Research Article

Advances in Mechanical Engineering

2017, Vol. 9(7) 1–11
Ó The Author(s) 2017
An improved dynamic model of DOI: 10.1177/1687814017710580
journals.sagepub.com/home/ade
preloaded ball screw drives considering
torque transmission and its application
to frequency analysis

Fuhua Li1, Yao Jiang2, Tiemin Li1,3,4 and Yunsong Du1

Abstract
A dynamic model of the ball screw drive is proposed in this article. It is revealed that when axial thrust is transmitted
between the ball screw and the nut, extra torque is generated synchronously which is not proposed in existing
dynamic models. And a physical model for studying the relationship between the thrust and the torque is proposed.
A lumped dynamic model is established, and a kinematic compatibility equation describing motion transmission
between rotary displacement and axial displacement is established. Then a preload model of a double-nut for analyz-
ing the force and the deformation is built. An approach to analyze the first resonant frequency of the proposed
model is described. Meanwhile, a tested bench with a novel preload-adjustable double-nut and two novel loading
mechanisms is constructed. The axial stiffness of the supporting bearings and the preloaded double-nut is tested
based on a measurement system. Then vibration tests are carried out to measure the first resonant frequency of the
ball screw drive. Finally, contrastive analysis between experimental results and simulated results of three models is
conducted. The results show that the proposed model agrees much better with the experimental results than the
discrete model and the hybrid model do.

Keywords
Ball screw, preload, frequency, modeling, double-nut

Date received: 18 September 2016; accepted: 26 April 2017

Academic Editor: Crinela Pislaru

Introduction nuts and it has general significance to study the pre-

loaded ball screw drives.
The demand for high productivity rates and high-
precision parts requires faster and more accurate feed
drives for machine tools.1–4 Because of high stiffness, 1
Institute of Manufacturing Engineering, Department of Mechanical
reliable operation, ability to mitigate the impact of Engineering, Tsinghua University, Beijing, China
2
inertial changes, and low cost, the ball screw is the most Institute of Instrument Science and Technology, Department of
Precision Instrument, Tsinghua University, Beijing, China
commonly used component in feed drives to convert 3
Beijing Key Laboratory of Precision and Ultra-Precision Manufacturing
rotary motion to linear motion.5–7 One of the impor- Equipments and Control, Tsinghua University, Beijing, China
4
tant characteristics of a ball screw drive is that its per- State Key Laboratory of Tribology, Tsinghua University, Beijing, China
formances can be altered by exerting different preloads,
Corresponding author:
such as the stiffness of the ball screw which can be Tiemin Li, Institute of Manufacturing Engineering, Department of
improved in this manner.1,8,9 In current commercial Mechanical Engineering, Tsinghua University, Beijing 100084, China.
products, preloads are almost indispensable in the ball Email: [email protected]

Creative Commons CC-BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License
(http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/
open-access-at-sage).
2 Advances in Mechanical Engineering

