Review of Machine Tool Metrology (Spindle)

Kok You Cheng (U084692M)

Department of Mechanical Engineering

In fulfillment of the requirements for

ME4105 Specialization Module for Precision Engineering

National University of Singapore

Year 2011/2012
Contents

List of Figures .......................................................................................................... iii

Chapter 1 Introduction ............................................................................................... 1

Motivation ............................................................................................................. 2

Paper Outline ......................................................................................................... 2

Chapter 2 Theories and Concepts (Spindle Metrology) .............................................. 3

2.1 What is a spindle? ............................................................................................ 3

2.2 Types of Spindle Error ..................................................................................... 4

2.2.1 Synchronous versus Asynchronous errors .................................................. 5

2.2.2 Spatial Errors............................................................................................. 6

2.3 Other Practical Handling Issues:....................................................................... 9

Chapter 3 Literature Review .................................................................................... 10

3.1 Methodologies for Error Separation ............................................................... 10

3.1.1 Multi-probe Error Separation Technique .................................................. 11

3.1.2 Multi-step Error Separation Technique .................................................... 12

3.2 Methodologies of Error Compensation ........................................................... 15

i
 3.2.1 Methods of Thermal Error Compensation based on linear and non-linear

Regression........................................................................................................ 16

3.2.2 Compensation Methods based on Neural Networks ................................. 17

3.2.3 Compensation Methods based on Fuzzy Logics ....................................... 18

3.2.4 Compensation Methods supported by a Grey System Theory................... 19

Chapter 4 Designing a Nano-Precision Spindle Calibrator ....................................... 22

4.1 Hardware (Components of a spindle calibrator):............................................. 22

4.2 Method of Data Acquisition and Analysis ...................................................... 23

4.3 Calibrator Testing .......................................................................................... 24

4.4 Quantification of other errors ......................................................................... 27

4.5 Error Separation Techniques Comparison with the Calibrator ........................ 30

Chapter 5 Conclusion .............................................................................................. 32

References............................................................................................................... 33

ii
List of Figures

Figure 1: Table 1: Key aspects of Precision Metrology .............................................. 1

Figure 2: Schematics of a spindle .............................................................................. 3

Figure 3: Types of Spindle Errors .............................................................................. 5

Figure 4: (left) Synchronous Error (right) Asynchronous Error .................................. 6

Figure 5: (a) Axial error motion (b) face error motion (c) radial error motion (d) tilt

error motion .............................................................................................................. 7

Figure 6: Measuring tilt error motion from (a) axial and face (b) radial...................... 8

Figure 7: 3 different types of error separation techniques ......................................... 10

Figure 8: Three Probe method for roundness and spindle error motion measurement 11

Figure 9: Schematic of multi-step error separation method ...................................... 12

Figure 10: Reversal techniques ................................................................................ 14

Figure 11: Comparison of error compensation based on various indirect compensation

methodologies ......................................................................................................... 16

Figure 12: An example of the Laser Doppler Vibrometer......................................... 20

Figure 13: (left) Setup with ball artifact (right) Setup with master axis artifact......... 22

Figure 14: Schematics of fitting the temperature data using a second order polynomial

................................................................................................................................ 24

iii
Figure 15: Schematics of a double ball artifact targeted by five measurement gages for

five degree of freedom measurements...................................................................... 29

Figure 16: An example comparison results .............................................................. 30

iv
Chapter 1

Introduction

Precision Metrology (a.k.a Dimensional Metrology) is the science of calibration of

equipment and using physical measurement equipment to quantify the dimensions

from and of any given object (etc. size, length, angle, and distance). The end goal of

precision metrology will be to achieve a high level of competency in the 4 following

aspects of measurements [1]:

Figure 1: Table 1: Key aspects of Precision Metrology

The continuously increasing needs for highly accurately machined components has

led to a significant amount of research focus in machine tool design and particularly

towards ways by which the geometric accuracy of machines may be upgraded and

maintained [2]. There has therefore grown up this specific field of metrology

concerned primarily with the geometric tests of the alignment accuracy of machine

tools under dynamic loading conditions.

1
Motivation

The successful manufacture of machine tool such as spindle, for use in a range of

applications critically depends on the ability to confirm their quality. The author seeks

to review current research literature on precision metrology, in particular the field of

spindle metrology, to better understand and to be equipped with knowledge in this

field. In addition, the author seeks apply the relevant knowledge to the design of a

spindle calibrator at the nano-precision level.

