Você está na página 1de 8

Microsyst Technol (2012) 18:15591566

DOI 10.1007/s00542-012-1592-z

TECHNICAL PAPER

Shock and vibration isolation of laptop hard disk drive


using rubber mount
Kyoung-Su Park Seungho Lim Young-Pil Park

Young-Bae Chang No-Cheol Park

Received: 26 September 2011 / Accepted: 11 June 2012 / Published online: 24 June 2012
Springer-Verlag 2012

Abstract Hard disk drives (HDDs) in laptop personal suffer critical failure when dropped or otherwise subjected
computers (PCs) are subject to impact-induced failure, as to external impacts. To minimize the likelihood of these
well as to operational vibrations. Elastic mounts with mechanical failures, shock and vibration analysis should be
cushioning materials are often used to minimize the like- incorporated into the design of system mountings for
lihood of failure in such cases. In this paper, we analyze the HDDs.
dynamic characteristics and shock response of the vibration The effects of shock and vibration on HDDs have been
isolation systems supporting HDDs. Anti-vibration per- extensively researched. Jayson et al. (2003) investigated
formance is investigated via position error signal and the effects of air bearing stiffness in response to external
acceleration transmissibility. Shock response is obtained shock. Zhu et al. (2006) employed topology optimization to
from a lumped parameter model, based on the nonlinearity improve the dynamic characteristics of HDD suspensions.
of the rubber mounts. Based on the results, we propose Muraoka et al. (2002) simulated the vibrations of the
guidelines for shock and vibration isolation systems, contact slider and gimbal of a flexible disk drive. In recent
including a dual rubber mount design. years, studies on the shock analysis of HDDs have become
more widespread in response to the increasing importance
of laptop portability. Harmoko et al. (2007) suggested a
1 Introduction more efficient approach for investigating the effects of
various HDD components on shock tolerance. Drop test
Mobile devices have become an important part of daily simulations and power spectrum analysis of a head actuator
life. This is especially true of laptop personal computers assembly were carried out by Shi et al. (2007). The shock
(PCs), which are portable enough to be used almost any- response of small-form-factor HDDs was analyzed by
where. However, the portability of laptop PCs is compro- Murthy et al. (2007) using the finite element method. In
mised if their components are not robust enough to endure addition, the latch design was introduced by Kuwajima
rugged operating environments. For example, the speakers et al. (2007) to enhance rotary shock resistance.
and optical disk drives (ODDs) installed in laptop PCs can However, only a few of these investigations have ana-
cause excessive vibrations, leading to user discomfort or lyzed or suggested improvements to the vibration isolation
unexpected power shutdowns in extreme cases. In partic- systems for laptop HDDs (Harmoko et al. 2009; Suwa and
ular, the read/write performance of an HDD is very vul- Aruga 1999), and none of these have systematically con-
nerable to excessive vibrations. Moreover, laptop PCs often sidered the rubber mount system, which is the most popular
method for isolating and supporting an HDD.
K.-S. Park  S. Lim  Y.-P. Park  N.-C. Park (&) In this research, we focused on the dynamic behavior of
Center for Information Storage Device, Yonsei University, a 2.5-in. laptop HDD supported by rubber mounts. Accel-
Seoul, Korea eration transmissibility and position error signal (PES)
e-mail: pnch@yonsei.ac.kr
were examined with respect to the type of rubber mount to
Y.-B. Chang qualify the anti-vibration performance. Since rubber
Seagate Korea Design Center, Suwon, Korea mounts are considerably deformed during external shock,