For the ball screw drives, some performances are kinematic compatibility equation describing the motion
always used as important indexes which are always con- transmission between rotary displacement and axial
cerned by engineers and researchers, such as resonant displacement is constructed. Then a preload model of a
frequencies, tracking errors, and positioning accuracy. double-nut is built to analyze the force and the defor-
In order to represent the characteristics as detailed as mation. An approach to analyze the first resonant fre-
possible, dynamic models have been proposed in some quency of the proposed model is described. In order to
researches. Feng and Pan proposed a lumped dynamic conduct experimental tests, a tested bench with a
model to study the resonant frequencies and preload preload-adjustable double-nut mechanism and two
variation of the ball screw drive. The variation of novel loading mechanisms is constructed. The axial
the preload can be diagnosed by spectrum analysis of stiffness of the supporting bearings and the preloaded
the processed experimental data and the results of the double-nut is tested. The vibration tests are carried out
mathematical dynamic model.10 Chen et al. developed to measure the first resonant frequency of the ball
a dynamic model of the high-speed ball screw drive sys- screw drive, and the relationship between preloads and
tem. The joint of the ball screw and the nut is equiva- the first resonant frequency is obtained. Finally, con-
lent to a constant linear spring in the model, and the trastive analysis between the experimental results and
system was formulated with Lagrange’s method.11 the simulated results of three models is conducted. The
Amain Kamalzadeh et al. proposed an elastic deforma- results show that the proposed model agrees much bet-
tion model of a ball screw drive and presented a new ter with the experimental results than the discrete model
strategy for mitigating the detrimental effect of elastic and hybrid model do. Therefore, this study provides an
deformation using the model to offset the position com- improved dynamic model of preloaded ball screw
mand. In the model, the equivalent axial stiffness was drives, and more accurate first resonant frequency can
obtained by cascading the stiffness of bearing, nut, and be obtained by the proposed model, which is significant
ball screw.3 S Frey et al. compared the hybrid model for better performance of ball screw drives.
with the discrete model in predicting the relevant eigen-
frequencies of the feed system and validated the results
through experiments. The stiffness of the ball nut is Force transmission analysis and dynamic
expressed as a linear spring in the two models.12 Jiang model of the ball screw drive
and Zhu13 analyzed the axial stiffness of the nut and A common ball screw drive is composed of a servo
supporting bearings using the Hertz contact theory of motor, a coupling, a ball screw, ball nuts, supporting
elastic mechanics and established the axial stiffness of bearings, sliders, linear guideways, a working table,
the ball screw drive with three springs connected in and other parts.14,15 A schematic of a ball screw drive
series. In the above researches, the stiffness characteris- is illustrated in Figure 1. The force transmission analy-
tics of the ball nut are always equivalent to a linear sis between the ball screw and the nut is analyzed in
spring, therefore, the axial force is the only interactive section ‘‘Force transmission analysis.’’ The dynamic
force between the ball screw and nut. This study puts model of the ball screw drive is presented in section
forward that the linear spring model can rarely reveal ‘‘Dynamic model.’’
the force transmission between the ball screw and the
nut accurately, and the calculation of the motion of the
working table is not precise enough. An improved Force transmission analysis
dynamic model is proposed in this study. The interactive force between the ball screw and the
The resonant frequency is an important performance nut is transmitted by rolling balls. For a two-point con-
index for ball screw drives, and the system bandwidth tact ball screw, the force analysis of a single ball is illu-
and the anti-jamming property can be improved by strated in Figure 2(a).16 The contact force for a ball is
increasing it.6 In order to analyze the resonant fre- P and can be decomposed into three component force
quency, dynamic models are commonly used. In this (Figure 2(b)), axial force Fa, tangential force Ft, and
article, the proposed model is applied on the analysis of
the first resonant frequency. Because of the nonlinear-
ity of the model, another calculation method is adopted
which is different from the one that the discrete model
adopts.
This article reveals that when axial thrust is trans-
mitted between the ball screw and the nut, extra torque
is generated synchronously. And a physical model for
studying the relationship between the thrust and
the torque is proposed. A lumped dynamic model
of the ball screw drive is established, and a Figure 1. A ball screw drive.
Li et al. 3

Figure 2. (a) Force analysis of a rolling ball, (b) decomposition of contact force, and (c) force projection.

Figure 4. Dynamic model of a ball screw drive.

Figure 3. Force analysis for all the balls. where z and R represent the number of the rolling balls
and the nominal radius of the ball screw, respectively.
radial force Fn (Figure 2(c)). The physical equations are On the basis of equation (2), the mathematic relation-
expressed as ship between the thrust and the torque can be repre-
sented by
Fa = P sin a cos f ð1aÞ
Tn = FR tan f ð3Þ
Ft = P sin a sin f ð1bÞ
Fv = P cos a ð1cÞ
Dynamic model
where a represents the contact angle between rolling
To analyze dynamic characteristics of ball screw drives
balls and grooves, and f represents the helix angle of
with different preloads, a lumped parameter system is
ball screws.
modeled in Figure 4. Four degrees of freedom (DOF)
Assuming that all the rolling balls are the same and
are analyzed for the system, including the rotary DOF
there is no geometry error for the balls and the grooves,
of the motor and the ball screw, the axially translational
the force analysis for all the balls is shown in Figure 3.
DOF of the ball screw and the working table. The inter-
The axial thrust F is constituted by the axial compo-
active forces of the contact areas between the ball screw
nent forces of all the balls, and the torque Tn is consti-
and the nut are equivalent to the thrust F and the tor-
tuted by the tangential component forces of all the
que Tn presented in section ‘‘Force transmission analy-
balls. Therefore, torque is generated synchronously
sis.’’ Jm and Jb represent the inertia of the motor and
when the balls transmit thrust. The working table is
the ball screw. Mb and Mt represent the mass of the ball
driven by the axial thrust, and the torque applied on
screw and the working table. kg is the torsional stiffness
the working table is balanced by the guideways. The
of the ball screw and the coupling. ke is the equivalent
thrust F and the torque Tn are expressed as follows
axial stiffness of the ball screw and the supporting bear-
F = zP sin a cos f ð2aÞ ings. um and ub represent the rotary displacement of the
motor and the ball screw. Xb and Xt represent the axial
Tn = zPR sin a sin f ð2bÞ displacement of the ball screw and the working table.
4 Advances in Mechanical Engineering

