Sei sulla pagina 1di 10

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMSI: REGULAR PAPERS, VOL. 56, NO.

11, NOVEMBER 2009

2519

Design of a Solar-Harvesting Circuit for


Batteryless Embedded Systems
Davide Brunelli, Clemens Moser, Lothar Thiele, Member, IEEE, and Luca Benini, Fellow, IEEE

AbstractThe limited battery lifetime of modern embedded systems and mobile devices necessitates frequent battery recharging
or replacement. Solar energy and small-size photovoltaic (PV)
systems are attractive solutions to increase the autonomy of
embedded and personal devices attempting to achieve perpetual
operation. We present a batteryless solar-harvesting circuit that
is tailored to the needs of low-power applications. The harvester
performs maximum-power-point tracking of solar energy collection under nonstationary light conditions, with high efficiency
and low energy cost exploiting miniaturized PV modules. We
characterize the performance of the circuit by means of simulation
and extensive testing under various charging and discharging
conditions. Much attention has been given to identify the power
losses of the different circuit components. Results show that our
system can achieve low power consumption with increased efficiency and cheap implementation. We discuss how the scavenger
improves upon state-of-the-art technology with a measured power
consumption of less than 1 mW. We obtain increments of global
efficiency up to 80%, diverging from ideality by less than 10%.
Moreover, we analyze the behavior of supercapacitors. We find
that the voltage across the supercapacitor may be an unreliable
indicator for the stored energy under some circumstances, and this
should be taken into account when energy management policies
are used.
Index TermsDCDC power conversion, embedded systems,
energy harvesting, maximum-power-point tracking (MPPT),
photovoltaic (PV) cells, power supply, wireless sensor networks
(WSNs).

I. INTRODUCTION
HE INTEREST in supply circuits that harvest energy from
the surrounding environment for powering embedded systems has been increasing over the last years [1][4]. Thanks
to the progress in low-power design, research has greatly reduced the size and the power consumption of distributed embedded systems, and the autonomy of these systems can be
further increased by energy-harvesting techniques. Nowadays,
small solar panels suffice to ensure continued operation, and
several photovoltaic (PV) harvesting circuits have been recently
proposed for this purpose [5], [6].

Manuscript received November 27, 2007; revised October 27, 2008. First
published February 18, 2009; current version published November 04, 2009.
This work was supported by the European Network of Excellence ArtistDesign.
This paper was recommended by Associate Editor C. K. Tse.
D. Brunelli and L. Benini are with the Department of Electronics, Computer
Sciences and Systems (DEIS), University of Bologna, 40136 Bologna, Italy
(e-mail: davide.brunelli@unibo.it; luca.benini@unibo.it).
C. Moser and L. Thiele are with the Computer Engineering and Networks
Laboratory, Swiss Federal Institute of Technology (ETH) Zurich, 8092 Zurich,
Switzerland (e-mail: moser@tik.ee.ethz.ch; thiele@tik.ee.ethz.ch).
Digital Object Identifier 10.1109/TCSI.2009.2015690

The output characteristics of a PV array vary nonlinearly


when temperature or irradiance conditions change. Therefore,
maximum-power-point tracking (MPPT) techniques are exploited for adjusting the operating point of the solar panel
in order to obtain the maximum output power from
the PV module. So far, MPPT methods have been roughly
classified into two groups: large-scale PV power systems,
generally making use of digital signal processors (DSPs) or
microcontrollers [7][9], and small-scale PV power systems,
usually without digital controllers. They are less accurate, but
they are cheaper with an advantageous cost efficiency in PV applications below 50 W [10], [11]. With the increased interest in
harvesting technology, a third class of MPPT methods, focused
on microscale PV power systems with some square-centimeter
area, has recently emerged. Compared to well-known industrial
PV systems, these approaches address the tracking of the MPP
with power consumption of a few milliwatts since the maximal
energy drawn from PV modules is very limited (i.e., hundreds
of milliwatts) due to the small size of the cells.
Many tracking techniques have been developed for the former
two classes in the last years [12]. The hill-climbing method
directly computes the MPP by measuring the gradient of the
output power in dependence of the output voltage [13]. A similar
technique, namely, perturb and observe [6], [14], continuously
perturbs the systems in order to detect the MPP, and the operating voltage of the panel oscillates around it. The incremental
conductance method [15] is based on the fact that the slope of
a PV power curve is zero at the MPP, positive on the left of the
MPP, and negative on the right. Through the help of a simple
calculation, a microcontroller periodically checks and tunes the
current operating point [16], resulting usually in good accuracy
and efficiency. The fractional open-circuit voltage (FVoc) technique [10] exploits the existing nearly linear relationship between
and the open-circuit voltage
under varying irradiance and temperature levels. Since the linear factor depends
on the characteristics of the used solar cell, it usually has to be
computed beforehand by empirically determining the desired
operating points for the specific cell. The technique is, however, very easy and cheap to implement and does not require
any digital controller. Similar to FVoc, fractional short-circuit
current methods rely on the fact that there also exists an approximately linear relationship between the short-circuit current
and
[7] and that the linear factor has to be determined for
each specific solar cell. Even if this ratio is usually more stable
than fractional Voc under irradiance and temperature variations,
during operation is usually more complicated for
measuring
low-power embedded systems that need extra current sensors,
adding complexity and cost to the system. Finally, [17] exploits