123
1560 Microsyst Technol (2012) 18:15591566

we considered the nonlinearity of the mounts by varying When the rubber mounts were not attached to or contacting
the stiffness and damping properties extracted from modal an HDD, they were vibrated by a magnetic exciter, and the
tests. The shock response was obtained via a 2-degree of frequency response function (accelerance) was measured
freedom (DOF) lumped parameter model, the validity of using an impedance head, while the compression ratio was
which was verified by a linear drop test. Finally, guidelines varied by a micro-stage, as shown in Fig. 2a. When the
based on the use of a dual rubber mount design are pro- rubber mounts supported an HDD, the compression ratio
posed for simultaneously improving the shock and vibra- due to the weight of the HDD was considered, as shown in
tion isolation performance. Fig. 2b. We attached two accelerometers, one to the jig and
the other to the bottom of the HDD spindle motor because
of its rigidity. An isolated laptop HDD was represented by
2 Vibration isolation a single-degree-of-freedom model. The estimated material
properties were automatically calculated by considering the
Nowadays, people often use laptops to enjoy multimedia preload induced by the HDD weight. From the measured
content, including movies and music. High-output speakers maximum value of the acceleration transmissibility (Tmax)
and ODDs create vibration problems, sometimes leading to and the damped natural frequency (wd), the damping ratio
power resets or increased operating time for copying data. and stiffness were calculated via Eqs. (1), (2), (3) using a
For vibration isolation, the 2.5-in. HDD is cushioned by 16 numerical search method based on the implicit Eq. (1)
rubber mounts, as shown in Fig. 1. Since the gap between (Kelly 2000).
the HDD and the laptop body frame should be 2 mm or less " p #1=2
(depending on the laptop specifications), it is impossible to 2 1 8f2
Tmax 4f  p
utilize rubber mounts of various shapes. Thus, the mounts 2 16f2 16f4  8f2  2 1 8f2
have a simple rectangular shape, and lumped parameters 1
for rubber (e.g., stiffness and damping ratio, excluding
shape factors) are the main focus in this research. This x2d
k m 2
research concentrates on vibration and shock response for 1  f2
the vertical direction since this is the direction that shock p
c 2f mk 3
and vibration are predominantly transmitted to laptops. The
tilt vibratory mode of the system is rarely excited, due to In this work, a silicon elastomer and five butyl rubbers
the symmetric positioning of the mounts. (which are widely employed in electrical appliances) were
used for vibration isolation. Figure 3 shows the measured
2.1 Material properties of rubber mounts

It is important to obtain exact material properties for


dynamic analysis. The experimental setup shown in Fig. 2
was employed to represent the stiffness and damping
properties of rubber mounts with lumped parameters.

Fig. 1 Vibration isolation system supported by rubber mounts Fig. 2 Experimental setup for estimating material property

123
Microsyst Technol (2012) 18:15591566 1561

Table 1 Estimated modal parameters for each of rubber samples


Rubber type Damped natural Max. Damping Undamped natural Stiffness Damping
frequency (Hz) transmissibility ratio frequency (Hz) (MPa) coefficient (Ns/M)

R1 83 1.69 0.386 89 0.031 42.81


R2 94 1.64 0.407 102 0.041 51.63
R3 376 3.59 0.146 380 0.559 68.43
R4 61 1.73 0.373 65 0.017 30.15
R5 116 2.35 0.241 119 0.055 35.46
R6 660 6.36 0.080 662 1.696 65.12

acceleration transmissibility and the simulated acceleration These vibrations amplify the acceleration response in
transmissibility based on the calculated material properties. a frequency range between 50 and 1,500 Hz, as shown
The values are summarized in the Table 1. The frequency in Fig. 4. Accordingly, it is necessary to isolate vibra-
responses agreed well, and hence the estimated material tions primarily in this frequency region. As Fig. 3
properties were appropriate. clearly shows, models R3 and R6 provide poor vibra-
tion isolation performance. However, aside from these
2.2 Vibration transmissibility two models, it is difficult to compare anti-vibration
performances. One method of quantifying the perfor-
Figure 4 shows the acceleration responses transmitted to mance is to integrate cumulatively a curve in the spe-
the HDD of a typical laptop from various sources, cific frequency range 01,500 Hz, as summarized in
including speaker and optical disk vibrations. Table 2.

