International Journal of Machine Tools & Manufacture 102 (2016) 3540

Contents lists available at ScienceDirect

International Journal of Machine Tools & Manufacture

journal homepage: www.elsevier.com/locate/ijmactool

Correlation between preload and no-load drag torque of ball screws

Chang-Guang Zhou a, Hu-Tian Feng a, Zeng-Tao Chen b, Yi Ou a,n
a
b

Department of Mechanical Engineering. Nanjing University of Science and Technology, Nanjing 210094, China
Department of Mechanical Engineering, University of Alberta, Edmonton, Alberta, Canada T6G 2G8

art ic l e i nf o

a b s t r a c t

Article history:
Received 7 September 2015
Received in revised form
24 November 2015
Accepted 25 November 2015
Available online 27 November 2015

This paper examines the relationship between the preload and no-load drag torque of ball screws. Based
on a new equation to calculate the normal contact load between balls and the screw raceway under no
external load, a new correlation between preload and no-load drag torque is proposed. Meanwhile, a
novel preload-adjustable ball screw mechanism and a drag torque measuring system are constructed.
Three LCM300 load cells and a force sensor are implemented for obtaining the experimental data from
the constructed systems. Preload and no-load drag torque are obtained and analyzed. It is found that
there is a large gap between the calculated preload by traditional formula and the measured value. The
experimental results, agreeing well with the theoretical values calculated in this paper, show there exists
a linear correlation between preload and drag torque in a preloaded ball screw mechanism. Furthermore,
the derivation of the drag torque would be decreased when increasing the preload, which can improve
the stability of the screw during its operation. This study provides an accurate correlation to obtain the
preload through the no-load drag torque for preloaded ball screws, which is signicant for better performance of ball screws as well as the CNC machine tools.
& 2015 Elsevier Ltd. All rights reserved.

Keywords:
Preload
Ball screws
Drag torque
Contact mechanics

1. Introduction
Ball screws, which can transform force and motion, are the
machine component most frequently used for computer numerical
control (CNC) machining equipment [1]. Two most important
features of the mechanism are its high positioning accuracy and
heavy load carrying capacity. In recent years, the growing demand
for high precision, high speed and high durability in CNC machine
tools has propelled high performance and reliability to be the most
important index for ball screws.
In order to eliminate backlash and provide the required rigidity
for dynamic processes, most applications in the eld of production
technology use preloaded ball screws of different levels [2]. For
example, oversized balls are often used in the single-nut ball
screw (4-point-contact) to provide preload for light to medium
loading applications and a pin is inserted between the double-nut
ball screw (2-point-contact) to give proper preload for medium
and higher loading applications [35]. Increasing the pretension
will raise the achievable dynamics as well as the maximal allowable thrust load on the feed drive. However, the service life of ball
screws would be decreased due to excessive wear and heat generation if a superabundant preload was applied. An appropriate
n

Corresponding author.
E-mail addresses: zhoucg1225@163.com (C.-G. Zhou),
ouyi3281289@163.com (Y. Ou).
http://dx.doi.org/10.1016/j.ijmachtools.2015.11.010
0890-6955/& 2015 Elsevier Ltd. All rights reserved.

and constant preload is the basic requirement for a ball screw to

retain its reliability in position accuracy and dynamic characteristic, which is signicant for the performance of CNC machine
tools. Therefore, it is important for researchers and engineers to
know the preloading value of ball screws precisely.
So far, a direct identication of the preload is generally associated with high complexity and expenses, and therefore, is not
used in practice [6]. Usually the preload level of a ball screw resulting from assembly is determined by measuring the drag torque
at 100 rpm when no external loads are applied [7]. In the most
commonly used formula of drag torque and preload of ball screws,
for example, according to the product technology manuals of NSK
[8], only the helix angle is taken into consideration, other parameters, such as the contact angle or the coefcient of friction are
not considered.
Verl and Frey studied the correlation between feed velocity and
effective preload in ball screw drives through a preload sensing
mechanism, while the preload level is un-adjustable for a certain
ball screw [6]. With a preload-adjustable ball screw feed drive
system, Feng and Pan examined the relationship between ball
screw preload variation and vibration signals [9]. However, the
drag torque of the mechanism presented in their study cannot be
diagnosed. Xu, et al. develop a new systematic creep analysis
model to calculate the friction of ball screws. Their study can be
used to calculate the mechanical efciency of ball screws (under
axial load) other than the drag torque, and the experimental verication is totally in doubt [10].

