REVIEW OF SCIENTIFIC INSTRUMENTS 85, 104502 (2014)

Ultra-low frequency vertical vibration isolator based on LaCoste

spring linkage
G. Li,1,2,3 H. Hu,1,2,3,a) K. Wu,1,2,3 G. Wang,1,2,3 and L. J. Wang1,2,3
1
Joint Institute for Measurement Science, Tsinghua University, Beijing 100084, China
2
State Key Laboratory of Precision Measurement Technology and Instruments, Tsinghua University,
Beijing 100084, China
3
Department of Precision Instrument, Tsinghua University, Beijing 100084, China
(Received 21 July 2014; accepted 28 September 2014; published online 15 October 2014)
For the applications in precision measurement such as absolute gravimeter, we have designed and
built an ultra-low frequency vertical vibration isolator based on LaCoste spring linkage. In the sys-
tem, an arm with test mass is suspended by a mechanical extension spring, and one end of the arm is
connected to the frame with flexible pivots. The displacement of the arm is detected by an optical re-
flection method. With the displacement signal, a feedback control force is exerted on the arm to keep
it at the balance position. This method can also correct the systematic drift caused by temperature
change. In order to study the vibration isolation performance of the system, we analyze the dynamic
characteristics of the spring linkage in the general case, and present key methods to adjust the natural
oscillating period of the system. With careful adjustment, the system can achieve a steady oscillation
with a natural period up to 32 s. This isolator has been tested based on the T-1 absolute gravime-
ter. A statistical uncertainty of 2 μGal has been achieved within a typical 12 h measurement. The
experimental results verify that the isolator has significant vibration isolation performance, and it is
very suitable for applications in high precision absolute gravity measurement. © 2014 AIP Publishing
LLC. [http://dx.doi.org/10.1063/1.4897488]

I. INTRODUCTION sary to suppress the cross-coupling between different freedom

Ultra-low frequency vibration isolation has valuable ap-
plications in many fields such as precision measurement,
scientific instruments, high-technology manufacturing, ad-
vanced imaging, etc.1 In terms of the motion direction, it is
classified into horizontal and vertical vibration. As vertical vi-
bration causes variation of the gravity potential energy, vibra-
tion isolation along vertical direction is more difficult.2 The
main sources of the vertical seismic noise are human activ-
ity and terrestrial micro-seismic activity generated by com-
plicated natural factors. The frequency of the human related
noise is generally above 0.5 Hz while the dominant spectrum
of the terrestrial micro-seismic noise extending from 0.05 Hz
to 0.5 Hz.3 To suppress the seismic noise significantly, the
vibration isolator requires a resonant frequency below 0.05
Hz, corresponding to a natural oscillating period of more than
20 s.
For example, in modern free-fall absolute gravity mea-
surement, the movement of a falling body is measured by a
laser interferometer. In order to suppress the disturbance of
ground seismic noise, a corner cube is suspended by a verti-
cal vibration isolator to behave as the reference point of the
interference measurement.4–6 The most famous vibration iso-
lator for absolute gravimeter is the Superspring,7, 8 a compo-
nent of the commercial free-fall absolute gravimeter FG5.4
The Superspring can achieve a period up to 40–60 s by servo
controlling the top end of a spring to follow the motion of the
isolated mass suspended by the spring. As a cost, it is neces-
a) Author to whom correspondence should be addressed. Electronic mail:
huhua@tsinghua.edu.cn.
degrees of the isolated mass, and the mechanical structure of
the Superspring is complicated.
In this paper, we will introduce a new vertical vibra-
tion isolator based on LaCoste spring linkage.9 The La-
Coste spring linkage has been applied widely in long period
seismometers10 and relative gravimeters.11 In addition, there
are also some researches using it for the vibration isolation,
particularly in the absolute gravimetry12 and the gravitational
wave detecting.13 First, we review the ideal theoretical model
of LaCoste linkage and analyze its dynamic characteristics in
the general case. Then the configuration and performance of
the isolator are described. Finally, the isolator is tested on the
homemade T-1 absolute gravimeter,5, 14 a prototype built in
Tsinghua University, and the experiment results are presented
to demonstrate the isolation performance.
II. THEORETICAL MODEL
The diagram of LaCoste linkage is shown in Fig. 1. The
mass M is fixed to the arm OB which is pivoted at O. A zero-
length spring AB attaches to the frame at A and attaches to the
arm at B. The meanings of the symbols in Fig. 1 are shown in
Table I.
The tension force of a spring follows that
FS = k(lS − l0 ), (1)
where k is the spring constant. If l0 = 0, the force-length curve
of the spring goes through the coordinate origin. Such a spring
is the so-called zero-length spring. Similarly, the spring is
a negative-length spring when l0 < 0, and a positive-length