1549-8328/$26.00 2009 IEEE

2520

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMSI: REGULAR PAPERS, VOL. 56, NO. 11, NOVEMBER 2009

the mean-value theorem to provide the analytic solution of a


point in a close neighborhood of the MPP. Thus, the system
never works on the actual MPP but close to it. Mathematical
solutions are stored in a lookup table of at least 64 kB and are
addressed for all possible combinations of the input variables,
such as the measured short-circuit current, open-circuit voltage,
and temperature.
Unfortunately, even if the experience in exploiting large-scale
PV modules has rapidly advanced in the last two decades, research on solar scavenger for small and low-power embedded
devices is characterized by additional new issues to tackle. In
fact, MPPT implementation is practicable only if the power consumed by the tracker is significantly lower than the amount of
output power that it gains. For this reason, most of the proposed
solutions for industrial electronics are unsuitable for low-power
embedded systems or must be reconceived at best. Moreover,
small-scale solar harvester systems usually experience a substantial interaction between the scavenger circuit and the powered device, so designing a solar harvester separately from the
embedded systems is not always feasible [6].
In this paper, we present a highly efficient solar energy harvester for batteryless embedded systems, starting from the circuit described in [18]. We present an extensive evaluation in
terms of simulations and measurements focusing on the power
consumption and efficiency of the PV harvester. The adoption
of a simple MPPT circuit allows us to shrink the size of PV modules and to reduce the capacity of the energy reservoir. In fact,
the harvester circuit consumes less than 1 mW and diverges from
the optimal situation by less than 10%. Since it does not use any
microcontroller or DSP for MPPT, the embedded system can
be shut down when unused to save energy. Moreover, it does
not require a precharged storage device (such as rechargeable
batteries), and it works even if the energy buffers are empty.
Experimental results show increments of global efficiency up to
80% and demonstrate the flexibility and robustness of our approach. All these features make the proposed solution suitable
for low-power embedded systems.
The advantage of solar energy over other forms of environmental energy is that the available solar energy can be predicted, at least to some extent. This allows one to implement
power management techniques and to plan and optimize the future system activity in order to achieve a more sustainable operation. These energy-harvesting-aware features might require
also the knowledge of the current available energy [19], [20]
to tune the system behavior. Clearly, if there is a relation between conversion efficiency and energy buffer level, the energy prediction will require more computing effort and may reduce the accuracy. Moreover, the charging process of an energy storage device (ESD) by means of a solar cell is a nonlinear process. The amount of energy that can be harvested depends on various factors, such as the voltage level of the ESD
and the incident-light intensity. Common MPPT systems exploit a combination of different storage devices (batteries and/or
supercapacitors) and methods that require microcontrollers or
power-consuming devices for the management [21]. Such hybrid systems use the advantages of both technologies providing
both high power and energy density. For example, a combination of thin-film solid-state batteries (TFBs) and supercapacitors