36

C.-G. Zhou et al. / International Journal of Machine Tools & Manufacture 102 (2016) 3540

Fig. 1. Normal force, sliding velocity and friction force formed at the ball-nut contacts.

There is a lack of literature discussing the correlation between

preload and drag torque of ball screws. And the existing formula is
not applicable to the cases of both the double-nut preloaded type
and the single-nut preloaded type with an offset pitch. Thus, a
new equation to calculate the normal contact load between balls
and the screw raceway under no external load was derived, and
then a new accurate correlation between no-load drag torque and
preload was built. With a novel preload-adjustable ball screw
mechanism and a drag torque measuring system constructed in
this research, the preload and drag torque at different preloading
levels and positions of the screw were measured. The experimental results verify the correlation derived in this paper.

2. Preload and no-load drag torque of ball screws

According to DIN ISO 34083 [7], the test instructions of preloaded ball screws are as follows
1. System preloaded (with or without end seals).
2. For the recording of the radial preload force, couple the ball nut
body to a load cell at the distance l (mm) from the axis of
rotation.
3. Take recordings of the force indicator at a rotational speed
100 rpm in both directions of rotation.
4. For lubrication, use a lubricant of ISO viscosity grade 100.
It can be seen from the test instructions that the rotational
speed is very low (100 rpm). According to Ref. [11], the contact
angle between the ball and the screw or the nut can be regarded as
45 degree when the rotational speed is lower than 1000 rpm.
2.1. Problem of current formula
According to Ref. [8], the relationship between preload and
drag torque in ball screws can be expressed as follows:

Tp =

0. 05 Fp Ph

tan 2

(1)

where , Tp, Fp and Ph are the lead angle, drag torque, preload and
lead of the ball screws, respectively.
Eq. (1) is based on the energy conservation principle, which
means the work done by the drag torque ( Tp 2 ) is equal to that
done by the preload ( Fp Ph ) with a coefcient of 0.05 . However,
tan

the preload is an internal force and does not do any work. And for
a certain applied preload, the contact angle would directly determine the normal force between the ball and the raceway, and
therefore, inuence the friction and drag torque, which is not reected in Eq. (1). Besides, for a ball screw mechanism of the
double-nut preloaded type and the single-nut preloaded type with
an offset pitch, the drag torque is applied by two nuts or two sides
of balls, which is not applicable for Eq. (1). This would lead to a
large gap between the calculated preload by Eq. (1) and the actual
value under the above conditions.
2.2. New correlation between preload and no-load drag torque
It has to be pointed out that, according to DIN ISO 34083 [7],
the ball nut body is coupled to a load cell during the measurement
of the drag torque. Thus, the obtained drag torque is just the
moment applied by the friction force on the ball-nut contact point
at a distance 1 (mm) from the screw axis, independent of the
moment between the balls or the ball and the screw. The friction
model in Ref. [10] includes those above three kinds of moment,
while the experiment verication is the drag torque test of the
preloaded ball screw under no load, which is the drag torque between the ball and the ball nut track. Thus the experiment verication is obviously not right. Other studies on the preload of ball
screws are based on the drag torque through Eq. (1) [6]. Thus,
theres no right formula of the correlation between no-load drag
torque and preload of ball screws in literature. Based on the kinematics and dynamics analysis between the ball and the nut, a
completely new model was built. As shown in Fig. 1, three coordinate systems are established rst: the global coordinate system, FrenetSerret coordinate system, and contact coordinate
system. The global coordinate system, (X,Y,Z), is xed in space with
its Z-axis coincident with the axis of the screw. The FrenetSerret
coordinate system [12], (t,n,b), is xed at the ball center (O) whose