Paper Outline

The following chapter reviews the existing theories and concepts in spindle metrology.

Chapter three discusses the state of the art in spindle metrology. Chapter four

discusses how the relevant knowledge may be applied to the design of a spindle

calibrator. The last chapter concludes this review paper with a summary of the

precision metrology techniques reviewed.

2
Chapter 2

Theories and Concepts (Spindle Metrology)

2.1 What is a spindle?

A spindle is a device that provides for rotation between components. It provides

stiffness, load capacity, accurate and repeatable rotation

Figure 2: Schematics of a spindle

Mechanical components of a spindle:

 Rotor

 Stator

 Bearings in between

The ideal motion of a spindle should be in a single degree of freedom, that of pure

rotation [3]. Movement in the remaining five degrees of freedom is undesired. One of

the research goals will be to eliminate these movements.

3
Potential Sources of Disturbances:

 Spindle and/or motorization problems

 External influences, such as thermal gradients, applied forces, external

vibration, or some other-named load

 Environmental influences, such as floor vibration, acoustic noise, and

electrical noise

 Other influences, such as air supply pressure fluctuations, etc.

The research directions of spindle metrology mainly focus on the separation,

compensation, identifications of the various forms of errors [4]. Especially at the

nano-scale, any error in the positioning of the spindle will lead to significant

inaccuracies in the finished products.

2.2 Types of Spindle Error

For any spindle, there are three most significant types of errors that researchers are

looking out for, to prevent during machine/spindle design or to compensate after

isolating the error [5]:

 Spatial–defined in terms of where it is measured on a spindle axial, face, radial,

and tilt motion

 Frequency–describe the error motion in terms of the fundamental synchronous

and asynchronous

 Sensitive direction–fixed (e.g., lathe) and rotating (e.g., boring machine)

4
 Figure 3: Types of Spindle Errors

2.2.1 Synchronous versus Asynchronous errors

If the artifact against which the displacement indicator gauges is perfectly round, the

displacement indicator will measure the total error motion of the spindle. The total

error motion for any of the measurement configurations can be described as having

synchronous and asynchronous components.

 Synchronous error motion is the portion of the total error motion that repeats

from revolution to revolution [6]. The asynchronous error motion is the

portion of the total error motion that does not repeat from revolution to

revolution. In mathematical terms, the synchronous error motion is made up of

harmonics that occur at integer multiples of the rotation frequency of the

spindle.

 Asynchronous error motion is usually reported as the maximum width of the

total error motion polar plot for all of the revolutions of data [7]. Figure 4

illustrates polar plotting of the total, synchronous, and asynchronous error

motion. In Figure 4a, the synchronous error motion is the bold line plotted

over the total error motion plot. Figure 4b shows the asynchronous error

5
 motion polar plot. The data shown consists of 8 revolutions with 256 samples

per revolution.

 Conversely, the asynchronous error motion is the portion of the total that does

not occur at integer multiples of the rotation frequency. In practice, the

synchronous and asynchronous can be determined in either the time or

frequency domain. When using a polar chart, the synchronous is easily found

by radially averaging the data at equal angular increments around the polar

plot. The asynchronous can be plotted after subtracting the synchronous from

each revolution of data.

Figure 4: (left) Synchronous Error (right) Asynchronous Error

2.2.2 Spatial Errors

In most circumstances, a spindle will have a non-rotating part, or stator, and a rotating

part, or rotor. Mechanical (balls, rollers, etc.) or fluid bearings (air, oil, etc.) are used

to allow rotation of the rotor with respect to the stator [8]. For the spindle, perfect

rotation is the ideal type of motion; motion in any other degree of freedom is

considered an error motion. If a reference coordinate is attached to the rotor with the

6
Z axis aligned with the axis of rotation, any linear motion in X, Y, Z, or angular

motion in θ x , or θ y is an error motion. Figure 5 shows schematics showing the

measurement of the spindle error motions.

Figure 5: (a) Axial error motion (b) face error motion (c) radial error motion (d) tilt error motion

From the figure below, the tilt error motion is calculated from the axial motion

measurement error, Z(θ), and the face error motion measurement, F(θ) , which are

separated by the distance r using the first equation below. The same calculation can be

made from two radial measurements, R1 (θ) and R2 (θ), separated by distance L as

shown in second equation below. The tilt error motion is designated as A(θ).

7
 Figure 6: Measuring tilt error motion from (a) axial and face (b) radial

All spindle error motion measurements are made with the displacement indicator

targeting a precision artifact mounted to the rotor of the spindle. The artifact used in

spindle metrology is typically a precision sphere, cylinder, or flat that has a high level

of roundness for a radial measurements or circular flatness for face measurements.