could be the optimal solution to bridge the mismatch between


available and required energy in small size. Unfortunately, getting the information about available stored energy to perform
power management techniques becomes too complex in systems
with low computation capability, such as wireless sensor networks (WSNs). For this reason, in this paper, we focus on supercapacitors as unique ESDs. They have better cycle life and temperature characteristics and can cope with unpredictable peak
power demands by wireless communication or advanced sensors (e.g., CMOS cameras in wireless video sensor networks
[22]). Furthermore, supercapacitors offer a simpler way to estimate the stored energy than batteries or TFBs, but we find that
the voltage across the supercapacitor may lead to misinterpretations if previous charging/discharging cycles or precharging
is not taken into account. Therefore, we investigated deeply the
characteristics of ultracapacitors and verified that the classical
, which is used to determine and
formula
optimize the efficiency of energy-scavenging systems in the literature [23], [24], needs a revision.
This paper is organized as follows. Related work will reviewed in the next section. Section III presents experiments that
show the behavior of supercapacitors used as storage devices
for PV harvester circuits. Starting from the description of the
used MPPT method, Section IV describes the implementation of
our PV harvester. Section V shows experimental results and the
achieved performance, followed by an analysis of the cost impact of such solutions, considering the scalability of the system
for small and distributed embedded systems (e.g., WSNs). Finally, Section VII concludes this paper.
II. RELATED WORK
Several techniques for the MPPT of PV arrays have been presented, and the number of proposed implementations has grown
significantly in recent years. Techniques vary in many aspects,
such as complexity, cost, or accuracy of the tracking method.
Large-scale PV power systems are out of the scope of this paper,
but a detailed survey with the great majority of articles presented
on MPPT can be found in [12] and [25].
Concerning energy harvesters using small and microscale PV
modules, [11] presents a cost-effective MPPT system that can be
directly integrated onto solar arrays. The authors focus on the
issue that it is more cost effective to design high-efficiency lowpower MPPT systems in order to scale down the PV array and
storage devices, resulting in a lower cost system that is suitable
to be utilized on wider application scenarios (e.g., distributed
embedded systems and WSNs).
References [23] and [24] propose systems that attempt to enable perpetual operation of low-power embedded systems. In
both solutions, the replenishment of the energy buffers is performed by a direct connection between the PV panel and the
storage device, which forces the operating point to the capac. Both solutions do not perform any MPPT, alitor voltage
though the size of the panel and the collected power permit some
forms of power management.
In [5], [6], and [21], low-power systems for tracking the peak
power point are presented. They exploit microcontroller and
analog circuits to track MPP during light variations. The size of
, which is enough
the adopted PV modules is greater than 20

BRUNELLI et al.: DESIGN OF A SOLAR-HARVESTING CIRCUIT FOR BATTERYLESS EMBEDDED SYSTEMS

2521

to provide tens of milliwatts and to perform an efficient power


collection. In particular, [21] supports different power sources
and tries to eliminate the overhead in cost and power consumption caused by a microprocessor-based algorithm for MPPT.
The harvesting unit is independent from the target system, and
it requires the presence of a rechargeable battery as secondary
buffer to work when the primary buffer is empty. The adopted
technique to estimate the position of the MPP relies on a light
sensor (e.g., photodiode), and they associate this information to
the solar-cell characteristic. Since the relationship between light
sensor and solar panel is usually almost empirical, this method
is less accurate than a simple fractional Voc. Additionally, the
sensor has a sizable power consumption that influences the energy budget of the system.
III. SUPERCAPACITOR ANALYSIS
The occasionally unexpected behavior of supercapacitors and
its influence on the performance of an energy scavenger have
never been investigated in the context of PV harvesters for lowpower embedded systems. In fact, supercapacitors are usually
employed in the power supplies of the driver and logic circuits
to allow them to operate in the absence of the primary power
source. Conventional ESDs, such as batteries and aluminum
electrolytic capacitors, must often be replaced during the lifetime of a product. Supercapacitors do not suffer from aging
effects, and they do not experience irreversible chemical reactions. Moreover, they do not suffer from dry-up problems, unlike
aluminum electrolytic capacitors. These features allow supercapacitors to fill the gap between conventional capacitors and batteries and to be an attractive solution as energy reservoirs for
low-power distributed embedded systems.
In Section II, we described several energy-harvesting systems that have opted for supercapacitors as at least one of the
used energy reservoirs. This papers always assume a behavior
that is identical to that of conventional capacitors with negligible internal losses. Supercapacitors accumulate charges at the
interface between the surface of a conductor and an electrolytic
solution. Even if they store energy in a similar way as lower
value capacitors, experiments showed, however, that supercapacitors do demonstrate a different behavior from normal capacitors under certain conditions, and that is not covered by
datasheets. The validity of the capacitor energy-content formula
is, however, of substantial importance to several applications
of embedded systems. In papers like [23] and [26], the information about available stored energy is used to estimate the
performance of the system and to perform power management
techniques.
Electrical charge is stored in the double layer of a supercapacitor when an external voltage is applied. The flow of
charges across the interface is not an instantaneous process, and
it is followed by a series of charge-distribution processes that
can take a considerably long time. To represent the combination
of many processes with different time behavior, a simplified
model consisting of three branches with time constants on the
order of a few seconds, a few minutes, and tens of minutes,
respectively, has been proposed [27], [28] and used for our
considerations.

Fig. 1. Internal charge-distribution process after the first charging cycle.


(a) Voltage decreasing during the delay interval. (b) Difference measured
energy and formula.