C.-G. Zhou et al. / International Journal of Machine Tools & Manufacture 102 (2016) 3540

moving trajectory is in line with the t-axis direction. The contact

coordinate system, (x, y,z), is xed at the contact points
(A) between the ball and nut raceway with the z-axis pointing to
the ball center and y-axis parallel to t-axis. The x-axis is dened to
be tangential to the balls circle formed on the bn plane. Point A
represents the contact point between the ball and nut raceway,
and point B represents the contact point between the ball and
screw raceway.
In Fig. 1, A and B are the projections of point A and point B in
b-axis; i and o represent the contact angles of the ball with
respect to the screw and nut, respectively; the angle between baxis and Z-axis is , which has the same value of helix angle; w is
the rotational speed of the screw, wm is the angular velocity of the
ball revolution along the screw raceway; wt , wn and wb are the
three absolute velocity components in the t-, n- and b-directions
due to balls spinning with an angular velocity of wR ; VSA is the
slipping velocity between the ball and nut raceway; Q and Ff are
the normal contact force and the friction force acting on the ball,
respectively; Ffn is the friction force acting on the nut, which is
equal and opposite to Ff . VSA , Ff and Ffn are parallel to each other;
the components of VSA along the x- and y-axis are Vy and Vx , respectively; and the angle between Vx and VSA is dened as , which
can be written as

Vy
= tan1
Vx

(2)

The relationship between Vy and Vx is given as [13]

Vy
2r + rb (b cos o n sin o )
= m m
Vx
r b t

(3)

For the rotational speed of the screw of only 100 rpm, can be
regarded as . For the ball screw used in the experiment, rm is

Q=

Fa
it Nc Zt sin o cos

37

(8)

where Fa is the axial load

Eq. (8) is the only formula currently used to calculate normal
contact load (Q ). However, it is found that Eq. (8) is just applicable
to the case under axial loads, not the case under no load. When
there exists an axial load ( Fa ), it is the servo motor that drives the
screw to rotate and produces the normal contact load between
screw-ball and ball-nut contacts, ultimately produces an axial
force to balance the axial load. Thus, the axial load (axial force) is
the component of the normal contact load. For a preloaded ball
screw under no load, the normal contact load is always there as a
result of the existing preload ( Fp ), whether the screw rotates or
not. Thus the normal contact load is the component of the preload,
which can be written as

Q=

Fp cos sin o
it Nc Zt

(9)

Eq. (9) is the formula to calculate the normal contact load for
the preloaded ball screw under no load, which has never been
reported in literature. Combined Eqs. (6), (7) and (9), the drag
torque of the preloaded ball screw can be rewritten as

Mf = Fp sin o (rm+rb cos o ) (cos )2

(10)

It has to be mentioned that the drag torque calculated by Eq.

(10) is just the torque produced by one nut, thus the total drag
torque (which is also the tested drag torque) of the preloaded ball
screw was

Mtest = 2Mf

(11)

Accordingly, the preload of the preloaded ball screw can be

written as

20 mm, rb is 2.9765 mm, m is 4.685 rad/s, b is -22.76 rad/s, n is

26.17 rad/s, t is 0.831 rad/s and is 89.3 degree. Therefore, VSA ,
Ff and Ffn can be regarded as parallel to t-axis.
The distance AZ from point A to Z-axis can be obtained as

AZ =

(rm + rb cos o o cos o i cos i )2

2
+ (rb o ) sin o tan

(4)

(5)

Thus, the drag torque of the preloaded ball screw can be

written as

Mf = it Nc Zt Ff (rm+rb cos o ) cos

(6)

where it and Nc are the numbers of the leads and the circles, respectively, Zt is the effective numbers of the balls in one circle. The
friction force, Ff , between the ball and the nut raceway can be
obtained as

Ff = Q

Mtest
2 sin o (rm + rb cos o )(cos )2

(12)

3. Functional principle of the preload-adjustable ball screw

where rm is the screw radius, rb is the ball radius, o is the deformation between the ball and the nut raceway, i is the deformation between the ball and the screw raceway. Under a preload of 2 KN, o and i both are about 0.4 mm, which can be
ignored compared to rm and rb . And [(rbo ) sin o tan ]2 can
also be ignored for tan only being 0.08. Therefore, Eq. (4) can be
simplied as