8
For most types of spindles, the form accuracy of available artifacts is typically at least

10 times better than the magnitude of the spindle error motions. When this is the case,

the form error of the artifact (typically 25 to 100 nm) is safely neglected and the

measured values from the indicator are treated as the error motion of the spindle.

However, the error motions of ultra-precision spindles approach or surpass even the

highest quality artifacts requiring either a reversal or some other error separation

technique to accurately determine the error motion.

2.3 Other Practical Handling Issues:

Practical issues related to handling of spindle and its associated machine that needs to

be taken into consideration:

 The indicators may not be perpendicular to the artifact.

 The indicators may not be lined up properly on the equator of a spherical

artifact.

 Structural vibration may cause motion that looks like spindle error.

 The apparent artifact surface is filtered by the indicator tip or capacitive

electrode size (this is often a benefit–short wavelength spindle errors are

visible, but short wavelength artifact errors are attenuated).

 Rotating sensitive direction test results differ from fixed sensitive direction

results.

 Data acquisition noise and resolution

 Temperature fluctuations from insufficient spindle cooling

The following chapter discusses the research directions in this field over the past

years.

9
Chapter 3

Literature Review

The research directions of spindle metrology mainly focus on the separation and

compensation of the various forms of errors. Especially at the nano-scale, any error in the

positioning of the spindle will lead to significant inaccuracies in the finished products.

3.1 Methodologies for Error Separation

In all spindle measurements, the displacement indicator will measure a combination

of the spindle error motion and artifact error. As the roundness of spherical master

artifacts approaches the error motion of the spindle under test, the information needs

to be separated [10]. Currently, a multitude of error separation methods have been

developed to separate the errors from machine and part. It is pertinent to carry out

separation to isolate and accurately determine the motion error from the artifact error.

The error may then also be analyzed and compensated. In the existing literature, there

are three key categories of methodologies:

Figure 7: 3 different types of error separation techniques

 Reversal technique requires two measurements. Both the artifact and

displacement indicator are moved between measurements. Reversal is

theoretically superior to other methods, but requires high-quality hardware to

10
 achieve nanometer-level results. A good choice for calibrating artifacts at low

speed.

 Multistep technique requires many measurements. Only the artifact is moved

in equally-spaced increments [11]. Best left for situations when moving the

displacement indicator is difficult.

 Multiprobe technique uses three or four probes simultaneously. Neither the

probes nor the artifact are moved. A good choice for high-speed applications.

3.1.1 Multi-probe Error Separation Technique

Figure 8: Three Probe method for roundness and spindle error motion measurement

This technique uses multiple gages to target an artifact, collecting redundant

information to separate errors [12]. This method, commonly known as a multi-probe

error separation method, uses three or more probes, to measure the combination of

spindle error motion and the artifact out of roundness. Figure above shows the

measurement of an out of round part with roundness profile, P(θ) , with three

displacement indicators, A, B, and C. Indicators B and C are separated in the XY

plane from indicator A by angles φ and ψ. As the artifact rotates on the spindle, three

simultaneous measurements are made, mA (θ), mB (θ), and mC (θ). This approach

works directly with the measured data even in the presence of asynchronous error

motion of the spindle.

11
Accurate knowledge of the separation angles of the probes and close matched

sensitivity of the gages are required. Ideally, the gages should be aligned to be in the

same axial plane so that if the spindle were perfect, the only difference in the

measurements would be a phase shift of the roundness. Much work has been done in

investigating the effect of the angular spacing of the gages [13]. The angular spacing

can distort the harmonic content of the measurement. For example, when the gages

are spaced evenly, low order harmonics are suppressed. The effect becomes less

severe for asymmetric arrangements but never offers complete separation of spindle

and roundness errors. Further research using four or more probes to reduce the

suppression of harmonics has been performed; however, the use of additional probes

can introduce additional errors from imperfect gage alignment.

3.1.2 Multi-step Error Separation Technique

The multi-step error separation method is a technique in which the artifact is indexed

in angular increments without moving or adding additional probes. Figure below

shows a schematic of the measurement method.

Figure 9: Schematic of multi-step error separation method

12
The probe or displacement indicator measures the error motion of the spindle from the

same orientation angle for each measurement. Each of the measurements can be

written in the form of below equation, which contains the radial error motion of the

spindle, R(θ ), and the artifact error, P(θ ) . The artifact form error, however, is phase

shifted by increments of φ in each measurement.

