Int J Adv Manuf Technol (2014) 70:957–963
DOI 10.1007/s00170-013-5339-5
ORIGINAL ARTICLE
Investigate the spindle errors motions from thermal change
for high-precision CNC machining capability
Ahmed A. D. Sarhan
Received: 16 December 2012 / Accepted: 24 September 2013 / Published online: 6 October 2013
# Springer-Verlag London 2013
Abstract As the demand for high speed and highly accurate
machines has significantly increased, error motions from ther-
mal change, which is up to 70 % of the total machining error,
is found to be the main hurdles to overcome in improving the
accuracy of CNC machine tools. In this research work, the
authors installed four eddy current displacement sensors in the
spindle structure near to the front bearing to monitor the
spindle offset in the bearing level, which is mainly attributable
to the thermal error motions of the spindle. In addition, anoth-
er three capacitance sensors are mounted on the machine table
level and aligned with the x-, y-, and z-axis of the machine to
monitor the spindle shift in the table level to find out the
correlation between temperature change and the thermal error
motions of the spindle. To measure the temperature changes,
we attached thermal sensors in the machine and cooling
system. The estimation of the spindle thermal displacement
based on temperature data from these thermal sensors can
provide more information for the monitoring of thermal error
motions of the spindle.
Keywords Machine tool spindle . Thermal error motions .
High-precision machining . Displacement sensors
A. A. D. Sarhan (*)
Center of Advanced Manufacturing and Material Processing,
Department of Engineering Design and Manufacture, Faculty of
Engineering, University of Malaya, Kuala Lumpur 50603, Malaysia
e-mail: ah_sarhan@yahoo.com
A. A. D. Sarhan
Department of Mechanical Engineering, Faculty of Engineering,
Assiut University, Assiut 71516, Egypt
1 Introduction
Modern machine tools must provide high-precision machin-
ing capability and also high productivity. Spindles are funda-
mental to most machine tools. Some people have gone so far
as to describe a machine told as a spindle with support hard-
ware and advanced design machine tools tend to confirm this
point of view [1] and [2]. Modern machine tools have stiff
structure and accurate positioning capabilities. However, the
weak link in a machine tool is often the spindle; so, spindle
analysis is highly needed to identify the factors that are lim-
iting manufacturing processes [3].
A rotating spindle generates an axis of rotation. This is a
theoretical line about which rotation occurs. This axis line has
six degrees of freedom or six different independent motions
that can be experienced. These six degrees of freedom are
linear motion in x-, y -, and z-axis directions; tilt along the axis
of rotation in the x- and y-axis directions; and rotation about
the z-axis line. An ideal spindle allows motion in only one
degree of freedom, which is rotation about the z-axis line.
Motions in the other five degrees of freedom are undesired
error motions. These undesired errors in motions cause deg-
radation in surface finish, roundness, and geometrical accura-
cy which in consequence, will lead to reduction in product
quality [4–6].
The undesired error motions can be classified as geometrical
errors, temperature rise errors, dynamic errors, etc. According
to relevant studies, errors caused by thermal changes are typi-
cally the largest positional error in a properly installed and
operating machine tool [7] and [8]. It is reported that
40–70 % of errors of machine tools result from the errors
caused by thermal changes [9], where the main source of heat
lies in the friction between the spindle and transmission belt,
958 Int J Adv Manuf Technol (2014) 70:957–963

Table 1 Specifications of the

machining center GV-503
Spindle Spindle speed (rpm) 200–20,000
Power 15 min./cont. (kW) 22/18.5
Tool interface 7/24 taper No.40 with nose face contact
Feed drive Max. rapid traverse rate (mm/min) 42,000
Max. feed rate (mm/min) 42,000
Max. acceleration rate (m/S2) x-axis:4.7, y-axis:5.1, and z-axis:7.9
Travel distance (mm) x-axis: 610, y-axis: 510, and z-axis: 460
Machine size Width × length × height (mm) 2,232×3,987×2,892
Mass (kg) 10,100
CNC servo system 64bit CPU (RISC processor) + high gain servo amplifier

