You are on page 1of 9

464 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 59, NO.

1, JANUARY 2012
Switching Delay Effects on Nonlinear
Piezoelectric Energy Harvesting Techniques
Mickal Lallart, Yi-Chieh Wu, and Daniel Guyomar
AbstractEnergy harvesting using piezoelectric elements re-
ceived much attention as vibrations are widely available and
as piezoelectric transducers feature high-power densities and
promising integration potentials. It has also been shown that
applying a nonlinear treatment on the output voltage of the piezo-
electric material can signicantly enhance the performance of the
device. This process consists of inverting the piezoelectric voltage
when the displacement is maximum, which therefore requires a
way of synchronization. In practical applications, however, a delay
may happen between the inversion and the actual occurrence of
an extremum. The purpose of this paper is to investigate the effect
of such a delay on the microgenerator performance and therefore
to predict the power output that can be expected under real cir-
cumstances. Theoretical analysis validated through experimental
measurements shows that the effect may not be the same for
positive or negative delays. It is also demonstrated that the effect
is not signicant as long as the delay is small. The acceptable delay
range also increases as the electromechanical system becomes
more coupled and/or less damped. Under such conguration, the
output power can even be slightly increased as the delay permits
controlling the tradeoff between energy extraction and damping
effect.
Index TermsEnergy conversion, energy harvesting, piezoelec-
tric devices.
I. INTRODUCTION
B
ATTERIES, which initially promoted the development of
portable electronic devices, have paradoxically become
a brake to this growth, particularly because of associated
maintenance and environmental issues (charging, replacement,
recycling process, and lifespan [1]). Hence, alternative power
sources obtained from ambient environment have recently
grasped peoples interest, particularly for powering up small-
scale electronic devices such as wireless sensors and sensor
networks [2][5]. As there are numerous potential energy
sources in daily life, identifying the suitable sources becomes
an important issue. For small-scale devices, a particular em-
phasis has been placed on systems scavenging the energy from
ambient sources such as heat, magnetic eld, or photons [1],
[6][9]. The works of Paradiso et al. [7] and Roundy et al. [1]
summarize the various sources of energy available and show
Manuscript received September 2, 2010; revised February 22, 2011; accepted
April 14, 2011. Date of publication April 29, 2011; date of current version
October 4, 2011.
The authors are with the Laboratoire de Gnie Electrique et Ferrolectricit
(LGEF), INSA-Lyon, Universit de Lyon, 69621 Villeurbanne, France (e-mail:
mickael.lallart@insa-lyon.fr; yi-chieh.wu@insa-lyon.fr; daniel.guyomar@
insa-lyon.fr).
Color versions of one or more of the gures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identier 10.1109/TIE.2011.2148675
the efciency of vibration energy conversion devices for small-
scale applications. In particular, using piezoelectric materials
to convert mechanical energy into electrical energy has been
intensively discussed over the last few years [10][17], as such
materials present high energy densities as well as promising
integration potentials.
In the domain of vibration energy conversion using
piezoelectric elements, nonlinear techniques introduced by
Guyomar, Richard et al. [18][25] have been proved to be
an efcient way for harvesting energy from vibrations. These
techniques, called synchornized switch harvesting on inductor
(SSHI), consist of quickly reversing the piezoelectric voltage
synchronously with the structure motion, using a switching
device and an inductor with the standard energy extracting
interface (the latter consisting of an acdc rectier and smooth-
ing/storage capacitor). The switch is set to turn on maximum
and minimum displacements. Such a treatment therefore intro-
duces a piecewise constant voltage which is proportional to the
sign of velocity and increases the piezoelectric voltage, leading
to a magnication of the conversion abilities and thus harvested
energy. In addition to energy conversion magnication, a strong
advantage of such techniques is their ability to be self-powered
[26], [27], using a very few amount of energy (typically less
than 3% of the electrostatic energy available on the piezoelec-
tric element).
Although the nature of the switching process allows wide-
frequency-range operations, owing to the maximum detection,
the necessity of synchronizing the switch with maximum value
detection would induce a phase delay (or a corresponding
time delay ) between the actual occurrence of a displacement
extremum and the instant of switching in real applications. This
delay would therefore limit the cumulative voltage magnica-
tion effect as well as shifting the piecewise constant voltage that
would consequently not be exactly proportional to the speed,
hence limiting the conversion enhancement. The fact that the
piecewise constant function is not exactly proportional to the
speed would also induce a stiffening effect [28].
Hence, the purpose of this paper aims at investigating this
switching delay effect on energy harvesting performance of
microgenerators based on this nonlinear process [19]. Theo-
retical model will be developed and experimental validation
assessed for both parallel SSHI and series SSHI techniques.
Theoretical predictions conrmed by experimental measure-
ments show that the switching delay effect is very limited
for a specic delay range which is actually increasing as the
electromechanical coupling becomes higher and/or the mechan-
ical losses decreases, even possibly outperforming the ideal
case (no delay) in the case of highly coupled weakly damped
0278-0046/$26.00 2011 IEEE
LALLART et al.: SWITCHING DELAY EFFECTS ON PIEZOELECTRIC ENERGY HARVESTING TECHNIQUES 465
Fig. 1. Equivalent mechanical model for piezoelectric structure.
electromechanical structures. It will also be shown that for the
parallel SSHI, negative delay has more impact than a positive
one on the microgenerator performance, while in the case of
the series conguration, the delay effect is symmetric and close
to the negative delay inuence in the case of the parallel SSHI
and is therefore more sensitive to a delay. The change in the
optimal load that maximizes the harvested power will also be
investigated, showing that it tends to increase for the parallel
SSHI and to decrease for the series SSHI, as the switching
delay increases. Finally, it will be demonstrated that the delay
inuence becomes negligible as the gure of merit given by
the product of the squared global coupling coefcient by the
mechanical quality factor increases, as the decrease of the
energy conversion abilities is balanced by a weaker damping
effect that permits a higher input mechanical energy.
This paper is organized as follows. Section II exposes the
principles of energy harvesting circuits and nonlinear process-
ing. Section III proposes a comprehensive view of the delay
origins as well as a theoretical development of energy harvest-
ing with nonlinear techniques considering the switching delay
and its effect on the performance of the microgenerator and
optimal operating points for several congurations. Section IV
presents results from experimental measurements as well as the
comparison between theoretical predictions and experimental
results. Finally, Section V briey concludes the paper.
II. ENERGY HARVESTING BASICS
In this section, the basic energy harvesting devices will be
presented. To simplify explanation, we consider that the system
is excited by a sinusoidal force at the resonant frequency and
that the harvested power expression is corresponding to average
power.
The considered energy harvesting device consists of a vibrat-
ing mechanical structure with bonded piezoelectric elements.
The structure with piezoelectric elements is modeled as a
{spring+mass+damper+piezo} system such as the one shown in
Fig. 1 [29], [30]. The energy harvesting system is composed by
a rigid mass M, a spring K
s
(representing the stiffness of the
host structure), and a viscous damper C (mechanical losses).
The external force F is applied on the rigid mass M.
According to Newtons second law of motion, the sum of
the force on the rigid mass leads to the governing (1) of the
Fig. 2. Standard interface circuit.
{spring+mass+damper+piezo} system
M u =F
T
K
E
u C u
F
T
=F V (1)
where F
T
is the total force exerted on the mass (external force
F and piezoelectric force), K
E
is the global stiffness in short-
circuit condition, and the force factor of the piezoelectric
element. u and V denote the structure displacement and output
voltage of the piezoelectric element, respectively.
The time domain integral of the motion equation multiplied
by the velocity expresses the energy balance of the system
t
0
+t
_
t
0
F udt =
1
2
M[ u
2
]
t
0
+t
t
0
+
1
2
K
E
[u
2
]
t
0
+t
t
0
+C
t
0
+t
_
t
0
u
2
dt +
t
0
+t
_
t
0
V udt (2)
giving the mechanical energy provided to the system and the
sum of the kinetic energy, potential elastic energy, mechanical
losses, and transferred energy over the considered time range
[t
0
; t
0
+ t].
In addition to the mechanical equation, it is also possible
from the constitutive equations of piezoelectricity to get the
expression of the current I owing out of the piezoelectric
element as [29], [30]
I = u C
0