Summing each of the measurements results in the first equation below. Solving it for

the radial error motion of the spindle gives the 2nd equation.

The artifact error can be expressed as a Fourier series as shown in the below equation

Reversal Techniques

Donaldson published a technique illustrating his technique to separate roundness from

radial error motion [14]. This method, which is analogous to straightedge reversal, is

shown schematically in the below figure.

In the first setup the angular orientation of the stator, rotor, artifact, and displacement

indicator are aligned as designated with the black fiducial marks in the below figure.

In the second setup, the displacement indicator and artifact are rotated 180º with

13
respect to the rotor and stator of the spindle. The measurements M1 (θ) and M2 (θ) in

figure 10 are a combination of the spindle’s radial error motion, R(θ), and the

roundness or form error of the sphere, P(θ) .

Figure 10: Reversal techniques

For the second setup the spindle errors reverse sign resulting in two equations that

require only algebraic manipulation to solve for the artifact roundness and spindle

error. Adding the two equations above results in the roundness or form error of the

artifact (first equation below), P(θ) while subtraction results in the error motion of the

spindle, R(θ ), (2nd equation below).

14
3.2 Methodologies of Error Compensation

The essence of machine tool error compensation lies in their measurement or

forecasting in real time and based on it introducing proper corrections to the control

system or to tool path [15]. Based on the method of determining errors, the following

methods of error compensation are distinguished:

 Direct sensor-based methods based on a direct error measurement by means of

sensors distributed in working space of a machine tool,

 Indirect sensor-based methods based on error forecasting with the use of their

mathematical models which utilise information e.g. from temperature

measurements in characteristic locations of machine tool structure, from

measurements of geometrical errors and from measurements of deformations,

 Indirect sensor-less methods based on error forecasting using a mathematical

model which utilises machine internal data, e.g. spindle rotational speed; in

these methods dedicated sensors placed on a machine tool are basically not

used – only these which are integral part of drive control systems for

controllable axes.

 Hybrid methods based on error forecasting based on mathematical models

which use information e.g. from temperature and deformation measurements

in characteristic points of machine tool structure and also take into

consideration internal data of a machine tool, such as spindle rotational speed,

feed, as well as ambient, motor, ball screw nut temperatures, etc.

For error compensation, usually indirect methods are applied in which quantities are

measured which are easier to acquire, e.g. in case of thermal errors – temperature in

specific machine tool points. Errors to be compensated are forecasted based on a

15
mathematical model of the entire machine tool or one of its assemblies which has a

decisive role in generating such error [16]. By applying indirect compensation

methods errors can be followed and monitored in real time, independently from the

cutting process, without the need of interrupting it. The basis of such monitoring in

case of thermal errors is constant observation of temperatures. However, a

disadvantage of indirect error compensation methods is the need of fitting a machine

tool in costly sensors and measurement systems. Below show some of the commonly

used types of Indirect Compensation Methods.

Figure 11: Comparison of error compensation based on various indirect compensation methodologies

3.2.1 Methods of Thermal Error Compensation based on linear and non-linear

Regression

 Indirect sensor-based compensation methods, in order to determine a

mathematical model of errors in form of a polynomial, usually utilize linear or

non-linear regression. Based on it corrections of machine tool errors are

introduced [17].

16
  Such function is mainly applied for compensating thermal errors. Input data of

these functions are temperature values acquired from sensors placed on a

machine tool, while their output is in form of needed correction values for

controllable axes.

 The main problem here is to localize these heat sources which in greatest

degree influence the change of machine tool accuracy, i.e. thermal

displacements accompanying temperature changes. This is usually

accomplished based on technical documentation of a machine tool and on a

numerical analysis of identification measurement results for temperatures and

thermal displacements.

 Such numerical analysis verified experimentally allows very precise

identification of the error model for a broad range of operational parameter

changes, machining conditions and environment influence. In places where,

according to the analysis, temperature rise significantly influences thermal

errors, temperature sensors are installed.

3.2.2 Compensation Methods based on Neural Networks

 In the field of machine tool development, neural networks are used mainly for

minimisation of thermal errors. Based on data acquired from the machine and

measurement system which define real position of spindle and temperature of

machine tool components, it is possible to define the difference between actual

and desired state, i.e. predicted thermal error of a machine tool in the analysed

controllable axis [18].

