Vibration Control Using Piezoelectric Transducers
J. Bert
October 29, 2007
(Submitted as coursework for Physics 210, Stanford
University, Fall 2007)
Controlling mechanical motion and damping unwanted
vibrations in flexible structures can be achieved using piezoelectric
transducers (piezos) [1]. Piezoelectric materials are crystals or
ceramics that generate an internal voltage when stressed. If a
piezoelelectric element is rigidly mounted to a flexible structure, then
vibrations and deformations in the structure are coupled to the attached
piezoelectric transducer. Using the voltage induced in the piezo as an
input signal the stress may be monitored or controlled using an external
shunt or feedback circuit.
This paper briefly
describes the physical origins of the piezoelectric effect and then
details how piezoelectric sensors and actuators are used in vibration
control devices. Vibration control is achieved by using the signal
generated from the piezo element as the input signal to an external
circuit. The circuit can either apply feedback by processing the input
piezo voltage and applying the output signal back to the piezo element, or
the circuit can simply dissipate vibrational energy by electrically
damping the piezo element.
|
Figure 1: Piezoelectric element equivalent circuit. |
The Piezoelectric Effect
Piezoelectric materials are typically ceramic or
crystalline in structure, with permanently aligned electric dipoles. The
dipoles allow for separation of positive and negative charges within the
material, but the symmetry of the crystal assures that there is no
internal electric field. However, if the crystal is stressed then the
crystal symmetry is broken and an internal electric field is generated
between surfaces of the piezo. If the internal field is not compensated,
for example by shorting one side of the piezo to the other, it results in
an induced voltage. The effect of stressing a piezo to generate a voltage
is called the direct piezo effect. The converse piezo effect involves
inducing a stress in the piezo element by a apply an external electric
field or potential.
Piezos are fabricated by adding electrodes to the
surfaces of the ceramic or crystal to either probe the stress induced
voltage or apply a local voltage to the element. An external field
applied parallel to the alignment direction of the dipoles within the
piezo material will cause the piezo to expand along the field direction
and contract in the two directions perpendicular to the field.
The external electrodes and the charge arrangement
within the piezo material give the piezo element a capacitance,
Cp. As a result the equivalent circuit for a piezo
element is the capacitance, Cp, and internal voltage,
Vp arranged in series as shown in figure 1.
Sensors and Actuators
To use a piezo as either a sensor or actuator for a
vibration control device it must be rigidly mounted to the vibration
sensitive part. This assures that any stress or strain in the static
structure is rigidly coupled to the piezo and vice versa. Piezos used as
sensors are often light and flexible to provide the best transmission of
the mechanical vibration into the piezo for detection. Piezos used as
actuators are often denser so that the stress induced in the piezo by an
applied voltage is large enough to stress the structure as well.
|
Figure 2: Flexible structure with collocated sensor
and actuator piezos. |
One of the simplest vibration control
circuits is a collocated pair of sensor/actuator piezo elements.
Collocated means that the two piezos are placed in the same position on
two different sides of a flexible structure, as shown in figure 2.
Assuming the two piezos are identical, meaning they have the same internal
capacitance, if the structure flexes the voltages induced in the two
piezos will be equal but 180º out of phase. After correctly
resolving the transfer functions between the vibrating structure and the
voltage generated in the sensor and between the voltage measured by the
sensor and the voltage applied to the actuator, feedback control can be
established between the sensor output and actuator to oppose the
vibration. This simple model is easy to imagine implementing; however, in
practice small errors in the transfer functions can quickly destabilize
the feedback.
Self-Sensing Piezos
A more
stable, although more complicated, feedback system can be set up with a
self sensing piezo actuator [2]. Such a setup requires just one piezo to
be mounted to the structure. The strain applied by the structure to the
piezo generates a voltage, Vp, which rather than being
fed back to a separate piezo can be used to compensate the same piezo that
generated the signal. The single piezo acts as both sensor and actuator
in the circuit.