V (3)
where C
0
is the clamped capacitance of the piezoelectric ele-
ment. From (3), it can be shown that the converted energy is the
sum of electrostatic energy stored on the piezoelectric element
plus the energy delivered to the connected electrical device

t
0
+t
_
t
0
V udt =
1
2
C
0
[V
2
]
t
0
+t
t
0
+
t
0
+t
_
t
0
V Idt. (4)
A. Standard Approach
The standard interface circuit (Fig. 2) is the most common
way for harvesting energy. It includes a diode rectier bridge
and a smoothing capacitor C
r
. The electrical energy is simply
provided to the storage capacitor through the acdc power con-
verter. The terminal electric load is modeled by an equivalent
resistor R
L
. When the absolute value of V is lower than the
storage cell voltage V
DC
,
1
the rectier bridge is blocked and the
1
Considered as constant as the product of R
L
and C
r
is much larger than
half a vibration period T/2.
466 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 59, NO. 1, JANUARY 2012
Fig. 3. SSHI interface circuits: (a) Parallel SSHI and (b) Series SSHI.
piezoelectric element is in open condition. When the absolute
value of V reaches the storage cell voltage V
DC
, energy is
transferred from the piezoelectric element to the storage ca-
pacitor through the diode rectier bridge. The bridge rectier
stops conducting when the absolute value of the displacement
u decreases (current cancellation).
When considering sine excitation, it can be shown that the
power expression yields [18], [19]
P =
V
2
DC
R
L
=
R
L

2
_
R
L

r
C
0
+

2
_
2

2
r
u
2
M
(5)
where
r
denotes the angular resonance frequency and u
M
the
displacement magnitude.
However, extracting electrical energy from vibrations re-
duces the amount of mechanical energy in the structure, hence
leading to damping effect. From the energy analysis of the
system, it can be shown that the displacement magnitude at the
resonance as a function of the applied force magnitude F
M
is
given by [18]
u
M
=
F
M
C
r
+
2R
L