 Data acquisition process can be realised at constant, rising or randomly

varying spindle rotational speed. Based on acquired data, a neural network

17
 must be trained with the values of needed spindle tip location correction by the

control system at specific operational parameters, such as spindle position,

ambient temperature, spindle rotational speed, etc. Data acquisition process

can last for a few to few tens of hours [19]. After this time has passed, it can

be assumed that the neural network can generalise the problem, which means

that it can provide proper error correction values for operating parameters

varying in predefined ranges.

 Based on experimental investigations, it can be assumed that the application of

a neural network for compensating errors can increase machine tool accuracy

even by 75-80%. However, with it arises a problem of unknown (implicit)

form of a function, based on which compensation is realised. Even if one

succeeds at defining such function in the explicit form, it will be much more

complicated than e.g. polynomial function acquired by means of the regression

analysis. This renders it difficult to implement compensation inmachine

control system and requires the application of an additional external computer.

 Another solution for such problem is the application of digital signal

processors (DSP’s) which, compared to standard processors, are able to

execute parallel mathematical operations. It must however be taken into

account that the less complicated is the function, the quicker are calculations

realised.

3.2.3 Compensation Methods based on Fuzzy Logics

 Fuzzy logics can be applied in all places where it is impossible to use a

mathematical model or when mathematical model could be excessively

complex.

18
  Fuzzy Logic Controller (FLC) is composed of three blocks: – fuzzification

block which transforms numeric values to linguistic, – inference block with a

set of rules, defuzzification block defining output values.

3.2.4 Compensation Methods supported by a Grey System Theory

 Grey system theory can be used to search for relationships between various

quantities [20]., for modelling and forecasting short-period time series in

which data is incomplete and uncertain, as well as to help in decision making.

 In machine tool error compensation, grey system theory is used to support

other mathematical modelling tools, such as neural networks, linear regression

and fuzzy logics.

 An example of utilising the GST method is the minimisation of temperature

sensors to be used for defining the correction function.

For mechanical micromachining processes, the trajectory that the tool tip follows as

the tool rotates is correlated to the dimensional form, and surface quality that can be

obtained. The tool-tip trajectory (except the tool form error) can be best analyzed

through measurements of the radial motions from a geometrically precise artifact

attached to the spindle. Furthermore, due to the dynamic response of the spindle (and

the supporting structure) to the excitation from the eccentricity and white noise, the

measured tool-tip trajectory exhibits a dynamic behavior (which is sometimes referred

to as the dynamic runout). These motions may be referred to as the speed-dependent

radial motions. These motions include radial error motion of the axis of rotation, out-

of-roundness of the tool (artifact), and eccentricity of tool (artifact).

19
As an important contributor to the tool-tip trajectory errors (and thus to the speed-

dependent radial motions), it is critical to develop methodologies to measure the

spindle error motions [21]. Various methodologies have been developed to measure

these errors. Some of them measures and separate the artifact error from the motion

errors while others that don’t are described below:

Figure 12: An example of the Laser Doppler Vibrometer

 One of the effective non-contact measurement techniques uses laser Doppler

vibrometers (LDVs) to measure the radial motions of an artifact while the spindle

is rotating. Rantatalo et al., for instance, measured the radial motions of an artifact

using a spindle up to a rotational speed of 24 000 r/min. The analysis was mainly

restricted to the effects of speckle noise and crosstalk on LDV measurements from

a rotating surface [22]. The results were compared to those acquired at low speeds

using capacitance displacement sensors. It was concluded that, if artifacts with

surface roughness less than half the wavelength of the incident light were utilized,

accurate radial motion measurements can be obtained using LDVs.

 Tlusty introduced the method of using two gauge heads to look at an eccentrically

installed ball, visualizing the measurements as Lissajous plot on oscilloscope. This

20
 method has been recognized as the rotating sensitive direction measurement in the

current spindle metrology standards.

 Bryan presented a method which uses a spherical target and two eccentric cams to

visualize radial error motion on an oscilloscope. This method has been recognized

as the fixed sensitive direction measurement in current standards [23].

 Vanherck and Peters modernized spindle metrology test setup by replacing

eccentric cams with rotary encoder, and using Fourier transform to process the

measurement data.

21
Chapter 4

Designing a Nano-Precision Spindle Calibrator

With the increasing demand for nano-machine tool metrology, it is pertinent to

advance the state of the art in spindle calibrator, to be capable of calibrating spindles

to nano-level precision. Therefore, it will be good to understand the methods to apply

the concepts, theories and knowledge in spindle metrology to design and build a

spindle calibration device, to resolve and quantify the nanometer level error motions

of ultra-precision axes of rotation. The calibrator needs to possess the ability to

calculate all error motions (radial, axial, and tilt for fixed and rotating sensitive

directions) of the spindle under test.