0034-6748/2014/85(10)/104502/6/$30.00 85, 104502-1 © 2014 AIP Publishing LLC

104502-2 Li et al. Rev. Sci. Instrum. 85, 104502 (2014)

After derivation, the torque equation becomes

   
l0
J θ̈ = cos θ kl1 l2 1 − cos θ0 − MlM (g + ÿO )
lS
 
l
+ kl1 l2 1 − 0 sin θ sin θ0 − KO θ. (4)
lS
When the arm is balanced, ÿO = 0, θ = 0, and θ̈ = 0. If θ
is limited in a very small range, sinθ ≈ θ , cosθ ≈ 1, |θ |2
 |θ |. Then, according to Taylor expansion, by ignoring
higher order terms of θ , Eq. (4) can be simplified as
J θ̈ = −MlM ÿO
 
kl1 l2 l0 sin θ0
− KO − kl1 l2 sin θ0 +  θ.
l1 2 + l2 2 + 2l1 l2 sin θ0
(5)
FIG. 1. Diagram of LaCoste linkage.
The total stiffness coefficient is
kl1 l2 l0 sin θ0
spring when l0 > 0. Obviously, an ordinary spring made by K = KO − kl1 l2 sin θ0 +  . (6)
normal manufacturing process is a positive-length spring. l1 2 + l2 2 + 2l1 l2 sin θ0
In the ideal theoretical model of LaCoste linkage, the If taking the damping effects into account, the dynamic equa-
spring is a zero-length spring and the attachment point A of tion becomes
the spring is right above the pivot O, which means l0 = 0 and
θ 0 = 0. Then the total torque exerted on the arm is − MlM ÿO = J θ̈ + ζ θ̇ + Kθ. (7)

T0 = hFS − MglM cos θ = (kl1 l2 − MglM ) cos θ. (2) Here, ζ is the damping coefficient.
Applying Laplace transformation to the dynamic equa-
If kl1 l2 = MglM , then T0 = 0 for all values of θ and the nature tion, we get the transfer function as
oscillating period of the arm is infinite.
In practice, the natural oscillating period is up to tens of θ (s) −MlM s 2
= . (8)
seconds. According to the vibration theory, a long period sys- yO (s) J s2 + ζ s + K
tem has suppression effect on seismic noise. In order to apply
Here, s is the complex variable of the transfer function.
the spring linkage as a vibration isolator, it is necessary to
Select one point on the arm OB as the output point D. The
analyze its dynamic characteristics. Besides, the theoretical
distance between D and the pivot O is lD .
model should have applicability in the general case, for ex-
The coordinate of D in the y axis is
ample, when l0 = 0 or θ 0 = 0. Some other factors such as the
restoring torque of the pivot should also be taken into account. yD = yO + lD sin θ ≈ yO + lD θ. (9)
The dynamic torque equation of the arm with respect to
the pivot O is The transfer function between yO and yD is