AZ = rm+rb cos o

Fp =

(7)

where is the COF in a ball screw mechanism which can be obtained through the product manual or the manufacturer.
The equation to calculate the normal contact load between a
ball and the raceway is written as [14,15]

A preload-adjustable ball screw mechanism is constructed to

examine the relationship between the preload levels and the drag
torque, which is shown in Fig. 2. The proposed mechanism consists of the following components: a screw shaft, two ball screw
nuts, two grooved drum round nuts, two disc springs, three tension and compression load cells ( FUTEK MODEL LCM300) and a
sleeve. Table 1 shows the specications of the ball screw.
The mechanism of the modied double-nut structure is built by
regulating the force between the master nut and the slave nut.
This is executed by rotating two grooved drum round nuts on the
specially designed slave nut with external threads. Two disk
springs are installed between the grooved drum round nuts and a
spacer. Three tension and compression load cells are installed
between the spacer and a sleeve. Once the grooved drum round
nuts are rotated to compress the disk springs, the exerting force
acts on the spacer, the load cells, the sleeve and the ange of the
master nut in sequence, then enlarge the space between the
master and slave nuts, squeezing the balls inside the master and
slave nuts. Therefore, a preload is applied to the ball screw. We can
then adjust the preload by compressing the disk springs. It has be
to pointed out that: The normal value of the ball diameter is
5.953 mm. However, when constructing the preload adjustable
mechanism, the diameter of the balls inside the master and slave
nuts is selected as 5.949 mm. This ensures the initial applied
preload is zero. The key ensures that the master and slave nut

38

C.-G. Zhou et al. / International Journal of Machine Tools & Manufacture 102 (2016) 3540

move in the axis direction of the screw without rotating. The

above two points guarantee that the force obtained by the load
cells is just the preload between the two nuts. According to the
output signal of the tension and compression load cells, the applied preload can be obtained in real time through the preload
acquisition system in Fig. 3. The dynamic load rating of the ball
screw used in this study is 35 KN, which is obtained from the data
sheet. The suggested normal preload of a ball screw is typically not
greater than 10% of its dynamic loading capability. Thus, the
maximum preload setting used in this study is 3.5 KN.

4. Experimental setup

Fig. 2. Preload-adjustable ball screw mechanism.1-Grooved Drum Round Nut (2),

2-Spacer, 3- Tension and Compression Load Cell (3) 4- Sleeve, 5- Disk Spring (2),
6-Slave Nut, 7-Key, 8-Master Nut.
Table 1
Specication of the preload-adjustable ball screw (without end seals).
Parameters

Value

Unit

Radius of the ball rb

Radius of the raceway rs
Nominal radius of ball screw rm
Helix angle
Contact angle i
Helical pitch Ph
Circles number  columns number i
Rotate speed w
Lubrication mode
Viscosity grade of the oil
Environment temperature

2.9745
3.304
20
4.55
45
10
61
100
oil
100
20 1

mm
mm
mm
degree
degree
mm
rpm

Celsius

In the experiment, the preload-adjustable ball screw mechanism is tested on the drag torque test bench. One end of the ball
screw is connected with the servo motor and the other end is xed
at the tailstock center. The circumferential movement of the nut is
constrained by the measurement device and connecting rod. As
Fig. 3 shows, a specially designed vertical adjustable support unit
on the work table is in contact with the screw shaft track during
the measurement. So the work table would feed in both directions
of the screw shaft feed axis as the ball screw nut does when the
shaft rotates. With the work table and the ball screw nuts feeding
at the same speed, the friction of the circumferential movement
measured by the pressure sensor multiplied by the arm of force is
the drag torque of the ball screw. And at the same time, the preload is recorded by the preload acquisition system (Force SensorsData Acquisition System-PC).
Before the start of the experiment, the ball screw ran for 5 min
to make it fully lubricated. Then, the drag torque was measured at
100 rpm with no axial load, in which total four preloading levels
(0, light, middle and heavy) were set with each level being repeated three times. When there is no preload applied on the nuts,
the measured drag torque turns out to be about zero, which means
the initial applied preload is zero and veries point in Section 3.
In the experiment, the length of force arm was set to be 52.5 mm,
the sampling frequency of the two kinds of force sensor (drag
torque and preload) were 12 and 1000 Hz. The experimental values of drag torque and preload are shown in Fig. 4. For the

Fig. 3. Acquisition system of the drag torque and preload.1- Tension and Compression Load Cells, 2-Rot, 3-Force Sensor, 4-Support Unit.