Figure 13: (left) Setup with ball artifact (right) Setup with master axis artifact

There are a few stages to the design of a spindle calibrator:

4.1 Hardware (Components of a spindle calibrator):

i) Type of spindle that you want to test

ii) The type of rotary table that the spindle is mounted on (ultra-precision table)

22
iii) Pilot to facilitate the quick and accurate reversal of artifacts with respect to the

spindle (nano-precision)

iv) Chuck locating for reversal of the artifact with respect to the rotor of the spindle

under test

v) Bolting the entire assembly to an universal measuring machine

vi) Capacitive gages targeting artifacts (nano-precision gages)

4.2 Method of Data Acquisition and Analysis

One possible method of analysis via using the Discrete Fourier Transform (DFT).

Low pass filtering and removal of the fundamental components of the data acquired

(e.g., eccentricity of an artifact for a radial error motion measurement) can be

performed in the Fourier domain by removing the content from the appropriate bins

on the complex vector obtained. This type of filtering imposes a brick wall cutoff

without distorting the phase of the components in the pass region, which is important

in roundness and error motion analysis.

The inverse DFT is then used to return the complex vector data to the original

sampled time domain. With the data back in the spatial domain, traditional

visualization techniques such as polar plotting can be applied to the data sets. These

frequency domain techniques work well for the low asynchronous error motions of

the spindles. Care is taken when measuring rolling element bearings where the

repeatability may influence the amount of leakage of spectral content into adjacent

bins in a data set.

23
4.3 Calibrator Testing

The objective of the calibrator testing is to determine a holistic error budgeting that

quantifies the accuracy of the spindle calibrator. Error budgeting originated from

Donaldson, the man who came up with the Donaldson reversal technique for error

separation. It has been used as a deterministic tool, to identify and estimate the type

and magnitude of errors in a machine. Once errors are incorporated into the budget,

the error sources may then be dealt with systematically and deterministically.

Components of the error budget may be determined by two ways:

1) Statistical methods (measured directly or modelled mathematically)

2) Other methods (etc., manufacturer specifications, calibration certificates or

estimates based on observations)

Figure 14: Schematics of fitting the temperature data using a second order polynomial

24
Components of the error budget to be tested:

i) To account for temperature fluctuation

For the spindle calibrator, temperature variation is a reasonable concern. Temperature

gradients cause expansion or contraction of the structural loop of the machine and

affect the measured error motion over the course of any test.

To investigate the role of temperature, simple drift tests may be carried out on the

spindle calibrator. The measurement gages then radially targets the artifact and the air

temperature may be measured with a thermistor within a few millimetres of the gage

location. The measured drift is then fitted by a suitable approximation and the errors

caused by temperature variation are conservatively estimated.

ii) To account for vibration

For any machine tool, energy sources in the machine and environment cause vibration

and affect the accuracy of any machine measurement. Ideally, the sources of vibration

should be eliminated to improve the accuracy of measurement. However, this is not

always possible; laboratories and factories alike will always have ambient vibration

caused by sources ranging from worker activity to operation of heavy machinery.

Isolation techniques and devices should be implemented to minimize the

transmissibility of ambient vibration to metrology machines. For the spindle calibrator,

the vibration level near the setup can cause disturbances with magnitudes much

greater than the nanometre level error motions that are being measured. This will then

be unacceptable. Ultra precision passive elastomer isolation mounts should be used to

reduce the low frequency transmissibility and higher frequency transmissibility noises.

25
Modal analysis of the calibrator structure should be carried out to determine the

usable bandwidth of the calibrator.

iii) To account for gage noise

This refers to the calibration of the measurement gages and this data should be

provided by the manufacturer and should be included as a term in the error budget.

Testing of the specifications given by the manufacturer may be done by performing a

gage test where the gage is capped with a small conductive cover, to offer a very

small structural loop (minimizing effects of vibration and temperature) and ensuring

proper ground. The gage should also be placed in a thermal insulated environment

with a thermistor to record the air temperature. Any noises are then measured.

Testing a combination of the above:

The above experiments approximate the contribution of each error budget term

independently. In practice, the vibration of the machine, noise in the gage, and

thermal drift happen simultaneously. Therefore, to estimate their combined

contribution, the calibrator is arranged with the measurement gages aimed at a non-

rotating target. Data is acquired with a signal analyser and post processed to look at

the combined structural error motion and thermal drift.