2
(R
L
C
0

r
+

2
)
2
. (6)
B. SSHI Approaches
From the electromechanical energy conversion analysis (4),
the average power P
conv
converted by the piezoelectric element
yields
P
conv
=
1
T
t
0
+T
_
t
0
V udt. (7)
This power depends on two factors, the time shift between
piezoelectric voltage V and the mechanical velocity u and
the voltage amplitude. Hence, to achieve a better power out-
put performance, the phase shift between V and u has to
be reduced, and, coincidentally, the voltage amplitude needs
to be increased. Besides, the converting process should have
minimum input energy.
To optimize average converted power, the so-called SSHI
approach has recently been investigated. The parallel SSHI [18]
is composed of a nonlinear processing circuit connected in
parallel with the standard interface described in Section II-A
and shown in Fig. 3(a). The nonlinear processing circuit is
composed of an inductor L in series with a digital switch S.
When the mechanical displacement reaches a minimum or a
maximum value, the switch is turned on. At this time instant,
an oscillating L C
0
circuit is established. The switch is kept
closed until the voltage on the piezoelectric element has been
reversed. As the inversion frequency is far greater than the
vibration frequency, the inversion process may be considered
as instantaneous. Inversion losses, however, exist during the
switching process, leading to an imperfect voltage inversion,
characterized by the inversion coefcient , giving the absolute
voltage ratio after and before the process.
The energy harvesting process is conducted before this volt-
age inversion period. Previous works [18], [19] showed that
the rectied voltage, harvested power, and the displacement
magnitude using the parallel SSHI are given by
V
DC
=
2R
L

R
L
C
0
(1 )
r
+

r
u
M
, (8)
P =
V
2
DC
R
L
=
4R
L

2
(R
L
C
0
(1 )
r
+)
2

2
r
u
2
M
, (9)
u
M
=
F
M
C
r
+
4R
L

R
L
C
0
(1
2
)
r
+2
[R
L
C
0
(1)
r
+]
2
(10)
typically allowing harvesting eight times more energy than in
the standard case [18], [19].
Instead of connecting the nonlinear processing circuit in
parallel with the piezoelectric insert and the rectier bridge,
the circuit may be connected in series as shown in Fig. 3(b),
constituting the series SSHI interface. In this case, the energy
harvesting process occurs as the switch turns on, in the same
time than the inversion process. The expression of the rectied
voltage V
DC
, average harvested power, and displacement mag-
nitude are then given by [19]
V
DC
=
2R
L
(1 +)
2R
L
C
0

r
(1 +) +(1 )

r
u
M
(11)
P =
4R
L

2
(1 +)
2
[2R
L
C
0

r
(1 +) +(1 )]
2

2
r
u
2
M
(12)
u
M
=
F
M
C
r
+
4
2
C
0
1+
2R
L
C
0
(1+)
r
+(1)
(13)
allowing an energy harvesting gain of approximately seven [19]
when using typical off-the-shelf components.
III. SWITCHING DELAY EFFECT
In the case of the SSHI approaches exposed in Section II, the
switch is assumed to perfectly turn on as the optimal displace-
ment occurs (minimum or maximum value). However, due to
non-perfect detection in practical implementation [26], [27], a
phase delay or the corresponding time delay exists between
the actual occurrence of an extremum displacement and the
instant the switch turns on. In a mechanical point of view,
there will be both a stiffness change [28] and a damping effect
[31] according to the phase delay. In the following section, the
detailed modeling of the effect of phase delay for parallel SSHI
and series SSHI will be proposed. It is, however, assumed that
the stiffness change is negligible so that the structure is still
excited at its resonance frequency.
2
2
Even in this case, the resonance frequency is comprised between the short-
and open-circuit resonance frequencies which are close to each other.
LALLART et al.: SWITCHING DELAY EFFECTS ON PIEZOELECTRIC ENERGY HARVESTING TECHNIQUES 467
Fig. 4. Implementation examples for the maximum detection.
Fig. 5. Self-powered switch.
A. Switching Delay Origins
Before investigating the effect of a delay in the switching
process, this section proposes to discuss the origin of this delay
in practical implementations. In Fig. 4, two examples for de-
tecting maximum signal are proposed. The rst example is the
simplied block diagramm of the actual self-powered switching
device proposed in [26], [27], depicted in Fig. 5, and constitutes
the most interesting solution as it is reliable and robust and can
even use the signal from the transducer itself to operate (i.e.,
no additional sensor is needed). However, because of the need
of synchronization, the switch signal command is not perfectly
generated on maximum and minimum displacement values.
As it can be seen in Fig. 4, the two proposed implementation
induce delay (positive for the rst one and negative for the
second one). This delay is mainly caused by the voltage gap of
discrete components (diodes and transistorssee Fig. 5) that
perform the detection and the switch command, and therefore
depends on the output voltage level of the transducer. Hence,
for low voltage outputs, the delay would be greater, as voltage
thresholds are constant. This would therefore lead to a lim-
ited cumulative voltage magnication (as the voltage inversion
would not be performed on maximum voltage) and a piecewise
Fig. 6. Waveforms of parallel SSHI technique with (a) negative switching
delay and (b) positive switching delay.
constant function no longer exactly proportional to the sign
of the speed. Both of these effects would therefore limit the
conversion enhancement as shown by Eq. (7).
In addition, it can be noted that, under broadband vibra-
tions, the performance of the SSHI may be degraded when
switching on each extremum, as a tradeoff between the number
of switching events and voltage increase appears. Therefore,
a frequency ltering has to be implemented to ensure that
the switching process occurs on global extrema. This ltering
however introduces an additional delay.
B. Parallel SSHI Approach
For the parallel SSHI technique, negative and positive de-
lays of switching occurrence will have different effects on the
system, as a positive switching delay ( and greater than
zero) would lead to an extra open-circuit state. Accordingly,
the section below will describe the switching delay effect for
negative and positive delays separately. Here, the negative delay
is dened as the case where the switch turns on before the
optimal displacement occurs with a specic error [Fig. 6(a)],
and contrarily, the case where the switch turns on after the
optimal displacement happens is dened as positive switching
delay [Fig. 6(b)].
1) Negative Switching Delay Effect: With a negative switch-
ing delay , the nonlinear process occurs before reaching
maximum or minimum displacement. As shown in Fig. 6(a),
the piezoelectric output voltage increases or decreases with
the mechanical displacement before reaching the extremum
displacement.
The displacement u

when the piezovoltage reaches V


DC
could be derived from the proportional relationship between
displacement and voltage when the system is in open-circuit
condition (I = 0)
V
DC
=