Qm and Qb are the rotary damping coefficients of the

motor and the ball screw; Bb and Bt represent the axial
damping coefficients of the ball screw and the working
table.
The dynamic equation of the ball screw drive can be
described as
2 32 € 3
Jm 0 0 0 um
6 0 Jb 0 0 76 €ub 7
7 6
6 7
6 76 7
4 0 0 Mb 0 54 X€ b 5
Figure 5. Motion decomposition of the working table.
0 0 0 Mt X€ t
2 32 _ 3
Qm 0 0 0 um commonly applied in ball screw drives and are analyzed
6 0 Q 0 0 76 u_ b 7
7 6 in this study. In order to study the relationship between
6 b 7 ð4Þ
+6 76 7 the axial thrust F and the axial contact deformation da,
4 0 0 Bb 0 54 X_ b 5
Hertz theory is adopted, which is introduced in section
0 0 0 Bt X_ t ‘‘Relationship between axial contact force and defor-
2 32 3 2 3
kg kg 0 0 um T mation.’’ In the case of preloading, the relationship
6 k 0 07 6 7 6 7
6 g kg 76 ub 7 6 Tn 7 between the axial thrust and the axial contact deforma-
+6 76 7 = 6 7 tion is analyzed in section ‘‘Axial deformation of
4 0 0 ke 0 54 Xb 5 4 F 5
double-nuts with preloads.’’
0 0 0 0 Xt F

Relationship between axial contact force and

Kinematic compatibility equation of the working table
deformation
The relationship between forces and components
motion is revealed by the dynamic model. Moreover, it Hertz contact theory is a classic theory to analyze nom-
is also needed to analyze kinematic transmission among inal contact stress and contact deformation of two elas-
the components of the ball screw drive. In the model, tic bodies, which satisfies the assumptions in Wang
elastic deformation occurring in the contact areas et al.18 and Zhu et al.19 The compression deformation
between the ball screw and the nut is considered, and of the two bodies resulting from the contact force is
torsional deformation of the ball screw is ignored. A represented as
compatibility equation describing relationship between (   2 )1=3 X 
axial displacement and angular displacement of compo- 2K 1 3 1  u21 1  u22 1=3
d=  r P2=3
nents is suggested. It is proposed that the displacement pma 8 2 E1 E2
of the working table is composed of three parts in the
equation: axial displacement Xt1 driven by rotary ð6Þ
motion of the ball screw, axial vibration Xt2 generated where K and ma are Hertz coefficients; u1 and u2 are
by axial vibration of the ball screw, and axial deforma- Poisson’s ratios of the two contact bodies; E1 and E2
tion Xt3 caused by interactive force of the contact areas are elastic moduli; Sr is the synthetic curvature at the
between the ball screw and the nut. Motion decomposi- contact point. Sr = r11 + r12 + r21 + r22, where r11,
tion of the working table is presented in Figure 5. The r12, r21, r22 are principal curvatures of the contact point
kinematic compatibility equation of the working table and can be obtained based on formulas in Wang et al.18
can be expressed as follows and Zhu et al.19
The Hertz contact theory describes the relationship
Xt = Xt1 + Xt2 + Xt3 ð5Þ
between normal force and normal deformation along
where Xt1 = uub/(2p), Xt2 = Xb, Xt3 = 2da; u is the the force direction. By means of geometric projection,
lead of the ball screw; da is the axial deformation. the axial contact deformation da of the rolling ball can
be expressed as follows

Analysis of axial contact force and d1 + d2

da = ð7Þ
deformation sin a cos f

There are several approaches to exert preloads, such as where d1 is normal contact deformation between the
using oversized balls, creating offset between leads, and ball and the ball screw, and d2 is normal contact defor-
adjusting spacers in a double-nut.2,17 Double-nuts are mation between the ball and the nut.
Li et al. 5

Figure 6. Force analysis of contact areas between the ball screw and the nut in dynamic state: (a) F2 < Fp and (b) F2 . Fp.