Fig. 3 Acceleration
transmissibility

R4
R6
R1
R2
R5 R3
Acceleration Transmissibility

R4
R1 R6
R2
R5 R3

Frequency[Hz]

123
1562 Microsyst Technol (2012) 18:15591566

rotational speed was 5,400 rpm, and PES signals with


frequencies between 0 and 2,970 Hz were validated via a
sensing limit determined by the Nyquist frequency theorem
for a given sampling frequency. Figure 5 shows the PES to
be measured and compared for various rubber mount
models. Table 2 summarizes the experimental results of the
acceleration transmissibility and standard deviation of the
PES signal during the operation. Here, 512 counts which is
PES unit represent a track. The acceleration transmissi-
bility and PES exhibited similar tendencies. Accordingly,
the vibration isolation performance of newly designed
rubber mounts could be effectively evaluated in terms of
PES performance by using acceleration transmissibility. In
addition, rubbers with low stiffness (such as R1 and R4) are
Fig. 4 Acceleration response near HDD excited by various sources more advantageous for vibration isolation.

Table 2 Measured PES and transmissibility 3 Shock analysis


Model PES (standard deviation, unit: counts) Transmissibility
3.1 Lumped parameter model
RRO NRRO Total PES

R1 8.32 10.8 13.6 248.5 Consider an isolated laptop HDD with a 2-DOF system, as
R2 10.9 9.57 14.5 301.5 shown in Fig. 6. The isolation system consists of the laptop
R3 10.6 15.5 18.8 1,423 body with rubber feet (the lower isolator) and the laptop
R4 10.8 8.93 13.9 179.5 HDD supported by rubber mounts (the upper isolator). The
R5 10.7 9.28 14.2 432.1 masses of the laptop body and HDD are respectively
R6 9.98 35.4 36.8 2,769 denoted by m1 and m2. The bodies move relative to the
base, so that the line of motion of the bodies coincides with
that of the base. The lumped masses are regarded as
absolutely rigid bodies.
2.3 Position error signal (PES) The rubber mounts are represented by a lumped stiffness
and a damping coefficient, which vary with the compressed
Although the vibration performance of an HDD can easily displacement. A large relative displacement between the
be estimated in terms of acceleration transmissibility, HDD HDD and laptop body causes the stiffness to increase. Based
performance associated with the data transfer rate between on experimental results, the stiffness is approximated via
the head and the disk is better reflected by the PES char-
acteristics (Chang et al. 2002). PES is an important index
for predicting the track mis-registration (TMR) level, and
also provides useful information about the characteristics
of the vibrations induced by the HDD itself. PES can be
classified in terms of two constituents: repeatable run-out
(RRO) and non-repeatable run-out (NRRO). The RRO
component of PES results from spindle imbalance and
servo track written errors, and the RRO signal is usually
proportional to spindle RPM. The NRRO component
(which is the remaining error) is mainly induced by
structural vibrations of the rotating disk, suspension actu-
ator, and external excitations due to air flow and user
handling.
In order to measure PES, the slider head of the HDD is
controlled to locate on a specific track, and PES is trans-
ferred to an external PC via a hyper terminal. In this
research, a 2.5-in. HDD with two platters was used. The Fig. 5 Measured position error signal (PES)

123
Microsyst Technol (2012) 18:15591566 1563

   
c2 c2 c1 k2 k2 k1
z1  z_1 z_2  z1 z2
m2 m1 m1 m2 m1 m1
7
c2 c1 k2 k1
z2 z_1  z_2 z1  z2  y 8
m1 m1 m1 m1
where z1 x2  x1 and z2 x1  y:
The equation of motion can be expressed in the matrix
form.
     " #
Z_ 0 I Z 0
y 9
Z C K Z_ 1
  "
#
Fig. 6 Lumped parameter model for shock analysis z1  mc22 mc21 c1
m1
where Z , C ; K
z2 c2
 mc11
m1
exponential curve-fitting, in accordance with the following "
#
 mk22 mk21 k1
Eq. (4). m1
k2
 mk11
k2 z2 k0 eaz2 4 m1