A. Charge Distribution
Supercapacitors experience several charge-distribution processes with different time constants, even in isolated and disconnected state. This makes it difficult to identify the process that
is responsible for voltage variations. After just being charged
for a short period, a disconnected supercapacitor will exhibit a
decreasing voltage. This decrease is mainly caused by a charge
distribution within branches. To be able to visualize the internal
charge-distribution processes of the supercapacitor, a measurement that is similar to that in [27] was performed (see Fig. 1).
A delay period of 10 min was inserted in between the charging
and discharging phases of a 22-F supercapacitor. The capacitor
was charged from an initial voltage of 12.5 V, and during the
delay, the charged supercapacitor was disconnected.
The graph in Fig. 1(b) shows the result of the experiment, in.
repcluding two different plots, denoted as and
resents the energy content of the supercapacitor computed by
measuring its output voltage and using the
formula, while is obtained by measuring first the energy delivered by the generator for the charging phase and then, during
discharging, measuring the energy delivered to the load. The
is the total energy provided to the supercapacitor
energy
represents the total energy delivduring charging, while
ered by the supercapacitor during discharging. Notice that the
supercapacitor does not behave as an ideal storage device as
is larger than
. Hence, we can define the cycle efficiency of
the supercapacitor as
(1)

2522

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMSI: REGULAR PAPERS, VOL. 56, NO. 11, NOVEMBER 2009

Fig. 2. Cycle efficiency (


). Charging and discharging patterns are
bounded by different lower and upper thresholds.

An important phenomenon that accounts for the nonideality


of the supercapacitor is the decrease of the supercapacitor
voltage during the delay interval shown in Fig. 1(a). No energy
is extracted from the supercapacitor, and still, the voltage
decreases. This is mainly caused by the charge-distribution
process that, in the short term, is more important than leakage.
Fortunately, this nonideal behavior becomes less significant as
the number of charging/discharging cycles increases, so
becomes a more reliable estimator of the stored energy with
errors that are lower than 10%. As a consequence with the
increment of the number of cycles, the difference between
and
decreases. The results on efficiency increment will be
discussed in the next section.
B. Multiple Cycles
In energy harvesters, supercapacitors can experience a wide
range of charging and discharging patterns, and it is obviously
impossible to test performance for every possible pattern. We
performed a set of experiments to investigate how supercapacitors perform when they are subjected to a sequence of identical
cycles. Experiments with different lower and upper boundaries
for the supercapacitor voltage were performed.
Apparently, when identical cycles are applied, the energy behavior of the supercapacitor becomes the same for each cycle
after an initial warm-up period. The evolution of cycle efficiency, as defined in (1), throughout the sequence (six cycles)
was examined for three different boundary configurations, and
the results are shown in Fig. 2. Charging and discharging between the boundaries [22.5 V] have lower efficiency for the
first two cycles but stabilize after an increase. Other than some
small variations, the other two cycle-efficiency sequences seem
already quite stable after the first cycle. In general, we can conclude that the performance of a supercapacitor is only influenced
by its most recent charging history. It takes no more than three
cycles to achieve quite-stable cycle efficiencies.
C. Leakage
Finally, leakage is also an important factor for supercapacitors, and it is shown in Fig. 3. The supercapacitors in the experiments were all charged by a voltage source waiting for the
end of any charge-distribution effects. In this way, switching off
the power supply, we can assume that voltage decrease is solely
caused by internal self-discharging. The figure clearly shows a

Fig. 3. Leakage effect of the supercapacitor. The energy content of the tested
supercapacitor is normalized with the maximum allowed energy content of the
respective device.

difference in leakage performance for the three different capacitance values. Even supercapacitors from the same manufacturer
do not show the same performance. In particular, the 33-F supercapacitor demonstrates a dissatisfying behavior, losing more
than 25% of its initial energy. In summary, our analysis shows
that supercapacitors are nonideal energy reservoirs, but they are
definitely viable as battery replacements in energy-harvesting
systems and applications where their limited memory effects,
as shown in Fig. 2, are extremely advantageous as well as the
high power density and the high flexibility in recharging. A key
advantage of supercapacitors with respect to rechargeable-battery technologies is that they can be discharged and recharged
a virtually unbounded number of times. Moreover, supercapacitor leakage is not a major concern in the case of solar energy
harvesting, as environmental energy is unlimited and becomes
available in fluctuating and periodic patterns (24 h), so a suitable size of the harvesting transducer will provide an average
energy intake that compensates the leakage effect. We point out,
however, that the nonideality of supercapacitors should be carefully taken into account when developing environmental energy
management policies [20], [26] that rely on the knowledge of
the available stored energy. Such an issue could be addressed
by measuring the ultracapacitor voltage and applying a correction factor to the formula of the energy, depending on the current state of the ESD. Another technique would be to retrieve
mathematical solutions stored in a lookup table, which will be
addressed for a combination of input variables and the number
of cycles experienced by the capacitor.
IV. SYSTEM DESIGN
A. MPPT Problem
In most cases, a solar cell panel consists of multiple solar
cell elements connected in parallel and/or series, where the current source of the solar cell is dependent on the intensity of
the incident light.
The characteristic of a PV module, when neglecting the
internal shunt resistance, is given by the following equation:
(2)