The equation between axial thrust and axial contact and the contact areas between the nuts and rolling balls.
deformation can be established as follows The axial deformation of the nuts A and B is the same
in value but opposite in direction. The axial deforma-
da = ðKe1 + Ke2 ÞFb 2=3 ð8Þ tion dax can be calculated as follows
where Fb is the axial thrust of a ball. The contact coeffi-  2=3
Fp
cient Kei can be expressed as follows dax = ðKe1 + Ke2 Þ ð11Þ
z
(  2 )1=3
Ki 3 1  u21 1  u22
Kei = + The working table is in a dynamic state. When the working
pmai 2 E1 E2
ð9Þ table moves, the axial thrust F is generated in the con-
1  X 1=3
r i = 1, 2 tact areas of the ball screw and the nut. The state in
5=3
ðsin a cos fÞ which F is smaller than Fp and the other state in which
F is bigger than Fp are analyzed respectively.
The axial stiffness of the nut with preload Fp is

1=3 Situation 1. In this situation, the preload is not less than

z2=3 FP
kn = ð10Þ axial thrust F. The directions of axial contact force for
Ke1 + Ke2
nuts A and B are opposite, as shown in Figure 6(a).
The contact angle a is relevant with contact force. The axial contact deformation caused by the axial
When the contact force is increased, the contact angle is thrust F for each nut is db. The increase of the axial
increased, which results in decrease of the contact coef- contact force for nut A is F1 relative to Fp, and the
ficient Ke. However, according to Huang and Ravani,20 decrease of the axial contact force for nut B is F2 rela-
the contact angle variation result from axial load is rela- tive to Fp. The equations used to calculate the rela-
tively small. Therefore, the influence of contact force tionship between axial force and axial deformation
variation on the contact angle is ignored in the follow- are expressed as equation (12). In the equations, Ke is
ing analysis. the sum of Ke1 and Ke2
   2=3
Axial deformation of double-nuts with preloads F1 + Fp 2=3 Fp
Ke  Ke = db ð12aÞ
z z
Assuming that the master nut and the slave nut are the
 2=3  
same, the axial force is the preload Fp for each nut when Fp Fp  F2 2=3
the working table is static and there is no external force Ke  Ke = db ð12bÞ
z z
applied on the working table. And the axial force of the
two nuts is resultant force of the preload Fp and the F1 + F2 = F ð12cÞ
thrust F when the working table moves. The two states
are discussed as follows.
Situation 2. In this situation, the preload is less than
axial thrust F. Directions of the axial contact force for
The working table is in a static state. The master nut A and nuts A and B are the same, which is different from the
the slave nut B only bear preload Fp in this state. When above state, and axial contact force of nut B is opposite
exerting a preload, elastic deformation occurs in the to the axial thrust F, as shown in Figure 6(b). In the
contact areas between the ball screw and rolling balls, initial process of exerting preloads, axial contact force
6 Advances in Mechanical Engineering

of the nut B decreases to zero. Then the direction of

the force is reversed and it keeps increasing till the bal-
ance state. The force F2 can be obtained by the follow-
ing equation. Equations (12a) and (12c) are suitable in
the situation
   2=3
F1 + Fp 2=3 Fp
Ke  Ke
z z
 2=3   ð13Þ
Fp F2  Fp 2=3
= Ke + Ke
z z