dLor −
→ − → yD (s) (J − MlM lD )s 2 + ζ s + K
= J θ̈ = FS h − MglM cos θ − KO θ − M lM × ÿO . = . (10)
dt yO (s) J s2 + ζ s + K
(3)
Here, J is the rotary inertia of the arm OB with respect to the If the spring linkage performs as a vibration isolator, yO is the
pivot O and K0 is the stiffness coefficient of the pivot and ÿO vibration noise and yD is the isolated response. According to
is the acceleration of the pivot O in the y axis. the transfer function in Eq. (10), for excellent isolation per-
formance, the location of the output point D should satisfy
the following relation:
TABLE I. Meanings of the symbols in Fig. 1.
J − MlM lD = 0. (11)
Symbol Meaning
If the mass of the arm OB is all centralized in the mass point
l1 Distance from Point A to Pivot O M, J = MlM 2
, lD = lM , the right output point D is just the mass
l2 Distance from Point B to Pivot O point M. In practice, however, the right output point should be
lM Distance from the mass point of M to Pivot O
selected according to the calculation about the mass distribu-
θ Angle between the arm and the horizontal
tion of the arm OB.
θ0 Angle between Line AO and the vertical
h Vertical distance from Pivot O to the spring
As the stiffness of the joints such as the pivots contributes
lS Length of the spring to the total stiffness of the system, the extension spring should
l0 Initial length of the spring provide negative stiffness to compensate the existing stiffness.
In Eq. (6), it is shown that there are two ways to adjust the
104502-3 Li et al.
total stiffness coefficient K to a value close to zero, i.e., to
achieve a very long natural oscillating period. One way is to
increase θ 0 , which means to move the top attachment point
of the spring towards the negative direction of the x axis. The
other way is to reduce the initial length l0 of the spring to
a negative value. As the coefficient of θ 0 in Eq. (10) is far
greater than that of l0 , the first way to adjust K has obvious
effect. In addition, it is necessary to keep K positive because a
stiffness coefficient equal to or below zero will lead to system
instability.
III. SYSTEM DESCRIPTION
A. Mechanical structure
The schematic of the spring linkage is shown in Fig. 2.
The arm is connected to the frame with flexible pivots and
is suspended by a mechanical extension spring. The top end
FIG. 2. (a) and (b) Set up of the zero-length spring linkage (A) flexible piv-
ots, (B) arm, (C) counterweight, (D) corner cube, (E) voice coil, (F) magnet,
(G) mirror, (H) travel limiter, (I) mechanical extension spring, (J) micropo-
sitioner, (K) flexible pivots, (L) adjusting lever, (M) holding spring, (N) mi-
crometer screw, (O) laser collimator, (P) optical detector, (D) spectroscope.
Rev. Sci. Instrum. 85, 104502 (2014)
of the spring is attached to an adjusting lever. A micrometer
screw is used to turn the adjusting lever and then the spring
force is changed to balance the gravity of the arm. The attach-
ment point of the extension spring to the adjusting lever can
be moved along the x direction.
The extension spring is made of spring steel. Its initial
length can be determined as follows. A spring oscillator is
constructed by hanging a mass vertically with the spring. Its
natural oscillating period follows the relation as

T = 2π m/k. (12)
Here, k is the stiffness coefficient of the spring and m is the
mass.
Then the initial length of the spring can be determined as
l0 = l − mg/k = l − g (T /2π )2 . (13)
Here, l is the practical length of the spring when the spring
oscillator is motionless, g is the gravity acceleration, and T
is the oscillating period when the spring oscillator oscillates
freely. We measured l = 204 mm and T = 0.985 s. Then the
initial length of the spring was calculated as l0 = −37 mm.
Afterwards, the extension spring is connected into the
linkage with some steel wire. Then the initial length of
the spring became the sum of the original initial length and
the length of the connected steel wire. Therefore, by vary-
ing the length of the connected steel wire, the resultant initial
length can be adjusted.
B. Optical detection and feedback control
As shown in Fig. 2, a laser reflection method is used to
detect the deflection of the arm. The output signal of the op-
tical detector contains information about the deflection angle
of the arm. To determine the detection sensitivity, the output
signal of the optical detector is measured as the deflection an-
gle of the arm changes. The sensitivity curve of the optical
detector is shown in Fig. 3. The fitting equation indicates that
the sensitivity was 2.3 mV/μrad. As the RMS (Root Mean
Square) noise of the output signal is 0.7 mV, the minimum
deflection angle of the arm that can be distinguished is about
0.3 μrad. The distance from the mirror to the pivots is about
FIG. 3. Sensitivity curve of the optical detection module.
104502-4 Li et al. Rev. Sci. Instrum. 85, 104502 (2014)

FIG. 4. (a)–(c) Natural oscillations of the spring linkage.