and that between spindle and ball bearings. Cooling is provided
by a built-in cooling system in the spindle housing; therefore,
the dynamic balance of heat flux stabilizes after a while, and so
do temperatures. However, temperature fluctuations of a few
degrees, plus or minus, are inherent [10] and [11]. These
thermal error fluctuation motions occur simultaneously on
different directions, so measuring the thermal error motions as
the spindle is rotating at operation speeds is the most important
theme to solve for the machine tool designer and also for the
industrial engineer who use this equipment on the workshop.
In this research work, we examined the machine tool spindle
and/or bearing error motions caused by the thermal changes. To
investigate these error motions, we have integrated the machine
with displacement sensors system at different levels, tempera-
ture sensors system, and a signal acquisition and data analysis
system to construct an independent module measurement sys-
tem of spindle, by means of which it can be easy to find out the
correlation between temperature change and thermal error mo-
tions for the spindle. This can be used to test the quality of
spindles and also provides the basis for compensation of ther-
mal error motion during practical manufacturing process.
2 Experimental set-up to investigate the spindle thermal
error motions
In order to investigate the machine tool spindle and/or bearing
thermal error motions, displacement sensors and thermal sen-
sors are installed on the spindle unit of a high-precision
machining center. The machine used in the study is a
vertical-type machining center (GV503 made by Mori Seiki
Co., LTD.). The spindle has constant position preloaded bear-
ings with oil–air lubrication and the maximum rotational
speed is 20,000 min−1. Table 1 shows the specifications of
the machining center used. Four eddy current displacement
sensors are installed on the housing in front of the bearings to
detect the radial motion of the rotating spindle. The specifica-
tions of the sensor are as follows: the diameter is Ø 5.4 mm
and the length is 18 mm; measurement range is 1 mm; nom-
inal sensitivity is 0.2 mm/V; dynamic range is 1.3 kHz; and
linearity is ±1 % of full scale. Table 2 shows the specifications
of the Eddy current displacement sensor. A thin collar with a
fine cylindrical surface is attached to the spindle as a sensor
target. Figure 1a, shows the Eddy current displacement sensor
locations. The two sensors, S1 and S3, are aligned oppositely
in the x direction, and the other two sensors, S2 and S4, are
aligned opposite in the y direction. In order to measure spindle
temperature, several thermocouples are attached to the spindle
structure. Those thermocouples include T1 and T2 shown in
Fig. 1a. The thermocouple T1 is installed in the same hole
where the displacement sensor S4 is installed, while the other
thermocouple T2 is installed on the body of the spindle unit,
which is near the windings of the built-in motor. The sampling
time and sensor resolution are 60 s and 0.1 °C, respectively.
In addition, another three capacitance sensors C1, C2, and C3
are mounted on the machining table and aligned with the x-, y-,
and z-axis of the machine as shown in Fig. 1b and in Fig. 2. The
sensors C1 and S3 are located in the same side in x direction, and
the sensors C2 and S4 in y direction. The target of the sensors C1,
C2, and C3 is a precision ball mounted onto the spindle using an
adjustable holder. Table 3 shows the specifications of the capac-
itance displacement sensor used in this experiment. The sam-
pling frequency for the measurement is 1 Hz.
3 Experimental result
The displacement and temperature are measured during the
spindle rotation without cutting. The spindle is started at 1,
500 min−1 and kept rotating for 1 h. After that, the spindle
Table 2 Specifications of the eddy current displacement sensor
Detection principle Eddy current
Measurement range (mm) 0∼1
Output scale (V) 0∼5
Sensitivity (mm/V) 0.2
Resolution (μm) 0.4
Dynamic range (kHz) 1.3 (−3 dB)
Int J Adv Manuf Technol (2014) 70:957–963 959

Fig. 1 Locations of the 68 mm

capacitance and eddy current
displacement sensors. a
Locations of the eddy current
sensors, b Locations of the The load center
Front bearing
capacitance sensors of the bearing
system
system