The coefficients k0 and a denote the initial stiffness and 3.2 Linear drop test
the level of nonlinearity (hereafter referred to as the
stiffness factor), respectively. Figure 7 shows the measured A linear drop tester was developed to verify the mathe-
stiffness with respect to the deflection, and its matical model, as shown in Fig. 8. A jig with an isolated
approximation using Eq. (4). In addition, the stiffness and HDD descended along the linear guide and collided with
damping properties of the rubber feet on the bottom of the damping sheets, which could be used to adjust the
laptop body are assumed to be constant, and other shock duration. Two accelerometers were attached, one to
parameters used in the simulation are listed in Tables 3 the moving jig of the drop tester and the other to the back
and 4. The equation of motion is given by Eqs. (5) and (6). of the HDD spindle motor. Without loss of generality, the
m1 x1 c2 x_ 1  x_ 2 k2 x1  x2 c1 x_1  y_ shock input acceleration from the experiment illustrated in
Fig. 9 was applied to the analytical model instead of a half-
k1 x1  y 0 5
sine pulse, which is difficult to realize accurately in a linear
m2 x2 c2 x_ 2  x_ 1 k2 x2  x1 0 6 drop test. The derived equation of motion was solved
numerically via a nonlinear technique, such as the fourth-
For base excitation, it is convenient to introduce the
order RungeKutta method. After solving this nonlinear
relative displacement and velocity such as z1 and z2 as
ordinary differential equation, we compared the shock
follows.
response with the experimental results. If the stiffness was
assumed to be linear, the shock response was underesti-
mated, as shown in Fig. 10, and the predictions were quite
different from the experimental results. On the other hand,
the nonlinear simulation considered the stiffness variation
with respect to the compression ratio, as shown in Fig. 11.
The experimental transmitted acceleration was in good
agreement with the results of the nonlinear simulation.

3.3 Design guideline for shock performance


improvement

The shock response is affected by the shape of the shock


acceleration pulse, which can be a half-sine, a versed-sine,
triangular, dual quadratic, or rectangular (Harris 1997).
Among these, we chose the half-sine acceleration pulse
with a maximum amplitude ( ypeak ) of 120 G and a shock
Fig. 7 Non-linear stiffness and its approximation duration (s ) of 2 ms, since it is the one commonly used in

123
1564 Microsyst Technol (2012) 18:15591566

Table 3 Applied system parameter transmissibility (output-to-input ratio), the product of the
Parameters Value
shock duration and the natural frequency of the isolation
system must be less than about 1.07, as shown in Fig. 12.
Mass of HDD 0.098 kg However, the maximum deflection of the isolator may
Mass of laptop body 1.8 kg increase beyond the gap between the HDD and the laptop
Stiffness of rubber feet (k1) 1.57 MN/m body. For a specific shock duration, the natural frequency
Damping ratio of rubber feet (c1) 0.21 may be low enough to be within region 1 of Fig. 12, sat-
isfying the limit of maximum deflection. Otherwise, it
would be advantageous to raise the natural frequency for
Table 4 Coefficients for exponential curve-fitting setting the transmissibility curve, not in region 3, but in
region 2. Figure 13 shows the transmitted acceleration
Rubber type Initial stiffness (k0) Level of nonlinearity (a)
considering the nonlinearity of the rubber mounts. These
R1 31,328 2,915 nonlinear properties amplified the shock transmissibility,
R2 40,985 2,728 degrading their performance as shock isolators. In general,
R3 558,910 970.3 rubber mounts with lower initial stiffness have a high
R4 16,716 3,191 stiffness factor, and vice versa. Accordingly, from the point
R5 55,267 2,268 of view of shock performance only, it is appropriate to
R6 1,696,100 539.5 choose high stiffness and a low stiffness factor.