150 mm, so a deflection angle of 0.3 μrad corresponds to a 0.03 Hz. That is below the dominate frequencies of the verti-
vertical displacement of 45 nm. cal seismic noise caused by human activity and the terrestrial
Fig. 4(a) shows a typical oscillation of the arm detected micro-seismic activity. In theory, the system can suppress the
by the optical detector. Its natural oscillating period is about seismic noise significantly.
15 s. It can be seen that the balance position of the arm suffers The Superspring is of a period up to 40–60 s, correspond-
from drift. The system will stop operating if the arm drifts ing to a resonant frequency below 0.025 Hz. The isolator
out of the work area. As the max magnitude of the output in Ref. 13 which applies an Euler-LaCoste linkage achieves
signal of the optical detector is ±10 V and the sensitivity is a vertical resonance frequency of 0.15 Hz. Compared with
2.3 mV/μrad, the work area in which the arm can be detected these isolators, our isolator has achieved a typical resonant
by the optical detector is about ±4.3 × 10−3 rad. In practice, frequency for ultra-low frequency vertical vibration isolation.
the deflection of the arm is limited in a smaller range. The ex-
tension spring is made of spring steel, whose elastic modulus
has large temperature coefficient. Therefore, the tension force IV. APPLICATION ON ABSOLUTE GRAVIMETRY
of the spring varies with temperature change, which leads to In order to test the isolation performance, the vibra-
systematic drift. tion isolator has been applied in the homemade T-1 absolute
In order to correct the systematic drift, a feedback con- gravimeter (Fig. 5). The T-1 absolute gravimeter is a classical
trol electronics is designed. The signal of the optical detec-
tor is processed by a Proportional-Integral-Derivative Control
(PID) circuit. A voice coil assembled between the arm and
the frame baseplate (shown in Fig. 2) transforms the current
to feedback force. Among the three terms of PID control, the
integral part plays the decisive role in the correction for the
systemic drift. It limits the deflection angle of the arm in a
very small range to keep the stability of the system. The pro-
portional part and the derivative part contribute to K and ζ
of Eq. (7), respectively. As the noise of the current will be
transformed to disturbance force by the voice coil, the output
signal of the feedback circuit should be filtered properly. With
the application of feedback control, the drift problem has been
solved, as what is shown in Fig. 4(b).
By moving the top attachment point of the extension
spring and varying its initial length, the natural oscillating pe-
riod of the spring linkage can be adjusted in a wide range.
The oscillation of the arm shown in Fig. 4(c) reaches a natu-
ral period of 32 s, corresponding to a resonant frequency of FIG. 5. Set up of the T-1 absolute gravimeter.
104502-5 Li et al.
FIG. 6. Results of a typical 12 h measurement of T-1 with the spring linkage
as the vertical vibration isolator.
free-fall style prototype designed and built in Tsinghua Uni-
versity. In the apparatus, a falling body is dropped in vacuum
and its free-fall movement is measured by a laser interferom-
eter. Then the gravity acceleration g is calculated according
to the measured time and distance pairs with fitting methods.
To suppress the disturbance of ground seismic noise, a corner
cube is attached to the test mass of the isolator to behave as
reference mirror of the laser interferometer.
Fig. 6 shows the result of a typical 12 h measurement
of T-1 with our isolator at the Changping gravity station
of National Institute of Metrology of China. For the grav-
ity measurement, one set of 16 drops was taken each half
hour. In Fig. 6, the dot is the measured mean value of
one set, the error bar is the standard deviation and the as-
terisk is the theoretical solid tide gravity influence. With
the 12 h measurement, a statistical uncertainty of 2 μGal
(1 μGal = 1 × 10−8 m/s2 ) is achieved, after applying correc-
tion for solid tides. With the developed vibration isolator, the
standard deviation of a single drop is about 30–40 μGal, while
the value without vibration isolation is about 200–300 μGal.
The disturbance of the ground seismic noise is suppressed
by the isolator significantly. Besides, the isolator can op-
erate continuously more than 12 h, which demonstrates
that the system is robust and suitable for long-term gravity
measurements.
Besides, we have also applied a commercial Superspring
on the T-1 absolute gravimeter at the Changping gravity sta-
tion. The statistical uncertainty with a 12 h measurement is
about 0.5–1 μGal. Reference 15 discusses the IMGC-02 ab-
solute gravimeter built by the Italian Istituto Nazionale di
Ricerca Metrologica (INRIM). The IMGC-02 employs a re-
fitted 20 s long-period seismometer as the vertical vibration
isolator. Within a single observation session which also lasts