Displacement sensor

122 m m
132 m m
View from upper side
S2

S3 Y S1
Collar Z
X
X 92 mm S 4 Thermocouple T1
(Width:15 mm)
Holder

a Locations of the eddy current sensors

b Locations of the capacitance sensors

speed is increased to 3,000 min−1 for another hour followed by
50-min rest time. After that, the spindle is started again
at 6,000 min−1 and kept rotating for 1 h. Figure 3a, b shows the
temperature of sensor T1 during the entire time of the exper-
iment and the temperature of sensor T2 at spindle speed of
6,000 min−1, respectively.
After the first spindle rotation term, the spindle temperature
is around 14 °C. The increase of the spindle temperature is
about 1.5 and 4.5 °C in the second and third spindle rotation
term, respectively. Figures 4 and 5 shows the output of the
displacement sensors S1∼S4 and C1∼C3 as a result of heat
being added to the machine spindle structure at different
spindle speed.
4 Data analysis and discussion
As shown in Figs. 4 and 5, all the sensor signals involve two
components of fluctuations: the transient component and the
periodic component. The transient fluctuation component can
be seen in the temperature measured by the thermocouple T1,
while the periodic fluctuation component can be seen in the
temperature measured by the thermocouple T2, as shown in
Fig. 3. From Figs. 4 and 5, it is observed that, the fluctuations
in the displacement measurement at different spindle
speeds are same, but the periodicity and range are
different for each spindle speed. The fluctuations peri-
odicity is 480, 300, and 240 s at the spindle speed of
1,500; 3,000; and 6,000 min−1, respectively. The tran-
sient fluctuation component is found to be related to the
thermal extension of the spindle, while the periodic fluctuation
component is found to be related to the cooling control of the
oil, which is circulating inside the spindle housing. In this
research work, the transient fluctuation component is used to
evaluate the spindle error motion caused by thermal changes
in the spindle and/or bearing.
From the results at the spindle side, shown in Fig. 4a–c,
where the spindle motion in x direction is measured by the
960 Int J Adv Manuf Technol (2014) 70:957–963

X S2

Y Displacement sensor

S3 S1

S4

Jacket of the cooling system

T2

Front bearing

Temperature sensors Eddy current displacement Sensors

Amplifier
S4 Real -time
S1 simulator
T1 L. P. Filter
16 bit
Amplifier A/D Board

12 bit
L. P. Filter A/D Board

PC

Amplifier

Table

C2
X

Capacitance sensors
C1
Y C3
Fig. 2 The monitoring system and the experimental set-up

eddy current sensors S1 and S3, the gap between the sensor S1 increment of −0.4 and −1.25 μm at the end of the second and
and sensor target is almost not changed during the entirety of the third spindle rotation terms at 3,000 and 6,000 min−1.
spindle rotation with 1,500; 3,000; 6,000 min−1. However, the From the results at the table side, shown in Fig. 5a–c, where
gap between the opposite sensor S3 and spindle is decreased by the spindle motion in x direction is measured by the
Int J Adv Manuf Technol (2014) 70:957–963 961

Table 3 Specifications of the capacitance displacement sensor a 2

Sensor S1
Detection principle Capacitance sensor 1.5 Sensor S
2
1 Sensor S
Measurement range (μm) 500 3

Displacement µ m
Sensor S4
Output scale (V) −10∼10 0.5
Sensitivity (V/μm) 0.04 0
Sensitivity switch Low -0.5
Bandwidth (kHz) 13.9 (−3 dB)
-1

-1.5
capacitance sensor C1, which is located at the same side of -2
sensor S3 but at the table level, the gap between the sensor C1
-2.5
and spindle is decreased by about −1.5 μm at the end of the
-3
0 10 20 30 40 50 60
Time min
a 19
Spindle speed: 1500 min-1
-1
1500 min b
18 -1 2
3000 min
rest time 1.5
Temperature C

17 6000 min
-1
1
16 Displacement µ m 0.5

0
15
-0.5
14
-1

13 -1.5 Sensor S
1
Sensor S2
-2
Sensor S
12 3
0 50 100 150 200 -2.5 Sensor S4

Time min -3
60 70 80 90 100 110 120
Temperature T1 during the entire of the experiment Time min
b Spindle speed: 3000 min-1
20
c 2
19.5 1.5

1 Sensor S
19 2
Displacement µ m
Temperature C

0.5 Sensor S1
18.5 0

-0.5
18
-1 Sensor S
3

17.5 -1.5
Sensor S4
-2
17
-2.5

16.5 -3
0 10 20 30 40 50 60 170 180 190 200 210 220
Time min
Time min Spindle speed: 6000 min-1
-1
Temperature T2 at 6000 min Fig. 4 The output of the spindle eddy current displacement sensors at
Fig. 3 The measured temperature signals. a Temperature T1 during the different spindle speeds. a Spindle speed 1,500 min−1. b Spindle speed
entire of the experiment. b Temperature T2 at 6,000 min−1 3,000 min−1. c Spindle speed 6,000 min−1
962 Int J Adv Manuf Technol (2014) 70:957–963

a 35
first spindle rotation term at 1,500 min−1. But, later this gap is
Sensor C1 increased by increment of +1.5 μm at the end of the third
30 Sensor C
2 spindle rotation terms at 6,000 min−1. These results indicated
Sensor C
3 that the spindle is shifted in the x direction, expanded, and
Displacement µ m