Analysis of the first resonant frequency

The first resonant frequency is an important perfor-
Figure 7. The tested bench.
mance index for ball screw drives. According to equa-
tion (4), as the dynamic equations of the proposed
model are intercoupling and the independent stiffness
matrix cannot be obtained when the resonant fre- Experimental setup
quency is calculated, the method of getting the resonant A tested bench is constructed to examine the relation-
frequency by calculating the eigenvalue of the charac- ship between preloads of the double-nut and the reso-
teristic equation about the mass and stiffness matrix is nant frequency of ball screw drives, which is shown in
not applicable. To calculate the resonant frequency, Figure 7. The tested bench mainly consists of a master
another method of inputting a pulse torque to the ball screw drive with a preload-adjustable double-nut
motor and then applying the Fourier transform to the mechanism and two loading ball screw drives with load-
displacement of the working table is used in this study. ing mechanisms. The front end of the master ball screw
As for the discrete model, the double-nut is always is supported by two angular contact bearings and the
equivalent to a linear spring. The entire drive system is rear end is supported by a deep groove bearing.
modeled using a lumped parameter system, and the The preload-adjustable double-nut mechanism
dynamic equation of the modeled system can be shown in Figure 8 consists of the following compo-
described as equation (14), where kn is the stiffness of nents: two ball nuts, two disc springs, an adjustment
the double-nut, k is equivalent friction coefficient in the nut, a flange plate, a bending plate, a thrust bearing,
direction tangential to the ball screw shaft rotation, and a pressure sensor (LW25150-1 klbf; product of
and g is axial displacement of the ball screw shaft Interface Co., USA). The preload of the double-nut
caused by rotation.10 The first resonant frequency can can be adjusted by rotating the adjustment nut. When
be obtained by calculating the eigenvalue of the equa- the adjustment nut is adjusted, the distance between
tion. As for the hybrid model, it is built up in finite-ele- the master nut and the slave nut is changed and the disc
ment-methods using a combination of lumped mass, springs are compressed. Then the preload is exerted on
springs, dampers, and Timoshenko beam element, as the master nut and the slave nut. In order to restrict
shown in Frey et al.12 The relationship between pre- rotary motion of the slave nut, a flange plate and a
loads and the first resonant frequency can be acquired bending plate are used to connect the mechanism with
by changing the double-nut stiffness for dynamic the working table. In case that the preload cannot be
models. applied on the nuts, the axial translational freedom of
2 32 € 3 2 32 _ 3 the nuts is unrestricted in the process of exerting pre-
Jm 0 0 0 um Qm 0 00 um loads. After the preload is set, the flange plate and the
6 0 Jb 0 0 7 6 u€ 7 6 0 Qb 007 6 u_ 7
6 76 b 7 6 76 b 7 bending plate are fastened together by means of bolt
6 76 7 + 6 76 7
4 0 0 Mb 0 54 X€ b 5 4 0 0 Bb 0 54 X_ b 5 screws, so that the preload-adjustable double-nut
0 0 0 Mt X€ t 0 0 0 Bt X_ t mechanism forms a rigid body.
2 32 3 2 3 The loading mechanism which is driven by a ball
kg kg 0 0 um T
6 k screw is designed to apply axial force on the working
6 g kg + dkgkn 0 0 7 6u 7 607
76 b 7 6 7
+6 76 7 = 6 7 table, as shown in Figure 9. In order to apply tension
4 0 0 kn + k e kn 54 Xb 5 4 0 5 as well as pressure, two die springs are installed on both
0 0 kn kn Xt 0 sides of the loading plate. The pressure of the springs
ð14Þ can be controlled by adjusting position of the loading
plate. A load cell (UMMA-200 kg; product of
Li et al. 7

Figure 8. The preload-adjustable double-nut mechanism.

Figure 9. The loading mechanism.