C.-G. Zhou et al. / International Journal of Machine Tools & Manufacture 102 (2016) 3540

39

Fig. 4. No-load drag torque and preload at three different preloading levels.
Fig. 5. Predicted preload versus calculated preload by Eq. (1) under light preload.

experimental results of the forward stroke are quite similar as

those of the backward stroke, only the results over the forward
stroke are presented here. It is evident that Fig. 4 exhibits a direct
correlation between the preload and no-load drag torque.

5. Results and discussion

It is worth mentioning that, for the geometry error of the ball
screw raceway, the preload can never be kept constant at a certain
preloading level. Thus the preloading level means the average
value of the preload over the whole stroke of the ball screw. And
thats the reason why the preload in Fig. 4 varies at different positions of the screw even under one preloading level. According to
Fig. 4, the no-load drag torque is proportional to the preloading
levels at a certain position, and the variation trend of the drag
torque is consistent with that of the preload at a certain preloading
level, which indicates that there is a linear correlation between
preload and drag torque in a preloaded ball screw mechanism. In
addition, due to the impact of the starting moment, the no-load
drag torque has a large uctuation on the initial stage, as shown in
Fig. 4.
To have a better understanding of the inuence of the preloading level on the no-load drag toques, the derivation of the noload drag torque under different preloading levels is shown in
Table 2. It can be seen in Table 2 that the derivation of the no-load
drag torque, T /Tmin , is decreasing with the increasing of the
preloading levels. This is because the contact stiffness of the ball
screw increases with the increasing of the preloading levels. Thus,
for a given ball screw, the derivation of the drag torque would be
decreased when increasing the preload, which can improve the
stability of the screw in the working process.
Fig. 5, 6 and 7 show the preload predicted by the new model,
i.e., Eq. (12) and the preload calculated by Eq. (1) at different
preloading levels (light, middle and heavy). It can be seen obviously that the preload predicted by the new model and the

Fig. 6. Predicted preload versus calculated preload by Eq. (1) under middle preload.

Table 2
Derivation of the no-load drag torque under different preloading levels.
Preloading levels

Tmin (Nm)

Tmax (Nm)

T (Nm)

T
%
Tmin

light
middle
heavy

0.63
1.48
2.49

0.87
1.80
2.87

0.24
0.32
0.38

38.1
21.6
15.3

Fig. 7. Predicted preload versus calculated preload by Eq. (1) under heavy preload.

actual value (tested preload) are very close, while the preload
calculated by Eq. (1) is much larger. The mean values of the drag
torque and preload under the three preloading levels tested and
the corresponding preloads predicted by the new model and Eq.

40

C.-G. Zhou et al. / International Journal of Machine Tools & Manufacture 102 (2016) 3540

Table 3
Predicted preload by equation (12) and calculated preload by equation (1).
Drag torque Preload (N) Predicted preload by
(Nm)
Eq. (12)

0.74
1.62
2.65

1048
2077
3237

Calculated preload by
Eq. (1)

Value
(N)

Relative
error

Value
(N)

Relative
error

953
2086
3413

9.1%
0.4%
5.4%

2623
5743
9394

150.3%
177.0%
190.2%

(1) are shown in Table 3. The relative error of the preload predicted by the new model is within 10% while the error of Eq. (1) is
more than 150%, which veries the theoretical analysis in this
study.