C
0
u

+B (14)
where B is an integration constant. From the equations of the
inversion and open-circuit condition
_
V
DC
=

C
0
u

+B
V
m
=

C
0
u
M
cos +B = V
DC
(15)
it is therefore possible to express the displacement value u

when the harvesting process starts


u

=
C
0

(1 )V
DC
u
M
cos (16)
468 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 59, NO. 1, JANUARY 2012
yielding the harvested energy over a single phase
E =
t
2
_
t

V
DC
Idt = V
DC
(u
M
cos u

). (17)
As the energy may also be expressed as E = (V
2
DC
/R
L
)
(T/2), it can be shown that the rectied voltage and harvested
power expressions yield
V
DC
=
2R
L

+R
L

r
C
0
(1 )
u
M
cos (18)
P =
4R
L

2
r

2
[ +R
L

r
C
0
(1 )]
2
u
2
M
cos
2
. (19)
The simplied expression of overall energy balance over half
a vibration period, given by
T
4
F
M
u
M

r
=C
2
r
u
2
M
T
4
+
1
2
C
0
V
2
DC
(1
2
)+
V
2
DC
R
L
T
2
(20)
allows the derivation of the displacement magnitude as a func-
tion of the input force
u
M
=
F
M
C
r
+
4R
L

_
R
L
C
0
(1
2
)
r
+2
(R
L

r
C
0
(1)+)
2
_
cos
2

. (21)
2) Positive Switch Delay Effect: For the positive switch
delay, the switch turns on after the extremum displacement
occurs [Fig. 6(b)]. As the inversion happens when the absolute
displacement is below its maximum value, the correspond-
ing waveform would be a combination of open-circuit state
(from b to c), energy harvesting process (from c to d), open-
circuit state (d to e), and inversion process (a to b). The effect
described here is therefore different from the case of negative
switching delay.
The voltage V
M
after the switch turns off has a correlation
with its corresponding displacement u
M
cos(), and so for the
rectied voltage V
DC
, yielding
_
V
DC
=

C
0
u
M
+D
V
M
=

C
0
u
M
cos +D
(22)
where D is an integration constant.
From period d to e, the system is in open-circuit condition,
and the output voltage changes with the displacement as the
current is null
V =

C
0
u +D

(23)
where D

is another integration constant dened by


V
DC
=

C
0
u
M
+D

. (24)
Combined with the expression of the voltage V
m
after the
inversion process
V
m
=

C
0
u
M
cos +D

= V
M
(25)
it can be shown, using a similar analysis than in the previous
section (considering that the harvesting time range is [t

; t

]),
Fig. 7. Waveforms of series SSHI technique with (a) negative switching delay
and (b) positive switching delay.
that the expressions of the rectied voltage and harvested power
yield
V
DC
=
R
L

r
[1 + cos (1 cos )]
+R
L

r
C
0
(1 )
u
M
(26)
P =
R
L

2
r

2
[1 + cos (1 cos )]
2
[ +R
L

r
C
0
(1 )]
2
u
2
M
(27)
and the value of the displacement magnitude, obtained from the
energy analysis over half a vibration period, is given by
u
M
=
F
M
C
r
+
2A
2
R
L

r
+
C
0

_
A+

C
0
(cos 1)
_
2
(1
2
)
with A=
R
L

r
(1+cos (1cos ))
+R
L

r
C
0
(1)
. (28)
C. Series SSHI Approach
The analysis of the series SSHI approach with a switching
delay is the same than in the case of the ideal series SSHI
technique, although the practical nonlinear device turns on with
a phase delay with respect to the occurrence of an extremum
displacement. Hence, for the series SSHI technique, the theo-
retical expressions are derived from the same analysis for both
negative and positive switching delays, shown in Fig. 7(a) and
7(b), respectively. The corresponding displacement when the
switch turns on equals u
M
cos().
The absolute piezoelectric voltages when the energy har-
vesting process starts (V
M
) and after the inversion process
(V
m
) vary with the displacement u
M
cos() plus an integration
constant H, as the piezoelectric element is left in open-circuit
condition almost all the time. Combined with the inversion
equation, this leads to the equation set
_
_
_
V
M
=

C
0
u
M
cos +H
V
m
=

C
0
u
M
cos +H
(V
M
V
DC
) = V
m
+V
DC
(29)
giving the expression of the energy harvested over a single event
E = C
0
V
DC
(V
M
+V
m
) (30)
allowing the derivation of the rectied voltage V
DC
as a func-
tion of the connected load
V
DC
=
2R
L

r
(1 +)
(1 ) + 2C
0
R
L

r
(1 +)
u
M
cos (31)
LALLART et al.: SWITCHING DELAY EFFECTS ON PIEZOELECTRIC ENERGY HARVESTING TECHNIQUES 469
Fig. 8. Performance comparison of standard, parallel SSHI, and series
SSHI techniques for different coupling factors, -: parallel SSHI technique,
: series SSHI technique: (a) k
2
Q
M
= 0.1877; (b) k
2
Q
M
= 0.4628;
(c) k
2
Q
M
= 0.9046; (d) k
2
Q
M
= 1.7306; and (e) k
2
Q
M
= 3.1845.
and leading to the harvested power expression
P =
4R
L
(
r
(1 +))
2
((1 ) + 2C
0
R
L