One complication for this scheme is that the induced
voltage cannot be probed directly. The external circuit shown in figure 3
is required to separate Vp from the internal capacitor
voltage, Q/Cp, and deduce the strain. The strain voltage
is fed back into the input of the same piezo to compensate the mechanical
vibration.
Assuming the leakage resistors
R1 and R2 are very large and the
operational amplifiers have a gain of one, leads to the following
expressions for the output voltage.
|
|
|
Figure 3: Feedback circuit for a self-sensing damping
piezo. K(t) is an additional feedback control function which
for this discussion is taken to transmit Vs directly
back to V. The piezo is represented by its equivalent circuit shown
in the purple box. |
Here Vp is the voltage induced in
the piezo and Cp is the capacitance of the piezo. This
circuit tries to recreate the effect of the capacitance of the piezo in
the first branch and then subtracts the recreated signal from the signal
generated by the piezo to find the induced voltage. The key to this
design is matching the external capacitor Cr with the
internal capacitance of the piezo, Cp. If indeed
Cr is chosen so that
Cr=Cp then subtracting
V1 and V2 from equations 1 and 2
yields:
The differential voltage, Vs is
directly proportional to the voltage induced in the piezo.
Similar external circuits have also been designed to
calculate the strain rate, dVp/dt [2], yet most of these
methods rely on accurately matching the internal capacitance of the piezo.
In practice this is very difficult to achieve, mainly because
Cp is highly dependent on the environment and operating
conditions. For example, operating the piezo in an external field may
drastically change it's intrinsic behavior. In any situation where a
feedback circuit is operated in a closed loop, meaning the output is fed
back to the input, small errors in the output signal may excite rather
than damp the system and can quickly snowball and destabilize it. Either
extremely accurate knowledge of the internal capacitance or a dynamic
adaptive feedback circuit is required for successful operation of closed
loop feedback.
Shunted Piezos
|
Figure 4: RL shunted piezo. |
A third scheme first proposed by R. L. Forward
involves using external electric shunt circuits connected to the piezo to
quickly damp mechanical vibrations in the structure [3]. Shunting the
piezo is a passive technique, so it avoids the problem of having
destabilizing errors build up in a feedback loop. The piezo changes the
mechanical energy into electrical energy which diverted into a resonant LC
circuit and electrically dissipated. An common shunt circuit is shown in
figure 4, which includes an external resistor in addition the LC resonator
formed by the internal capacitance of the piezo, Cp and
an external inductor, L [4]. The inductor is selected to damp a
specific vibrational mode with frequency ωi.
The same problem of knowing the internal capacitance
of the piezo, Cp comes up again; however, in this case
an error results in a slightly different vibrational mode being damped
rather than a destabilization of the circuit. The main drawback to using
an RL shunt is that the shunt is only able to damp one mode. To introduce
further vibration control one must add more passive elements.
One possible technique, proposed by Wu et al. uses a parallel RL
shunt for each damped mode wired in series with a blocking LC
circuit[5-6]. The circuit diagram for two mode damping is shown in figure
5.
In a parallel LC circuit the inductor passes low
frequencies and the capacitor passes high frequencies, only the resonance
frequency of the circuit, ω=1/(LC)1/2, will be
blocked by the filter. To damp two modes in the base structure,
ω1 and ω2,
L1Cp is tuned to ω1
and L2Cp is tuned to
ω2, these circuits will directly shunt the
respective modes. In the blocking circuits,
L1Cp* is tuned to
ω2 while
L2Cp* is tuned to
ω1. This prevents the shunts from being exposed
to modes that are damped elsewhere in the circuit.
|
Figure 5: Equivalent impedance for two mode
shunt damping circuit. |
Wu's design is an elegant way to add more modes
without retuning each previous branch, but the method does not scale well.
For every mode that is shunted with an RL circuit that mode must be
blocked on all other branches. To damp three modes requires six blocking
LC circuits in addition to the three RL shunts, requiring
2n2 passive elements to damp n modes. To
complicate things even further damping very low frequencies requires
fairly large inductors on the order of tens of henrys. These inductors
are often replaced by gyrator circuits that use an operational amplifier,
two resistors and a capacitor to accurately simulate an inductor.