DACELL Co., Korea) is installed to test the tension to test the axial displacement thereof, as shown in
and compression in real time. Two loading mechanisms Figure 11. The value of the first sensor is the axial
are adopted and distributed symmetrically. The para- deformation of the supporting bearings. The difference
meters of the constructed master ball screw drive are between the second sensor and the third sensor is defor-
listed in Table 1. mation of the double-nut.
In order to test the stiffness of the nut with preload,
the approach that measures the axial deformation of
Results and discussion the nut after applying specified axial force on the nut is
always adopted. The preload should be set at first and
Tests of the ball screw drive stiffness
then test the axial deformation under the action of axial
The stiffness of the key components has a great influ- load. Therefore, loading mechanisms and a preload
ence on the ball screw drive frequency; therefore, it is regulating mechanism are needed, which improve the
important to obtain accurate stiffness values. The stiff- experimental requirements greatly. What is more, many
ness identification experiment is conducted on the experiments with different preloads need to be con-
tested bench. In order to test the axial stiffness of the ducted to get more accurate relationship between pre-
supporting bearings and double-nut, three laser displa- loads and axial stiffness of the double-nut. The
cement sensors are used. The first one is placed beside approach that the preload can be equivalent to be the
the front end of the ball screw to test axial displace- external axial load when axial stiffness is tested is
ment of the circular part which is fixed to the ball taken. Therefore, the axial stiffness can be tested con-
screw, as shown in Figure 10. The second one is placed tinuously and the experimental efficiently can be
at the location beside the rear end of the ball screw to improved greatly.
test axial displacement of the ball screw end face. The In order to validate the approach, three experiments
third one is placed behind the working table and is used with preloads of 0, 335, and 649 N are performed,
8 Advances in Mechanical Engineering

Table 1. The parameters of the constructed master ball screw drive.

Parameters of the ball screw drive Given value Unit

Working table mass Mt 61.71 kg

Ball screw mass Mb 3.20 kg
Inertial moment of the motor Jm 4.45 3 1024 kg m2
Inertial moment of the ball screw Jb 3.81 3 1024 kg m2
Viscous damping coefficient of the guideway of the table Bt 0 N m/s
Viscous damping coefficient of the supporting bearing of the ball screw Bb 0 N m/s
Rotation viscous damping coefficient of the motor Qm 0 N m s/rad
Rotation viscous damping coefficient of the supporting bearing of the ball screw Qb 0 N m s/rad
Pitch of the ball screw u 10 mm
Nominal diameter of the ball screw d 25 mm
Diameter of the ball db 4.762 mm
Number of the working balls z 75
Pressure angle between balls and grooves a 45 °
Helix angle of the ball screw f 6.98 °
Dimensionless radius of curvature of grooves 0.52
Elastic modulus E1, E2 210 GPa
Poisson ratio u1, u2 0.26
Axial stiffness of the ball screw shaft kb 1.15 3 108 N/m
Torsional stiffness of the ball screw and the coupling kg 2083 N m/rad
Equivalent friction coefficient in the direction tangential to the ball screw shaft rotation k 0
Axial displacement of the ball screw shaft caused by rotation g 3.18 3 1023 m/rad

Figure 11. Tests of the axial displacement of the ball screw

Figure 10. Tests of the axial displacement of the ball screw rear end and working table.
front end.

RMSE for the curve representing preload of 335 N is

respectively. Axial load is applied by the two loading 0.20 mm, and for the curve representing preload of
mechanisms on the working table, and range of the 649 N is 0.25 mm relative to the curve representing pre-
axial load for the three experiments is 0–900, 0–565, load of 0 N. The difference between them is relatively
and 0–251 N, respectively. The constant moving velo- small and can be ignored. Therefore, the test approach
city of the loading plates is 0.3 mm/s. Figure 12 shows of the double-nut stiffness that the preload can be
the experimental results of the double-nut. The origins equivalent to the axial load to test the deformation of
of the curves representing preload of 335 and 649 N are the double-nut is applicative.
set to be 4.8 and 8.7 mm, and the abscissa is sum of pre- In order to test the axial stiffness of the supporting
load and axial load. bearings, axial load is applied on the working table by
The index root mean square error (RMSE) is the loading mechanisms. The range of the axial load is
adopted to evaluate the difference between the curve 0–900 N. The test results of the first laser displacement
representing preload of 0 N and other two curves. The sensor and the axial load are presented in Figure 13.
Li et al. 9

a direct influence on the degree of axial–torsional cou-

pling. As the transmission ratio of the tested bench is
10 mm/rev which is relatively low, the discrepancy of
the first mode between couple system and decoupled
system is very small.14 Therefore, the axial–torsional
coupling of the first mode is ignored, and the first mode
is considered to be pure axial. When the first resonant
frequency is calculated by the proposed model, the dis-
crete model, and the hybrid model, the stiffness of the
double-nut is acquired from the fitting curve, as shown in
Figure 14, and the region of the preload is [0 N 900 N].
In the simulation process of the proposed model, the sol-
ver ode45 is chosen to solve the differential equations
and the step size is 0.0001. The simulated results of the
Figure 12. Relation between axial deformation and axial load
two models are also shown in Figure 15.
with three preloads.