r
(1 +))
2
(u
M
cos )
2
. (32)
As previously, the expression of the displacement magnitude
as a function of the force magnitude (showing the damping
effect) can be obtained from the energy analysis over half a
vibration period
T
4
F
M
u
M

r
= C
2
r
u
2
M
T
4
+
1
2
C
0
_
1 +
1
_

_
2
C
0
u
M
cos 2V
DC
__
2
C
0
u
M
cos
_
(33)
yielding
u
M
=
F
M
C
r
+
4

_
1+
1
_
cos
2

_

C
0
A

_
with A

=
2R
L

r
(1 +)
(1 ) + 2C
0
R
L

r
(1 +)
. (34)
D. Theoretical Comparison
The harvested power difference between the delayed and
ideal SSHI approaches in relation with the switching delay (nor-
malized with respect to vibration period) for ve different cases,
are shown in Fig. 8 for both parallel SSHI and series SSHI
techniques. These charts are indexed with respect to the gure
Fig. 9. Power as a function of the load and switching delay for parallel
and series SSHI technique, k
2
Q
M
= 0.1877 (a) parallel SSHI technique and
(b) series SSHI technique.
Fig. 10. Power as a function of the load and switching delay for parallel
and series SSHI technique, k
2
Q
M
= 3.1845 (a) parallel SSHI technique and
(b) series SSHI technique.
of merit given by the product k
2
Q
M
of the mechanical quality
factor Q
M
, reecting the available mechanical energy in the
structure, by the global electromechanical coupling coefcient
k
2
, which gives the effective amount of available energy that
can be converted into electrical energy.
As k
2
Q
M
increases, the curve becomes more at and shows
that the harvested power is the same than in the ideal case.
This could be explained by the tradeoff between damping effect
and converted energy. As the switching delay increases, the
converted power decreases and under a constant driving force,
this leads to the higher displacement amplitude as the damping
decreases as well. For a highly coupled structure, this increased
displacement will induce an increase in the available mechan-
ical energy, resulting in a higher harvested power. Hence, the
increase of the available mechanical counteracts the decrease
in conversion efciency.
These results show the effectiveness of SSHI techniques
within a specic switching delay, particularly for parallel SSHI
technique with positive delay switching. It is also interesting
to note that, after a critical value of k
2
Q
M
, the parallel SSHI
technique with a delay may exhibit a better performance than in
the no-delay case for a positive delay value. From the gures,
it also shows that the series SSHI is more sensitive to the
switching delay, although the difference with the parallel SSHI
in the case of a negative delay is rather small for low and high
k
2
Q
M
. Finally, it can be noted that, contrary to the series SSHI,
the parallel SSHI features a non-symmetric behavior, leading to
a different effect on the harvested power whether the delay is
negative or positive.
The top view of 3-D surface plots (Figs. 9 and 10) show the
harvested power as a function of resistance load (normalized
with respect to the optimal load value in the standard case when
no damping effect is considered [18]) and the switching delay
(in the ratio of a vibration period) for two values of the gure
470 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 59, NO. 1, JANUARY 2012
Fig. 11. Experimental structures: (a) relatively weakly coupled structure and
(b) relatively highly coupled structure.
of merit k
2
Q
M
. The optimal resistance load tends to toward
innity for parallel SSHI technique and tends to zero for series
SSHI technique as the absolute switching delay increases (par-
ticularly for highly coupled, weakly damped structuresi.e.,
high k
2
Q
M
) due to the different harvesting processes. Such
an effect can be explained by the fact that, as the switching
delay increases, the converted energy will decrease. Hence,
for the parallel SSHI technique, the resistance load has to be
increased to have a higher V
DC
to increase converted energy
(higher piezovoltage in phase with the speed). In the case of
series SSHI, the piezovoltage shift and increase (denoting the
conversion enhancement) is obtained by reducing the rectied
voltage, and therefore the resistance load would approach to
zero. As previous, the parallel SSHI exhibits a nonsymmetric
behavior facing the delay, contrary to the series SSHI.
IV. EXPERIMENTAL VALIDATION AND DISCUSSION
This section aims at validating the theoretical predictions
presented in the previous section through experimental mea-
surements made on simple electromechanical structures, con-
sisting of cantilever beams.
A. Experimental Setup
In order to validate the switching delay effect with the
theoretical results for both weakly coupled and highly coupled
systems, two structures with different k
2
Q
M
are used. The
experimental test structures are composed of a 40- and a 60-mm
long cantilever beams clamped at one end to a rigid structure
with bonded piezoelectric inserts on their surface (Fig. 11).
These two structures are taken as the relatively weakly coupled
and highly coupled devices.
The beams are driven at their rst resonance frequency by
an electromagnet which is controlled by a dSPACE control box
and ControlDesk interface. This digital signal processor is also
used for driving the digital switches (and selecting the delay),