25
tilted along the axis of rotation in the x direction. As for the
20 spindle motion in y direction, which is measured by the
sensors S2, S4, and C2, the spindle shows similar tendency.
15 As for the spindle motion in z direction, the gap between
10
the sensor C3 and spindle is increased remarkably with the
temperature increase, as shown in Fig. 5a–c. It is increased by
5 increments of 5, 15, and 33 μm at the end of the first, second,
and third terms of the spindle rotation with 1,500; 3,000; and
0
6,000 min−1. These results indicated that the transient fluctu-
-5 ation component is related to the thermal extension of the
0 10 20 30 40 50 60
spindle, while the periodic fluctuation component is related
Time min
to the cooling control of the oil.
Spindle speed: 1500 min-1
b 35
Sensor C1 5 Conclusion
30 Sensor C
2
Sensor C Displacement sensors and thermocouples are installed in the
3
Displacement µ m

25
spindle structure of a machining center to investigate the
20 machine tool spindle and/or bearing error motions caused by
the thermal changes. We present how to monitor the spindle
15
error motions caused by thermal change more precisely by
10 using displacement sensors system at different levels to
construct an independent module measurement system of
5
spindle, by means of which it can be easy to find out the
0 correlation between temperature change and thermal error
motions for the spindle. Accordingly, the displacement sensor
-5 signals involve two components of fluctuations: the transient
60 70 80 90 100 110 120
Time min fluctuation component related to the thermal extension of the
spindle and the periodic fluctuation component related to the
Spindle speed: 3000 min-1
cooling control of the oil. Due to the thermal changes in the
c 35 spindle and/or bearing, the spindle is expanded, shifted, and
tilted along the axis of rotation in the x- , y- and z -axis
30
directions.
Displacement µ m

25 Sensor C1
Sensor C
2
20 Sensor C3

15 References

10
1. Postlethwaite SR, Ford DG, Morton D (1997) Dynamic calibration of
CNC machine tools. Int J Mach Tools Manuf 37(3):287–294
5
2. Heisel U, Gringel M (1996) Machine tool design requirements for
high-speed machining. Ann CIRP 45:389–392
0
3. Xu S, Wang Y, Zhao J, Le G (2008) A study of fault monitoring in CNC
-5
machining of free-form surfaces based on NN-wavelet analysis—tech-
170 180 190 200 210 220 nical communication. Mach Sci Technol V12(3):PP405–PP416
Time min 4. Sarhan AAD, El-Zahry RM (2011) Monitoring of tool wear and
surface roughness in end-milling for intelligent machining. Int J
Spindle speed: 6000 min-1 Phys Sci 6(10):2380–2392
Fig. 5 The output of the table side capacitance displacement sensors at 5. Lin Z-C, Chang J-S (2007) The building of spindle thermal displace-
different spindle speeds. a Spindle speed 1,500 min−1. b Spindle speed ment model of high speed machine center. Int J Adv Manuf Technol
3,000 min−1. c Spindle speed 6,000 min−1 34(5-6):556–566
Int J Adv Manuf Technol (2014) 70:957–963
6. Sarhan A, Sayed R, Nassr AA, EL-Zahry RM (2001) Interrelationships
between cutting force variation and tool wear in end-milling. J Mater
Process Technol 109(3):229–235
7. Kim S-M, Lee S-K (2005) Spindle housing design parameter opti-
mization considering thermo-elastic behavior. Int J Adv Manuf
Technol 25(11-12):1061–1070
8. Bryan J (1990) International status of thermal error research (1990).
CIRP Ann - Manuf Technol 39(2):645–656
963
9. Cao H, Holkup T, Altintas Y (2011) A comparative study on the
dynamics of high speed spindles with respect to different preload
mechanisms. Int J Adv Manuf Technol 57(9-12):871–883
10. Tseng P-C (1997) A real-time thermal inaccuracy compensation meth-
od on a machining centre. Int J Adv Manuf Technol 13(3):182–190
11. Weck M, McKeown P, Bonse R, Herbst U (1995) Reduction and
compensation of thermal errors in machine tools. CIRP Annals 44(2):
589–598