about 12 h, the gravimeter achieves a statistical uncertainty of
1.8 μGal. The performance of our isolator, achieving a statis-
tical uncertainty of 2 μGal with a 12 h measurement, is close
to that of these apparatuses.
For a single drop, the interference fringe signal of the
laser interferometer is collected by a high-speed signal acqui-
sition card. Then the digitized data are processed by computer
software to detect the time of zero-crossings. Afterwards, the
Rev. Sci. Instrum. 85, 104502 (2014)
FIG. 7. Analysis of the data residuals. (a) Fitting residuals for a single drop.
(b) Spectrum of residuals for a single drop.
gravity acceleration is calculated by fitting the time and dis-
tance pairs. The fitting residuals reflect the disturbance to the
laser interference measurement. A typical residuals plot is
shown in Fig. 7(a).The amplitude of the residuals on a single
drop is about 2 nm. The spectrum of the residuals which is ob-
tained by fast Fourier transform (FFT) is shown in Fig. 7(b).
Most of the power in the residuals is concentrated in three
frequencies centered at about 30 Hz, 130 Hz, and 1250 Hz.
Similar spectrums are found in the fitting residuals of other
drops.
The sources of the three frequencies are not clear at
present. We surmise that the frequency of 1250 Hz is caused
by the bell mode of the falling body, which refers to the low-
est order body mode of the falling body itself when it rings
like a bell after given an impulse. The other two frequencies
are surmised to be caused by the mechanical resonances of
the isolator, which probably refer to the vibration modes of
the mechanical extension spring. Similar discussion is found
in Refs. 16 and 17. Further research is needed to verify the
sources of the fitting residuals and its influences to the mea-
sured g values.
V. CONCLUSION
We have designed and built a vertical vibration isolator
based on LaCoste spring linkage. Applying the special me-
chanical characteristics of extension spring, LaCoste spring
linkage achieves a long natural oscillating period with a rela-
tively simple structure. It is a beneficial attempt to use it as a
vertical vibration isolator.
The current urgent problem in the spring linkage is
the systematic drift caused by temperature change. We have
solved the problem with PID feedback control. In the future,
we plan to try some research to reduce the systematic drift
104502-6 Li et al.
from the source, for example, to control the temperature in-
side the system, and to make the spring with material of low
temperature coefficient such as the constant elastic alloy.
The isolator has been applied in the T-1 absolute
gravimeter and a statistical uncertainty of 2 μGal was
achieved with a typical 12 h measurement. The experiment
results demonstrate that the isolator is robust and suitable
for the vertical vibration isolation for high precision absolute
gravimeter.
ACKNOWLEDGMENTS
The authors would like to thank the Tsinghua Uni-
versity Initiative Scientific Research Program (Grant No.
2013THZ05) for providing support for this research.
1 F. H. Liao, X. P. Li, X. D. Chen, Z. Y. Yuan, and W. Li, Mech. Sci. Technol.
Aerosp. Eng. 31, 1411 (2012).
2 J. Winterflood and D. G. Blair, Phys. Lett. A 300, 122 (2002).
3 D. B. Newell, S. J. Richman, P. G. Nelson, R. T. Stebbins, P. L. Bender,
and J. E. Faller, Rev. Sci. Instrum. 68, 3211 (1997).
4 J. E. Faller, Metrologia 39, 425 (2002).
Rev. Sci. Instrum. 85, 104502 (2014)
5 T. M. Niebauer, G. S. Sasagawa, J. E. Faller, and F. Klopping, Metrologia
32, 159 (1995).
6 H. Hu, K. Wu, S. Lei, G. Li, and L. J. Wang, Acta Phys. Sin. 61, 099101
(2012).
7 R. L. Rinker and J. E. Faller, “Precision measurement and fundamental
constants II,” in Proceedings of the Second International Conference (NBS-
SP-617) (National Bureau of Standards, Washington, DC, 1984), pp. 411–
417.
8 P. G. Nelson, Rev. Sci. Instrum. 62, 2069 (1991).
9 L. J. B. LaCoste, Jr., J. Appl. Phys. 5, 178 (1934).
10 L. C. Ye, Acta Geophys. Sin. 24, 415 (1981).
11 D. Crossley, J. Hinderer, and U. Riccardi, Rep. Prog. Phys. 76, 046101
(2013).
12 S. Grams, Diploma thesis, University Erlangen-Nuremberg, Erlangen,
2008.
13 M. A. Hosain, A. Sirr, L. Ju, and D. G. Blair, Rev. Sci. Instrum. 83, 085108
(2012).
14 C. H. Rothleitner, S. Scitlov, H. Merimeche, H. Hu, and L. J. Wang,
Metrologia 46, 283 (2009).
15 G. D’Agostino, S. Desogus, A. Germak, C. Origlia, D. Quagliotti, G.
Berrino, G. Corrado, V. d’Errico, and G. Ricciardi, Ann. Geophys. 51, 39
(2008).
16 T. M. Niebauer, R. Billson, B. Ellis, B. Mason, D. V. Westrum, and F.
Klopping, Metrologia 48, 154 (2011).
17 H. Hu, S. Svitlov, C. Rothleitner, J. Schafer, J. Zhang, and L. J. Wang,
Metrologia 47, 575 (2010).
Review of Scientific Instruments is copyrighted by the American Institute of Physics (AIP).
Redistribution of journal material is subject to the AIP online journal license and/or AIP
copyright. For more information, see http://ojps.aip.org/rsio/rsicr.jsp