although these latter can be made truly self-powered, using
a negligible amount of energy available on the piezoelectric
element [27], [32]. The driving force amplitude are 0.0031 and
0.0047 N for the weakly coupled and highly coupled struc-
tures, respectively. A displacement sensor with a sensitivity of
8 mm/V is used to measure the displacement amplitude. With-
out control and at the resonance frequency, the displacement
amplitudes close to the free ends are 2.9 mm for the rst beam
and 0.84 mm for the second. The system parameters, obtained
using the same procedure than in [31], are listed in Table I.
TABLE I
PARAMETER IDENTIFICATION FOR EXPERIMENTAL MEASUREMENTS
Fig. 12. Theoretical and experimental comparison for weakly coupled struc-
ture: (a) parallel SSHI technique and (b) series SSHI technique.
Fig. 13. Theoretical and experimental comparison for highly coupled struc-
ture: (a) parallel SSHI technique and (b) series SSHI technique.
The storage capacitance value is set to 2.2 F and load resistor
R
L
is made varying in order to change the value of the voltage
V
DC
, which is monitored using a multimeter, thus allowing the
derivation of the harvested power (P = V
2
DC
/R
L
).
B. Results and Discussion
Experimental results (Figs. 12 and 13) show a good agree-
ment with the predictions given by the model. The structures
used in the experiment could have a harvested power from
0.16 mW 0.2 mW for case 1 and 0.3 mW 0.33 mW for
case 2. The harvested power is reduced because the switching
delay induces a piecewise voltage that is reduced and no longer
exactly proportional to the sign of the velocity. As expected, the
series SSHI technique is more sensitive to the switching delay
in the case of a positive one, so that the curve of optimum power
is sharper than in the case of the parallel SSHI.
For a weakly coupled structure (case 1), a range of 0.15
0.15 delay ratio induces a respective harvested power drop of
LALLART et al.: SWITCHING DELAY EFFECTS ON PIEZOELECTRIC ENERGY HARVESTING TECHNIQUES 471
TABLE II
DELAY RANGE (IN RATIO OF VIBRATION PERIOD)
FOR A 50% POWER DROP
0.1 mW (50% less power) and 0.07 mW (35% less power) for
parallel SSHI technique and 0.08 mW for series SSHI (50%
less power) for both negative and positive delay. In the case of a
highly coupled structure, the same range of delay ratio induces
a harvested power drop of 0.09 mW (27% less), 0.03 mW (9%
less) for parallel SSHI technique with negative and positive
switching delay, respectively, and 0.09 mW for series SSHI.
As expected, the increase of k
2
Q
M
will increase the accepted
range of switching delay to get the same harvested power per-
formance (Table II). For parallel SSHI technique with a positive
switching delay, the enhancement of accepted range of switch-
ing delay is larger than with a negative delay. For a positive
delay range of 0.1 period ratio, the decrease of harvested power
is only 0.005 mW. Therefore, implementing the parallel SSHI
with an approach that induces positive delay is more effective.
V. CONCLUSION
This paper investigated the effect of a switching delay of
the nonlinear SSHI techniques. Both theoretical modeling and
experimental validation have been performed, showing that the
parallel SSHI approach features a non-symmetric behavior as a
function of the delay, contrary to the series SSHI technique.
From the theoretical predictions, it is shown that to have the
same harvested power, parallel SSHI technique with positive
switching delay permits a larger delay range. With the increase
of the gure of merit given by the product of the squared
coupling factor k
2
and mechanical quality factor Q
M
, the curve
of harvested power is more at and the accepted delay range
is enhanced. For a low-coupled structure (k
2
Q
M
= 0.1877)
within the delay range [0.1; 0.12] (in a ratio of a period),
the harvested power drops by 20% compared to the ideal case.
For a higher coupling factor, k
2
Q
M
= 3.1845, the delay range
increases to [0.19; 0.235] for the same performance, as the
conversion ability decrease is compensated by a higher me-
chanical energy caused by a weaker damping effect. Moreover,
because of the tradeoff between damping effect and converted
energy, parallel SSHI technique with positive switching delay
might have a slightly better performance than in the ideal case
(no delay).
Compared to the parallel SSHI technique, the series SSHI ap-
proach is more sensitive to a delay in the switching process, par-
ticularly for positive delays. Within the same range of switching
delay, the reduction of harvested power is greater in the series
SSHI case. For a weakly coupled system (k
2
Q
M
= 0.9046)
with the switching delay ranging from 0.1 to 0.1 period ratio,
the harvested power will decrease by 12.6% for the series
SSHI but less than 6% for the parallel SSHI. The series SSHI
technique, which is different from the parallel SSHI, harvests
power at the same instant as the inversion process for both
negative switching delay and positive switching delay, and has
a symmetric performance for negative delay and positive delay.
Both theoretical predictions and experimental validations
showed that nonlinear techniques applied under a specic