The axial stiffness of the supporting bearings is

Vibration tests
9.6 3 107 N/m based on the experimental result.
Therefore, the equivalent axial stiffness of the ball Vibration tests are carried out to measure the resonant
screw and the supporting bearings is 5.2 3 107 N/m. frequency of the ball screw drive, as shown in
The axial stiffness of the double-nut can be obtained by Figure 16. The devices used for the vibration test con-
testing axial deformation of the double-nut under the sist of an impact hammer, a signal amplifier, an acceler-
action of axial load. The relationship between the axial ometer (356A16; product of PCB Piezotronics Co.,
stiffness kn and the preload can be acquired by means USA), a digital analyzer INV 3018C, and data acquisi-
of testing the axial stiffness of the double-nut with dif- tion and signal processing (DASP) system. The sensitiv-
ferent preloads, as illustrated in Figure 14. It is obvious ity of the accelerometer is 103 mV/g, and an impulse
that the fluctuation of the stiffness curve is mostly hammer with a force sensor and white plastic tip is used
caused by the measurement. In order to get rid of the to excited the end face of the ball screw. Acceleration
effects, a fitting curve is obtained according to the signals and impact force signals are recorded by the
experimental outcomes, as shown in Figure 14. digital analyzer INV 3018C, and frequency response
functions (FRFs) are obtained with the help of the
DASP system. The first resonant frequency can be
Simulation analysis acquired according to the FRFs. Vibration tests are
According to Diego et al.,14 the first mode obtained by conducted on the ball screw drive with different pre-
equation (14) is mainly axial, but it is also coupled with loads, and the correlation between preloads and the
torsional vibration of the ball screw. What is more, the first resonant frequency of the ball screw drive is pre-
transmission ratio is an important parameter which has sented in Figure 15.

Figure 13. (a) Experimental results of three laser displacement sensors and (b) axial load applied on the working table.
10 Advances in Mechanical Engineering

increases. The curves clearly show that the proposed

model fits the experimental results better than the com-
monly used discrete model and hybrid model do. The
percentage of the difference between the discrete model
and the hybrid model is about 1%, which shows good
concordance with Frey et al.12 It is observed that at the
beginning of the curve, the errors between the experi-
mental results and the simulated results of the two
models are nonnegligible. It is because that the theore-
tical double-nut stiffness is zero when there is no pre-
load. However, actually even if no preload is exerted
by the preload-adjustable mechanism, there are still a
few preloads in the double-nut because of the machin-
ing and assembling, and they cannot be measured on
Figure 14. The relationship between preloads and the stiffness the tested bench. The average of the absolute frequency
of the double-nut. errors is adopted to evaluate the difference between the
simulated results and the experimental results. The rela-
tionship between the average of the absolute frequency
errors and the original offset of the preload is shown in
Figure 17. Simulation errors of the proposed model are
less than those of the discrete model and hybrid model
all the time. It is clear that the proposed model can get
more accurate first resonant frequency of ball screw
drives than the discrete model and hybrid model do.

Conclusion
This article has proposed a dynamic model of ball
screw drives. It is revealed that when axial thrust is
transmitted between the ball screw and the nut, extra
torque is generated synchronously and is proportional
to the thrust. Based on the dynamic model and the con-
Figure 15. First resonant frequency comparison between structed tested bench with a novel preload-adjustable
simulated results and experimental results.
double-nut mechanism and novel loading mechanisms,
the influence of preloads on the first resonant frequency
It can be seen in Figure 15 that the first resonant fre- is tested. And a comparison of the first resonant fre-
quency shows an increasing trend and the increasing quency of preloaded ball screw drives between experi-
speed slows down as the preload of the double-nut mental results and simulated results of the discrete

Figure 16. Vibration tests.