the electronics industry (especially in laptop HDDs), and it


is easily simulated with a linear drop table. Thus, we
assumed the following shock input acceleration:
  
y sin ps t ; 0\t\s
y peak 10
0; ts
The shock response in the time domain was obtained
based on this input condition. The most important criterion
for characterizing shock performance is the maximum
value of the transmitted acceleration, which is directly
related to the read/write reliability of a laptop HDD. In
addition, the relative displacement is limited to prevent
secondary shock to the HDD and laptop body, since the
marginal space between them has been greatly diminished
to satisfy the increasingly slim packaging requirements for
laptops. Fig. 9 Shock input acceleration from linear drop test
Figure 12 presents a physical interpretation of the shock
transmissibility of a linear system. In order to decrease the

Fig. 10 Shock response comparison between experiment, linear and


Fig. 8 Linear drop tester non-linear simulation

123
Microsyst Technol (2012) 18:15591566 1565

Fig. 12 Shock transmissibility of single dof system with linear


Fig. 11 Stiffness variation stiffness

4 Dual rubber mount design

Flexible rubber mounts can cause collisions between the


HDD and laptop body. Moreover, because of their high
nonlinearity, they amplify the transmitted acceleration. On
the other hand, from the point of view of vibration isola-
tion, the rubber mounts must be flexible enough to isolate
501,500 Hz vibrations. As a consequence, there is a trade-
off relationship between vibration and shock performance.
Because of this, we propose the dual rubber mount design
shown in Fig. 14. Under ordinary conditions, external
vibrations are isolated by the thick rubber mounts, which
Fig. 13 Shock transmissibility for the system in Fig. 5 with respect
have low stiffness. Conventional vibrational motions to stiffness factor and initial stiffness of rubber mount
induced by speakers and ODDs are much smaller than
shock responses, and in this case the HDD remains out of
contact with the thin rubber mounts. For shock input, the where a1, a2, k01, and k02 are the stiffness factor and initial
thinner mounts are attached to the HDD, which increases stiffness of the thin and thick rubbers, respectively. The
the stiffness. As a result, the shock response can be lowered equation of motion was the same as in Eqs. (5), (6), (7), (8),
while maintaining the antivibration performance. (9).
The former lumped parameter model was modified to Figure 15 shows the maximum transmitted accelerations
numerically analyze this dual rubber mount system. The of various dual mounts for an external shock producing an
modified model included both the nonlinearity and the input acceleration pulse with an amplitude of 120 G and a
contact condition of the rubber mounts with respect to shock duration of 2 ms. There were some errors between
the gap between the HDD and laptop body. The material the simulation and the drop test. Furthermore, the drop test
properties of the thin rubber mounts were extracted from did not guarantee reproducibility. The major source of
the experimental setup shown in Fig. 2a. The HDD weight these errors was the manufacturing tolerance for the gap (d)
was not considered for the thin rubber mounts because between the thick and thin rubbers. In addition, the input
(unlike the case of the thick rubber mounts) there was no acceleration profile measured by the drop tester was similar
preload in the initial state. Hence, an equivalent rubber to the versed-sine pulse, which induces a lower shock
mount could be represented as follows: response than the half-sine pulse. Although there were
some errors, the results of the simulation and the drop test
k21 z1 k01 ea1 z1 11 showed similar overall tendencies. In this study, R1 and R4
k22 z1 k02 ea2 dz1 12 were selected for the thick rubber mounts because they
 effectively isolated external vibrations. Various rubber
k z ; z1 \d
k2 21 1 13 mounts were considered for the thin rubber mounts. In
k21 z1 k22 z1 ; z1  d comparison with simple rubber mount designs, the

123
1566 Microsyst Technol (2012) 18:15591566

1. Acceleration transmissibility and PES are compatible


criteria for vibration isolation.
2. The stiffness of a rubber mount can be measured from
the acceleration transmissibility by considering the weight
of the HDD. Since a rubber mount has highly nonlinear
stiffness with respect to the compression ratio, it must be
represented by a nonlinear component for shock analysis.
Otherwise, the shock response is underestimated.
3. Dual rubber mounts improve the anti-shock perfor-
mance up to 15 % in comparison with simple rubber
mounts, while maintaining the anti-vibration performance.