switching delay would have very limited reduction in harvested
power and remain very effective in energy scavenging from
vibrating structures equipped with piezoelectric inserts. It is
therefore concluded that the nonlinear techniques are not com-
promised in terms practical implementation. Another point is
that, in the case of the parallel SSHI, the implementation of
the maximum and minimum detection should induce a positive
delay for better performance.
REFERENCES
[1] S. Roundy, P. K. Wright, and J. Rabaey (2003, Jul.). A study of
low level vibrations as a power source for wireless sensor nodes.
Comput. Commun. [Online]. 26(11), pp. 11311144, Ubiquitous Com-
puting. Available: http://www.sciencedirect.com/science/article/B6TYP-
47CWTY0-1/2/cf6278ee1732c2c1d225901b19270a10
[2] J. Baker, S. Roundy, and P. Wright, Alternative geometries for increasing
power density in vibration energy scavenging for wireless sensor net-
work, in Proc. 3rd Int. Energy Convers. Eng. Conf., San Francisco, CA,
Aug. 2005, pp. 959970.
[3] M. Lallart, D. Guyomar, Y. Jayet, L. Petit, E. Lefeuvre, T. Monnier,
P. Guy, and C. Richard (2008, Sep.). Synchronized switch harvesting
applied to self-powered smart systems: Piezoactive microgenerators for
autonomous wireless receivers. Sens. Actuators A, Phys. [Online]. 147(1),
pp. 263272. Available: http://www.sciencedirect.com/science/article/
B6THG-4SBHX0X-1/2/f50eff86a0007f31c07cbe724267d643
[4] V. Gungor and G. Hancke, Industrial wireless sensor networks: Chal-
lenges, design principles, and technical approaches, IEEE Trans. Ind.
Electron., vol. 56, no. 10, pp. 42584265, Oct. 2009.
[5] J. Rocha, L. Goncalves, P. Rocha, M. Silva, and S. Lanceros-Mendez,
Energy harvesting from piezoelectric materials fully integrated in
footwear, IEEE Trans. Ind. Electron., vol. 57, no. 3, pp. 813819,
Mar. 2010.
[6] N. G. Elvin, A. A. Elvin, and M. Spector (2001, Apr.). A self-powered
mechanical strain energy sensor. Smart Mater. Struct. [Online]. 10(2),
pp. 293299. Available: http://stacks.iop.org/0964-1726/10/i=2/a=314
[7] J. A. Paradiso and T. Starner, Energy scavenging for mobile and wire-
less electronics, IEEE Pervasive Comp., vol. 4, no. 1, pp. 1827,
Jan.Mar. 2005.
[8] A. Nasiri, S. Zabalawi, and G. Mandic, Indoor power harvesting using
photovoltaic cells for low-power applications, IEEE Trans. Ind. Elec-
tron., vol. 56, no. 11, pp. 45024509, Nov. 2009.
[9] P. Khaligh, A. Zeng, and C. Zheng, Kinetic energy harvesting using
piezoelectric and electromagnetic technologiesState of the art, IEEE
Trans. Ind. Electron., vol. 57, no. 3, pp. 850860, Mar. 2010.
[10] P. Glynne-Jones, S. Beeby, and N. White (2001, Mar.). Towards
a piezoelectric vibration-powered microgenerator. Proc. Inst. Elect.
Eng.Sci., Meas. Technol. [Online]. 148(2), pp. 6872. Available:
http://link.aip.org/link/?ISM/148/68/1
[11] G. Ottman, H. Hofmann, A. Bhatt, and G. Lesieutre, Adaptive piezoelec-
tric energy harvesting circuit for wireless remote power supply, IEEE
Trans. Power Electron., vol. 17, no. 5, pp. 669676, Sep. 2002.
[12] C. D. Richards, M. J. Anderson, D. F. Bahr, and R. F. Richards (2004,
May). Efciency of energy conversion for devices containing a piezoelec-
tric component. J. Micromech. Microeng. [Online]. 14(5), pp. 717721.
Available: http://stacks.iop.org/0960-1317/14/i=5/a=009
[13] F. Lu, H. P. Lee, and S. P. Lim, Modeling and analysis of micro piezoelec-
tric power generators for micro-electromechanical-systems applications,
Smart Mater. Struct., vol. 13, no. 1, pp. 5763, Feb. 2004.
[14] J. Carmo, L. Goncalves, and J. Correia, Thermoelectric microconverter
for energy harvesting systems, IEEE Trans. Ind. Electron., vol. 57, no. 3,
pp. 861867, Mar. 2010.
[15] W. Li, S. He, and S. Yu, Improving power density of a cantilever piezo-
electric power harvester through a curved l-shaped proof mass, IEEE
Trans. Ind. Electron., vol. 57, no. 3, pp. 868876, Mar. 2010.
[16] A. Tabesh and L. Frechette, A low-power stand-alone adaptive circuit
for harvesting energy from a piezoelectric micropower generator, IEEE
Trans. Ind. Electron., vol. 57, no. 3, pp. 840849, Mar. 2010.
[17] R. Dayal, S. Dwari, and L. Parsa, Design and implementation of a direct
AC-DC boost converter for low voltage energy harvesting, IEEE Trans.
Ind. Electron., vol. 58, no. 6, pp. 23872396, Jun. 2011.
472 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 59, NO. 1, JANUARY 2012
[18] D. Guyomar, A. Badel, E. Lefeuvre, and C. Richard, Toward energy har-
vesting using active materials and conversion improvement by nonlinear
processing, IEEE Trans. Ultrason., Ferroelectr., Freq. Control, vol. 52,
no. 4, pp. 584595, Apr. 2005.
[19] E. Lefeuvre, A. Badel, C. Richard, L. Petit, and D. Guyomar (2006, Feb.).
A comparison between several vibration-powered piezoelectric genera-
tors for standalone systems. Sens. Actuators A, Phys. [Online]. 126(2),
pp. 405416. Available: http://www.sciencedirect.com/science/article/
B6THG-4HTCTHJ-2/2/7cb0b88fafc922d3a2754b50e1429455
[20] K. Makihara, J. Onoda, and T. Miyakawa, Low energy dissipation elec-
tric circuit for energy harvesting, Smart Mater. Struct., vol. 15, no. 5,
pp. 14931498, Oct. 2006.
[21] Y. C. Shu, I. C. Lien, and W. J. Wu, An improved analysis of the SSHI
