
scanning tunnelling spectroscopy (STS) 119, separation 180-3
127 sphericityerror 137
scatter-based instrumental design 51 sphericity interferometer 178-80
scattered light 59 spin die: calibration of 282
scattered light: light-emitting diode (LED) spindie camber: machine tool 215
source 59 spin die imbalance: dynamies of machine tool
scattering 51 spindie 223
Schmaltz mieroscope 57 split-point drill 238,242
performance curves 58 spread: readings of 266
Schmaltz technique 55 spring cut: machining elasticity effects 241
optieal seetioning principle 57 squareness 172,234
Sei 77 stacking faults: atomic level 119
Sdr 78 standard deviation 265,266
Sds 77 standard uncertainty 269
secondary electron image (SEI) teehnique 108 static AFM 125
secondary eleetron yield 107 statistieal measures 265
secondaryelectrons 105 statistieal parameters 299,313
seleetable filters 13 statistical process control (SPC) 246,260-2
self-eentring iris 154 Std 77
self-sharpening grinding wheel 209 Stedman diagrams 258,317-19
semieonductor deteetor 112 basis for 317
semieonductor memory surfaee 109 for instrument assessment of comparative
sensitive funetion: of reflective surface 57 performance 318
servo-errors: when interpolation milling 234 range and resolution 318
servo-spikes: milling interpolation 233 stylus instrument 318
sharpness: topography 77 stereo sc an electron photomicrograph 62
shot -blasting 6 stochastie output 8
signal-to-noise ratio (SNR) 286 Str 77
simple polarisation 62 straightness 153
sine-centres 144 stress affected layer (SAL) 187
sine wave amplitudes 34 stress corrosion 195
sintering process 6 stride: archaie measurement 251
size and scale 252 stylus cone angle 38
skewness 26-7,237,313-14 stylus damage 281-3
of assessed profile 26 stylus deflection 283
of topography-height distribution 76 stylus fine crown (cresting) 288
stylus forces 11,39,50,66 textured metal sheets 93,94
stylus misalignment 39 texturised surface features 85
stylus speed and dynamies 72 thermal damage 207
stylus-to-component alignment 283 thermalevent 206
summit and rider method 144 thermal field emission (TFE) 116
supervised learning: neural network 95 thermal gradient sensitivity 189
surface alterations 188,190,191 thermal growth effects 223
surface characterisation technique 71 thermally error-mapped instruments 273
surface chatter marks 142 thermo-mechanieal generation 205
surface concavity 216 three-dimensional analysis software 79
surface condition 190 three-dimensional characterisation 74
surface cracks 203 three-dimensional contour analysis 48
surface delamination and fatigue 203 three-dimensional holographic interferometer
surface description 258 98
surface diffraction physies 60 three-dimensional measurements 71
surface displacement 191 three-dimensional parameter characterisations
surface engineering 8,85,243 76
surface filtering 80,83-4 three-dimensional profilometer usage 80
surface form measurement 46 three-dimensional surface texture analysis 54
surface generation method 4 threshold operator utilisation 80,86
surface geometry 13 TiC: tool coating 207
surface grinding 210 time-temperature transformation (TTT curve)
surface integrity 3,62,181,188,192,196,205, 244
207,211 TiN 207
surface integrity generator 206 TiN-coated split-point drill 241
surface layer 188 tip condition measurement standards 275,276
surface magnetism 122 tip fianking 46
surface metallurgy 205 topographieal "change tree" chart 91-2
surface mieroscopy 103 topographieal features 121
surface mount technology 45 topographie al height profiles 111
surface of revolution 137 topographical observation: secondary electron
surface potential 122 imaging 108
surface profile parameters 22 topographical surface detail analysis 103
surface roughness 8,55 topologieal characterisation 90
surface science 121 topologieal surface texture characterisation 91
surface sensitivity 131 torque-controlled machining (TCM) 214
surface technology 3 total height of profile 24
surface texture analysis 66,274-81 total integrated scatter (TIS) 52
surface texture artefacts 10,275,323 total profile 20
surface texture descriptors 6 touch-trigger probe 289