Reproducibility and Repeatability test:

Repeatability of the calibrator measurements is an indicator of the performance and

the resolution of the spindle calibrator. To determine the repeatability, synchronous

radial error motion measurements are made over a period of time to the type of

spindle chosen. The maximum asynchronous error motion for each test should also be

26
recorded. Repeatability after Donaldson reversal is demonstrated by performing

multiple reversals with the same spindle arrangement.

The repeatability can be concluded by comparing the synchronous error motion tests

at each point sampled within each revolution of the spindle. For these tests, the

maximum deviation between the forum tests for each spindle should be in the

nanometres range, peak to peak.

If the tests demonstrate the sub-nanometre repeatability of both the spindles and the

measurement procedure, the error motion of the spindle can be confidently mapped

for a variety of applications.

A short overview:

Testing on the calibrator should shows that sub-nanometre repeatability and

reproducibility with nanometre uncertainty levels is obtainable for measurement of

ultra-precision spindles. Bandwidth limitation by the Structural modes of vibration

should be determined. Temperature variation should be compensation with a suitable

fit to the data. The noise of the measurement gage should be quantified to be in the

nanometre range. Finally, repeatability and reproducibility must be tested and results

should show that synchronous radial error motion measurements performed to have

standard deviation in the nanometre range.

4.4 Quantification of other errors

The other errors that need to be quantified are the asynchronous error motion & radial,

tilt and face error motion. However, before quantification of synchronous radial, axial,

and tilt error motions of the spindle, it is necessary to quantify the asynchronous

27
radial and axial error motions of the spindles. The asynchronous error motion of axes

of rotation can result in poor surface finish in machining or errors reading tracks in a

hard disk drive.

Asynchronous Error

Asynchronous error motion refers to the portion of error motion that does not repeat

from revolution to revolution. In practice, it is difficult to accurately measure the

exact asynchronous values for a spindle. The first reason is, when making spindle

measurements, any motion that is not synchronized with the rotation of the spindle

will appear as asynchronous error motion of the spindle. External sources of error

include vibration of the test stand, temperature effects, and electrical noise,

contributors which are not inherent to the axis of rotation. Secondly, the asynchronous

error motion cannot be resolved using reversal techniques.

However, a proper estimation of the asynchronous error motion of a spindle can be

determined. This is possible by making the structural loop of the spindle very small

and stiff so that vibration and temperature effects on the measurement may be

minimized. Mounting the displacement indicator directly on the spindle stator

targeting the rotor may be the best way to quantify the asynchronous motion of the

spindle. One method of estimating the asynchronous motion of a spindle is by using a

Single Point Asynchronous Motion (SPAM) test. This SPAM test does not require a

complete spindle measurement for estimating the actual asynchronous values.

Other errors (radial, tilt and face error motion)

28
Figure 15: Schematics of a double ball artifact targeted by five measurement gages for five degree of freedom
measurements

The B89.3.4M Axes of Rotation standard appendix outlines the procedures for

making multiple measurements on a spindle to determine the error motion at any axial

or radial position. Two radial measurements at different axial locations allow the tilt

error motion to be calculated for a designated sensitive direction. This may be

accomplished in practice by using a cylindrical or double ball artifact or the master

axis method. Once the tilt error motion is known, an axial error motion measurement

allows the face error motion to be calculated. Figure 15 shows this schematically on a

double ball artifact. By using five measurements, the radial, tilt, and face error motion

for any sensitive fixed or rotating sensitive direction can be determined.

29
4.5 Error Separation Techniques Comparison with the Calibrator

Figure 16: An example comparison results

If a versatile spindle calibrator hardware using the rotary table and reversal chuck

which allow Donaldson reversal has been implemented, multi-probe, and multistep

error separation techniques may be performed to separate the radial error motion of

the spindle from the form error of the artifact. An experiment to compare the fixed

sensitive direction radial error motion and sphere form error obtained from Donaldson

reversal, a multistep test, and a multi-probe error separation test may be carried out.

30
The objective of such comparison will check whether if the calibrator is suitable for

all the different types of error separation techniques and how the accuracy of each

technique compares with each other.