interface in piezoelectric energy harvesting, Smart Mater. Struct., vol. 16,
no. 6, pp. 22532264, Dec. 2007.
[22] L. Garbuio, M. Lallart, D. Guyomar, C. Richard, and D. Audigier, Me-
chanical energy harvester with ultralow threshold rectication based on
SSHI nonlinear technique, IEEE Trans. Ind. Electron., vol. 56, no. 4,
pp. 10481056, Apr. 2009.
[23] J. Liang and W.-H. Liao, An improved self-powered switching interface
for piezoelectric energy harvesting, in Proc. IEEE Int. Conf. Inf. Autom.,
2009, pp. 945950.
[24] J. Qiu, H. Jiang, H. Ji, and K. Zhu, Comparison between four piezoelec-
tric energy harvesting circuits, Frontiers Mech. Eng. China, vol. 4, no. 2,
pp. 153159, Jun. 2009.
[25] S. Mehraeen, S. Jagannathan, and K. Corzine, Energy harvesting
from vibration with alternate scavenging circuitry and tapered can-
tilever beam, IEEE Trans. Ind. Electron., vol. 57, no. 3, pp. 820830,
Mar. 2010.
[26] C. Richard, D. Guyomar, and E. Lefeuvre, Self-Powered Electronic
Breaker With Automatic Switching by Detecting Maxima or Minima of
Potential Difference Between Its Power Electrodes, Patent No. PCT/
FR2005/003000, Jul. 6, 2007.
[27] M. Lallart, E. Lefeuvre, C. Richard, and D. Guyomar, Self-powered
circuit for broadband, multimodal piezoelectric vibration control, Sens.
Actuators A, Phys., vol. 143, no. 2, pp. 277382, May 2008.
[28] D. Guyomar, M. Lallart, and T. Monnier, Stiffness tuning using a low-
cost semi-active nonlinear technique, IEEE/ASME Trans. Mechatron.,
vol. 13, no. 5, pp. 604607, Oct. 2008.
[29] A. Badel, M. Lagache, D. Guyomar, E. Lefeuvre, and C. Richard, Finite
element and simple lumped modeling for exural nonlinear semi-passive
damping, J. Intell. Mater. Syst. Struct., vol. 18, no. 7, pp. 727742,
Jul. 2007.
[30] A. Erturk and D. J. Inman, Issues in mathematical modeling of piezoelec-
tric energy harvesters, Smart Mater. Struct., vol. 17, no. 6, p. 065 016,
Dec. 2008.
[31] A. Badel, G. Sebald, D. Guyomar, M. Lallart, E. Lefeuvre, C. Richard,
and J. Qiu, Piezoelectric vibration control by synchronized switching
on adaptive voltage sources: Towards wideband semi-active damping,
J. Acoust. Soc. Amer., vol. 119, no. 5, pp. 28152825, May 2006.
[32] M. Lallart and D. Guyomar (2008, Jun.). An optimized self-powered
switching circuit for non-linear energy harvesting with low voltage
output. Smart Mater. Struct. [Online]. 17(3), p. 035 030. Available:
http://stacks.iop.org/0964-1726/17/i=3/a=035030
Mickal Lallart was born in 1983. He graduated
from Institut National des Sciences Appliques de
Lyon (INSA Lyon), Lyon, France, in electrical en-
gineering in 2006, and received the Ph.D. degree in
electronics, electrotechnics, and automatics from the
same university in 2008, where he worked for the
Laboratoire de Gnie lectrique et Ferrolectricit
(LGEF).
After working as a Postdoctoral Fellowin the Cen-
ter for Intelligent Material Systems and Structures
(CIMSS) at Virginia Polytechnic Institute and State
University, Blacksburg, in 2009, he was hired as an Associate Professor in
the Laboratoire de Gnie lectrique et Ferrolectricit. His current eld of
interest focuses on vibration damping, energy harvesting, and structural health
monitoring using piezoelectric, pyroelectric or electrostrictive devices, as well
as autonomous self-powered wireless systems.
Yi-Chieh Wu was born in 1985. She received the
B.S. degree in mechanical engineering from Na-
tional Cheng Kung University, Tainan, Taiwan, in
2007. After her undergraduate study, she continued
her Masters studies in mechanical engineering in
Europe and received the M.S. degree jointly from
Trinity College Dublin, Ireland, and the Institut
National de Sciences Appliques (INSA) de Lyon,
France, in 2010. She is currently working toward the
Ph.D. degree in the Laboratoire de Gnie lectrique
et Ferrolectricit at INSA.
Her research interests include nonlinear switched piezo devices for energy
harvesting and power applications of piezoelectric materials.
Daniel Guyomar received the Masters degree in
mechanical engineering and the Ph.D. degree in
acoustics and vibrations fromCompigne University,
France, and the Ph.D. degree in physics from Paris
VII University, Paris, France.
In 19821983, he worked as a Research Asso-
ciate in uid dynamics at the University of Southern
California (USC), Los Angeles. From 1983 to 1984,
he was a National Research Council Awardee at the
Monterey Naval Postgraduate School (CA), working
to develop transient wave radiation modeling. In
1985, he was hired by the Schlumberger group to lead several projects dealing
with ultrasonic imaging. He then moved to Thomson Submarine Activities
in 1987 to manage the research activities in the eld of physical under-
water acoustics. He co-created two start-ups involved in ultrasonic devices.
He is presently a full-time Professor at the Institut National des Sciences
Appliques de Lyon, Villeurbanne, France, where he manages the Electrical
Engineering and Ferroelectricity Laboratory (Laboratoire de Gnie lectrique
et FerrolectricitLGEF). His present research interests are in the eld of
smart materials and systems: semiactive vibration control, wave control, energy
harvesting, piezotransformers, electroactive materials, and nonlinear/hysteretic
modeling of these materials.

You might also like