Acknowledgments This work was supported by the National


Research Foundation of Korea (NRF) grant funded by the Korea
government (MEST) (No. 2011-8-0167).
Fig. 14 Lumped parameter model for shock analysis of dual rubber
mount
References

200 Chang YB, Park DK, Park NC, Park YP (2002) Prediction of track
mis-registration due to disk flutter in hard disk driver. IEEE
190
Max. Transmitted Acceleration [G]

Single Trans Magn 38(2):14411446


Rubber R4
180 Single
Harmoko H, Yap FF, Vahdati N, Gan S, Liu M, Shi BJ (2007) A more
Rubber R1 efficient approach for investigation of effect of various HDD
170 (R1 / R5) (R4 / R5)
components on the shock tolerance. Microsyst Technol 13:1331
(R4 / R4)
160
(R1 / R4) 1338
Harmoko H, Yap FF, Vahdati N, Li C (2009) Design and analysis of
150 shock and random vibration isolation of operating hard disk
140
drive in harsh environment. Shock Vib 16:143154
Harris CM (1997) Shock and vibration handbook, 4th ed. McGrall-
130 Hill, New York, pp 31.2031.23
Jayson EM, Murphy J, Smith PW, Talke FE (2003) Effects of air
120 Simulation Rubber Mounts
Experiment (Error Bound) (Thick Rubber / Thin Rubber) bearing stiffness on a hard disk drive subject to shock and
110 vibration. J Tribol Trans ASME 125:343349
Kelly SG (2000) Fundamentals of mechanical vibrations, 2nd ed.
100
McGrall-Hill, Singapore, pp 155157
Kuwajima H, Ueno Y, Umeda M, Inaji T, Ochi A, Matsuoka K (2007)
Fig. 15 The maximum transmitted acceleration of dual rubber mount
New type latch for hard disk drive. Microsyst Technol 13:1417
1424
transmitted acceleration decreased up to about 15 % while Muraoka Y, Ryoson H, Kawashima K, Kondo M, Kaneta Y, Okazaki
maintaining the anti-vibration performance, as shown in Y (2002) Vibration analysis of a contact slider/gimbal of a
flexible disk drive. Microsyst Technol 9:4147
Fig. 15.
Murthy AN, Pfabe M, Xu J, Talke FE (2007) Dynamic response of
1-in. form factor disk drives to external shock and vibration
loads. Microsyst Technol 13:10311038
5 Conclusion Shi BJ, Shu DW, Wang S, Luo J, Meng H, Ng Q, Lau JHT, Zambri R
(2007) Drop test simulation and power spectrum analysis of a
head actuator assembly in a hard disk drive. Int J Impact Eng
In this research, shock and vibration analysis of a laptop 34:120133
HDD on rubber mounts was carried out. Lumped param- Suwa M, Aruga K (1999) Evaluation system for residual vibration
eters, including stiffness and damping ratio, were measured from HDD mounting mechanism. IEEE Trans Magn 35(2):868
873
considering both HDD weight and the nonlinearity of the
Zhu DL, Wang AL, Jiang T (2006) Topology design to improve HDD
mounts. A lumped parameter model for shock analysis was suspension dynamic characteristics. Struct Multidisc Optim
developed and verified by a linear drop test. In addition, a 31:497503
dual rubber mount design was proposed to reduce the
shock response. The following conclusions can be drawn
from this research.

123

Você também pode gostar