31
Chapter 5

Conclusion

The growing miniaturization of consumer products in various industries reflects the

need for greater amount of research attention into the field of precision spindle

metrology. Innovation of new spindle metrology techniques that overcomes current

limitations plays an important role in product quality control and experimental

research purposes. In this review paper, the key aspects of precision metrology field

that have been reviewed are highlighted below:

 Concepts and theories related to spindle metrology are reviewed

 An extensive literature review has been carried out. The research directions of

spindle metrology mainly focus on the separation and compensation of the

various forms of errors.

 Application of the spindle metrology relevant knowledge to design a nano-

precision spindle calibrator is discussed

32
References

[1] Tlusty J. System and methods of testing machine tools. Microtecnic

1959;13(4):162–78.

[2] Bryan J, Clouser R, Holland E. Spindle accuracy. American Machinist

1967;111(25):149–64.

[3] Vanherck P, Peters J. Digital axis of rotation measurements. Annals of CIRP

1973;22(1).

[4] Rantatalo M, Tatar K, Norman P. Laser doppler vibrometry measurements of a

rotating milling machine spindle. IMechE Event Publications 2004;2:231–40.

[5] Castellini P, Montanini R. Automotive components vibration measurements by

tracking laser doppler vibrometry: advances in signal processing. Measurement

Science and Technology 2002;13(8):1266–79.

[6] Halkon B, Rothberg S. Rotor vibration measurements using laser doppler

vibrometry: essential post-processing for resolution of radial and pitch/yaw vibrations.

Transactions of ASME, Journal of Vibration and Acoustics 2006;128(1):8–20.

[7] Marsh ER, Grejda RD. Experiences with the master axis method for measuring

spindle error motions. Precision Engineering 2000;24(1):50–7.

[8] Vallance RR, Marsh ER, Smith PT. Effects of spherical targets on capacitive

displacement measurements. Journal of Manufacturing Science and Engineering,

Transactions of the ASME 2004;126(4):822–9.

33
[9] Liu C-H, Jywe W-Y, Lee H-W. Development of a simple test device for spindle

error measurement using a position sensitive detector. Measurement Science and

Technology 2004;15(9):1733–41.

[10] King-Fu H, Vallance RR, Grejda RD, Marsh ER. Error motion of a kinematic

spindle. Precision Engineering 2004;28(2):204–17.

[11] Grejda R, Marsh ER, Vallance RR. Techniques for calibrating spindles with

nanometer error motion. Precision Engineering 2005;29(1):113–23.

[12] Smith P, Vallance R, Marsh E. Correcting capacitive displacement measurements

in metrology applications with cylindrical artifacts. Precision Engineering

2005;29(3):324–35.

[13] Marsh ER, Couey J, Vallance RR. Nanometer-level comparison of three spindle

error motion separation techniques. Journal of Manufacturing Science and

Engineering, Transactions of the ASME 2006;128(1):180–7.

[14] Lu X, Jamalian A. A new method for characterizing axis of rotation radial error

motion: Part 1. Two-dimensional radial error motion theory. Precision Engineering

2011;35(1):73–94.

[15] Lu X, Jamalian A, Graetz R. A new method for characterizing axis of rotation

radial error motion: Part 2. Experimental results. Precision Engineering

2011;35(1):95–107.

[16] Donaldson RR. A Simple Method for Separating Spindle Error from Test Ball

Roundness Error. CIRP, 21 (1), 1972; 125-126

34
[17] Estler WT, Evans CJ, Shao LZ. Uncertainty Estimation for Multiposition Form

Error Metrology. Precision Engineering, 21 (2-3), 1997; 72-82.

[18] Chetwynd DG, Siddall, GJ. Improving the Accuracy of Roundness Measurement.

J. of Phys. E: Sci. Inst., 9 (7), 1976; 537–544.

[19] Whitehouse DJ. Some Theoretical Aspects of Error Separation Techniques in

Surface Metrology. J. of Phys. E: Sci. Inst., 9 (7), 1976; 531-536.

[20] Mitsui K. Development of a New Measuring Method for Spindle Rotation

Accuracy by Three Points Method. Proc. 23rd Int. MTDR Conf, 1982; 115-121.

[21] Moore D. Design Considerations in Multiprobe Roundness Measurement. J. of

Phys. E: Sci. Inst., 22 (6), 1989; 339-343.

[22] Zhang GX, Wang RK. Four-Point Method of Roundness and Spindle Error

Motion Measurements. CIRP, 42 (1), 1993; 593-596.

[23] Zhang GX, Zhang YH, Yang ZL, Li Z. A Multipoint Method for Spindle Error

Motion Measurement. CIRP, 46 (1), 1997; 441-445.

35