Li et al. 11

international conference on mechatronics and automation,

Harbin, China, 5–8 August 2007, pp.2194–2199. New
York: IEEE.
5. Okwudire CE and Altintas Y. Hybrid modeling of ball
screw drives with coupled axial torsional, and lateral
dynamics. J Mech Design 2009; 131: 071002-1–071002-9.
6. Erkorkmaz K and Kamalzadeh A. High bandwidth con-
trol of ball screw drives. CIRP Ann: Manuf Techn 2006;
55: 393–398.
7. Tsai PC, Cheng CC and Hwang YC. Ball screw preload
loss detection using ball pass frequency. Mech Syst Sig-
nal Pr 2014; 48: 77–91.
8. Lin CY, Hung JP and Lo TL. Effect of preload of linear
guides on dynamic characteristics of a vertical column-
spindle system. Int J Mach Tool Manu 2010; 50: 741–746.
Figure 17. Simulated frequency errors of the three models. 9. Jiang S and Mao H. Investigation of variable optimum
preload for a machine tool spindle. Int J Mach Tool
Manu 2010; 50: 19–28.
model, hybrid model, and the proposed model is con- 10. Feng G-H and Pan Y-L. Investigation of ball screw pre-
ducted. The results show that the proposed model load variation based on dynamic modeling of a preload
agrees much better with the experimental results than adjustable feed-drive system and spectrum analysis of
the discrete model and the hybrid model do. Therefore, ball-nuts sensed vibration signals. Int J Mach Tool Manu
an improved dynamic model of ball screw drive is pro- 2012; 52: 85–96.
11. Chen J-S, Huang Y-K and Cheng C-C. Mechanical
posed for researchers and engineers, and it can obtain
model and contouring analysis of high-speed ball-screw
more accurate first resonant frequency of preloaded
drive systems with compliance effect. Int J Adv Manuf
ball screw drives. Tech 2004; 24: 241–250.
12. Frey S, Dadalau A and Verl A. Expedient modeling of
Declaration of conflicting interests ball screw feed drives. Prod Eng 2012; 6: 205–211.
13. Jiang S and Zhu S. Dynamic characteristic parameters of
The author(s) declared no potential conflicts of interest with
linear guideway joint with ball screw. J Mech Eng 2010;
respect to the research, authorship, and/or publication of this
46: 92–99.
article.
14. Diego A, Vicente Togelio L, et al. Modeling and vibra-
tion mode analysis of a ball screw drive. Int J Adv Manuf
Funding Tech 2012; 58: 257–265.
15. Liu Y, Wang J, Zhao T, et al. Stiffness model for a ball
The author(s) disclosed receipt of the following financial sup-
screw drive system. J Tsinghua Univ 2011; 51: 601–606.
port for the research, authorship, and/or publication of this
16. Changhong WU. Study on axial contact stiffness of ball
article: This work is supported by the National Natural
screws. Master Thesis, Jilin University, Changchun,
Science Foundation of China (grant no. 51675292) and the
China, 2008.
National Science and Technology Major Project of China
17. Okwudire CE. Improved screw-nut interface model for
(grant no. 2015ZX04014021 and grant no. 2016ZX04004004).
high-performance ball screw drives. J Mech Design 2011;
133: 041009-1–041009-10.
References 18. Wang D, Lu Y, Zhang T, et al. Effect of stiffness of roll-
1. Verl A and Frey S. Correlation between feed velocity ing joints on the dynamic characteristic of ball screw
and preloading in ball screw drives. CIRP Ann: Manuf feed systems in a milling machine. Shock Vib 2015; 2015:
Techn 2010; 59: 429–432. 1–11.
2. Altintas Y, Verl A, Brecher C, et al. Machine tool feed 19. Zhu J, Zhang T and Li X. Dynamic characteristic
drives. CIRP Ann: Manuf Techn 2011; 60: 779–796. analysis of ball screw feed system based on stiffness char-
3. Kamalzadeh A, Gordon DJ and Erkorkmaz K. Robust acteristic of mechanical joints. J Mech Eng 2015; 51:
compensation of elastic deformations in ball screw 72–82.
drives. Int J Mach Tool Manu 2010; 50: 559–574. 20. Huang HTT and Ravani B. Contact stress analysis in
4. Zhou Y, Peng F and Chen J. Torsion vibration analysis ball screw mechanism using the tubular medial axial rep-
of lead-screw feed drives with changeable table position resentation of contacting surfaces. J Mech Design 1997;
and work-piece mass. In: Proceedings of 2007 IEEE 19: 8–14.