6. Conclusions
In the most commonly used formula of drag torque and preload
of ball screws, as expressed by Eq. (1), only the helix angle is taken
into consideration, other parameters, such as the contact angle or
the coefcient of friction are not considered. Thus, considering the
helix angle, contact angle, friction coefcient, ball diameter, sliding
direction and normal contact force of the ball-nut contact, a new
correlation between preload and no-load drag torque is built. The
experimental results show that, there is a linear correlation between preload and drag torque in a preloaded ball screw mechanism, and the relative error of the preload predicted by the
new model is within 10% while the error of Eq. (1) is more than
150%, which veries the present theoretical analysis. According to
the experimental results, the derivation of the drag torque would
be decreased when increasing the preload for a given ball screw.
This is because higher preload leads to higher contact stiffness,
thus the achievable dynamics of the ball screw mechanism is
raised. The outcome is signicant for better performance of ball
screws as well as the CNC machine tools.

Acknowledgment
The authors gratefully acknowledge the Key Laboratory of
Performance Test and Reliability Technology for CNC Machine Tool

Components of Chinese Machinery Industry for providing the test

bench and experiment materials. The insightful comments of the
anonymous reviewers are highly appreciated. The present study
has been supported by the National Science and Technology Major
Projects of China (2012ZX04002021).

Appendix A. Supplementary material

Supplementary material associated with this article can be
found in the online version at http://dx.doi.org/10.1016/j.ijmach
tools.2015.11.010.

References
[1] H. Nurhadi, Y.-S. Tarng, Experimental approached optimisation of a linear
motion performance with grey hazy set and Taguchi analysis methods (GHST)
for ball-screw table type, Int. J. Adv. Manuf. Technol. 44 (12) (2009) 149160.
[2] A. Verl, S. Frey, T. Heinze., Double nut ball screw with improved operating
characteristics, CIRP Ann. -Manuf. Technol. 63 (1) (2014) 361364.
[3] J.S. Chen, I.C. Dwang, A ballscrew drive mechanism with piezo-electric nut for
preload and motion control, Int. J. Mach. Tools Manuf. 40 (4) (2000) 513526.
[4] J. Otsuka, Nanometer-/subnanometer-positioning technology using feeding
mechanism., Int. J. Autom. Technol. 2 (1) (2008) 411.
[5] Kripa K. Varanasi, A. Nayfeh Samir, The dynamics of lead-screw drives: loworder modeling and experiments., Transactions -Am. Soc. Mech. Eng. J. Dyn.
Syst. Meas. Control. 126 (2) (2004) 388396.
[6] A. Verl, S. Frey, Correlation between feed velocity and preload in ball screw
drives., CIRP Ann. -Manuf. Technol. 59 (1) (2010) 429432.
[7] N. N., DIN ISO 3408-5 Acceptance conditions and acceptance tests, 2008.
[8] SKN, Precision machine components catalogue, NSK, Tokyo, 2003.
[9] Guo-Hua Feng, Yi-Lu Pan, Investigation of ball screw preload variation based
on dynamic modeling of a preload adjustable feed-drive system and spectrum
analysis of ball-nuts sensed vibration signals, Int. J. Mach. Tools Manuf. 52 (1)
(2012) 8596.
[10] Nannan Xu, et al., Modeling analysis and experimental study for the friction of
a ball screw, Mech. Mach. Theory 87 (2015) 5769.
[11] Chin Chung Wei, et al., Kinematic Analysis of the Ball Screw Mechanism
Considering Variable Contact Angles and Elastic Deformations, J. Mech. Design125 4 (2004) 717733.
[12] L.I.N. MING-CHING, Design and mechanics of the ball screw mechanism [D],
Mechanical Engineering of University of Wisconsin-Madison, USA (1989), p.
2326.
[13] M.C. Lin, S.A.V. Ravani, Design of the Ball Screw Mechanism for Optimal Efciency., J. Mech. Des. 116 (3) (1994) 856861.
[14] S. Shimizu, H. Shimoda, C.S. Sharma, Re-Evaluation of Basic Dynamic Load
Rating and Life Formula for a Ball Screw, Tribol. Trans. 50 (1) (2007) 8895.
[15] Xuesong Mei, et al., Study on the load distribution of ball screws with errors.,
Mech. Mach. Theory 38 (11) (2003) 12571269.

ID
784300

Title
Correlation between preload and no-load drag torque of ball screws

http://fulltext.study/journal/748

http://FullText.Study

Pages
6