Gyrator circuits are often cheaper, less bulky and more reliable than very
large inductors. If gyrator circuits are used to replace the inductors
then 4n2-2n operational amplifiers are required to damp
n modes. Operational amplifiers are active rather than passive
elements and require external power. Replacing inductors with gyrator
circuits is often necessary, but it makes scaling the circuit to damp more
modes even more difficult.
Many other passive shunt circuits have been designed
but all suffer from the same scaling problem [7-8]. They work well if the
piezo element is used to damp a small number of vibrational modes;
however, they do not scale well and therefore cannot be used to damp
multiple vibrational modes. To address a large number of vibrational modes
other solutions must be sought.
Shunted Feedback Circuits
This paper has covered using a post-processed signal
to feedback and control the motion of the piezo and applying external
passive shunt to electronically damp vibrations of the piezo. Both have
been shown to exhibit debilitating problems, especially when applied to a
large number of modes. A third solution combines these two methods by
using feedback to digitally generate an impedance which dynamically shunts
the piezo. A dynamic impedance eliminates the problem of having to
address all vibrational modes individually, and feeding back on an
external impedance circuit rather than the piezo itself is more stable.
A schematic of a feedback circuit proposed and implemented by Moheimani
et. al is shown in figure 6B [9]. In this implementation the
admittance Y(t) is used as the control for the circuit.
Gvv(t) is the transfer functions that determines how
voltage applied to the piezo transfers to motion in the structure is
related to the voltage the piezo senses, and the transfer function
Gvw(t) determines how a vibration in the structure,
w, is related to the induced voltage. Both of these transfer
functions affect the total induced voltage in the piezo,
Vp(t). Cp is the capacitance of the
piezo and V(t) is the voltage dropped across the admittance
element. The following expressions describe the behavior of the circuit
in figure 6A.
Two individual feedback loops within
the circuit feed back using the voltage induced in the piezo and the
voltage across the admittance. The challenge of designing this feedback
circuit is determining an algorithm for the admittance. Many possible
solutions exists and each must be optimized to the particular
piezo/structure coupling to achieve peak performance. The admittance
suggested by Moheimani et. al has been shown to reduce mechanical
resonances by 20 dB [9].
|
Figure 6: A) Schematic of a piezo feedback circuit
where the control is the admittance of the shunt circuit. B) Lump
diagram of the feedback loop assuming the admittance,
Y(t) is controlled. |
|
|
Summary
This overview of piezoelectric elements
used in vibration damping circuits has covered three basic damping designs
including self sensing feedback, passive shunts, and shunted feedback. The brevity
and introductory nature of this paper cannot due full justice to the technical
complexity as well as the variety of work done in this field. The interested
reader is encouraged to refer to the references within this paper to gain a more
comprehensive view of the subject.
© 2007 J. Bert. The author grants permission to
copy, distribute and display this work in unaltered form, with attribution
to the author, for noncommercial purposes only. All other rights,
including commercial rights, are reserved to the author.
References
[1] S. O. R. Moheimani, IEEE TCST 11,
482 (2003).
[2] J. J. Dosch, D. J. Inman, and E. Garcia, J. Intel.
Mat. System and Structures 3, 166 (1992).
[3] R. L. Forward, J. Appl. Opt. 18,
690 (1979).
[4] N. W. Hagood and A. von Flotow, J. Sound
Vibr. 146, 243 (1991).
[5] S. Y. Wu, US Patent 5 783 898 (1998).
[6] S. Y. Wu, Proc. SPIE Symp. Smart Strctures
Materials Passive Damping Isolation 3327, 159 (1998).
[7] J. J. Hollkamp, J. Intel. Mat. System and
Structures 5, 49 (1994).
[8] S. Behens and S. O. R. Moheimani,
Proc. SPIE Symp. Smart Strctures Materials Damping Isolation 4697,
217 (2002).
[9] S. O. R. Moheimani, A. J. Fleming, and S. Behens,
Electron. Lett. 18, 442 (2001).