surface texture tieks 5-6 traceability
surface topography 8, 71 of end standard length calibrations 257
characterisation by neural networks 94 of measurement 255
resulting from milled surfaces 216 traceability chart 257
utilising ANNs 96 trackability 72
surfacelsubsurface miero-hardness 191 trackability effect 72
Svi 78 transducer design 11
transmission electron mieroscope (TEM) 104,
114,116,119-20
T sub-nanometric elemental analysis 120
Talyfine instrument 59,60 transmitted electron image 114
Talystep instrument 11,64,73 traverse length 16, 18
taper section angle (TSA) 202 traverse speed 14-15,72-4
taper section magnification (TSM) 202 trepanning effect 220
tapered sectioning 60-2, 103, 199,200,202 tribological action 62, 303
tapping: AFM applications 127 tribologieal cutting effect on surface 196
Taylor curves 197-8 tribologieal factors 75
Taylor's general cutting tool wear relationship tribology 85
205 triple-boring heads 223
ten point height 76,302 trochoidal or three-Iobed shape 141
irregularities 23 tumbling harmonie effect 224
ten point Rz parameter derivation 302 tumbling three-Iobed harmonic shape
ternary element 119 223
ternary manufacturing envelopes (TMEs) 238, turning 192,211,240
240,242 two-dimensional profilometer usage 80
typeA van der Waals theory 123
evaluation 269 variable-inductance pick-up 40
standard uncertainty calculation 270 vertical bandwidth by height range/resolution
type Al 292
calibration arte fact 276 vertical distortions of profile 15
wide grooves with flat bottoms 274-5 vertical skid motion 13
type A2 vibration tendencies: bearings 137
calibration artefact 276 Vickers and Knoop indentors 198-9
wide grooves with rounded bottoms 275 Vickers hardness 198,325-6
type B Vickers hardness number (VHN) 198
evaluation, standard uncertainty calculation virtual gauging concepts 258
270 viscosity 122
evaluations 269 visual and tactile method: roughness
type BI: narrow grooves 275 comparison blocks 8,10
type B2 275 visual enhancement of a surface 109
type B2 calibration artefact 276 volume studies 80,82
type B3 277
type Cl 278
calibration artefact 278 W
type C2 278 wavelength filter 19
calibration artefact 278 waviness 3,4, 16,88
type C3 279 waviness filtering 13
calibration artefact 278 waviness profile 20
type C4 279 waviness profile mean line 20
calibration artefact 278 wedge of air: optical interferometry 54
type D1 wedging effect: hydrodynamics 148-50
calibration artefact, undirectional irregular weighting and filtering of roundness 168
profile 279 weighting factor: filtering 168
undirectional irregular profile 279 weighting function: filtering 15
type D2 white etching 203
calibration arte fact, circular irregular profile white layers 188,203,205,210
279 white phase 203
circular irregular profile 280 Whitworth, Sir Joseph (1803-1887)
type EI 280 254
profile co ordinate measurement artefact 280 Wilkinson, John (1774) 253
type E2 281 window assessment: roundness assessment
type F 169
standards 281 window averages: surfaces 15
typical cut-offs 14 window filtering effect 170
typical surfaces obtained by face or peripheral window slope 169-70
milling operations 215 windows 15,168-70
typical thicknesses of surface layers for Windows-based software options 48
engineering components 245 wiper inserts: milling cutters 215
witness marks 307
working clearances and motor drive
U configurations 223
ultra-precision "Nanosurf IV" instrument workpiece deformation by machining 194
292-3 workpiece roundness error 181
ultrasonic force microscope (UFM) 128 workpiece temperature: establishing 273
ultrasonic vibrations 128
uncertainty budget 271-3
metrology instrument 273 X
spreadsheet 272 X-rayanalysis 107
uncertainty issues 264 X-ray diffraction 119
undeformed chip thickness 225 X-ray photoelectron spectroscopy (XPS) 104,
unfiltered profile 15 130-2
unidirectionallay 25
unit event: surface integrity 187
unsupervised learning: ANN training 95 y
untempered martensite (UTM) 204,206 yardstick: archaic length measurement 251

V Z
valley fluid retention index 78 Zerodur: thermally stable material 63,65,
van der Waals force 123,127 292