BRUNELLI et al.: DESIGN OF A SOLAR-HARVESTING CIRCUIT FOR BATTERYLESS EMBEDDED SYSTEMS

Fig. 4.

Fig. 6. Nearly linear relation between V


conditions of irradiance.

I V and P V plots of the used PV module.

2523

and the pilot cell under various

To overcome both problems, we employ an additional small


PV module acting as a reference module. Fig. 6 shows how the
pilot cell follows almost linearly the behavior of the main PV
module during light variations. The plot displays the operating
over the
of the main
voltage of the pilot cell
and the
of
array. For clarity, we also plot the same
the main module to verify the similarity of the variations. As
shown, the behavior is nearly linear, so it is possible to exploit
as a reference signal
the voltage of the reference cell
for tracking the position of the MPP
(3)
Fig. 5. Fractional open-circuit voltage relation between V
various conditions.

and V

under

where is the generated current,


is the reverse saturation
current, is the electronic charge, is a dimensional factor,
is the Boltzmann constant, is the temperature (in kelvins), and
is the series resistance of the cell. The characteristic plot of
the PV module adopted in our solar harvester is shown in Fig. 4.
The aim of an MPPT method is to harvest as much energy
as possible from the available solar panel by operating continuously as close as possible to its MPP adapting to irradiance
and temperature condition variations. Among the numerous proposed methods, the fractional open-circuit voltage technique is
the most effective one in terms of cost and power consumption
for microscale PV systems.
Fig. 5 shows the nearly linear proportional relationship
of a PV
between the operating voltage at the MPP
module and the open-circuit voltage
under different
irradiance conditions. By a periodical brief disconnection of
the PV module from the circuit, the MPP can be estimated by
. This conceptually simple solution presents
measuring its
two important practical drawbacks: 1) The system experiences
a temporary drop of power from the panel, and 2) this control
method is not a real tracking method but a quasi-power point
tracker. In fact, the harvester tunes the tracker according to
sampling, and it does not try to find
the value of the last
continuously the current MPP. If the sampling rate is low,
or the environmental conditions change quickly, this method
decreases the accuracy and responsiveness.

The reference cell has been carefully selected to represent the


characteristics of the principal PV array, and since it is exposed
to the same light conditions as the main solar cell, it is reasonable to suppose that both PV modules will be built in the same
manufacturing process. For the proposed harvester, we adopt the
CPC1824 [29], a monolithic PV string of solar cells of only 9
. Providing continuously feedback irradiance information,
we can perform continuous tracking, and the power from the
main cell will always be used to power the embedded system or
stored into the energy buffer. Moreover, using a pilot cell has
several additional advantages: 1) It is not necessary to provide
a power supply for a light irradiance sensor, as in [21], and (2)
as the pilot cell can be extremely small, it does not waste power
that could be used for powering the system, it does not increase
the system size and cost, and it does not require extra components, such as photodetectors and signal-conditioning circuits.
B. Harvester Platform
The hardware architecture of the solar scavenger is shown
in Fig. 7. It is realized using commercial off-the-shelf (COTS)
parts, and it consists of three units: the MPP regulator, the
MPP tracker, and the MPP power supply. The MPP regulator
that is usually exploited in MPPT techniques is a buck power
inductor. It operates in continuous
converter with a 2.2is lower
mode because the maximum capacitor voltage
than the nominal operating voltage of the solar cell. Since the
input power
varies continuously with the atmospheric conditions, the circuit adjusts dynamically the output

2524

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMSI: REGULAR PAPERS, VOL. 56, NO. 11, NOVEMBER 2009

Fig. 7. Conceptual diagram of the harvester platform.

pulsewidth-modulated (PWM) signal to track the maximum


operating point supported by a small capacitance of 470 nF at
the input. In our work, narrow pulses with variable duty ratio
are generated by the tracking unit to drive a PMOS transistor
with minimum power losses due to parasitic resistance and
switching actions.
The core of the scavenger is the MPP tracker that attempts
to obtain the maximum achievable power from the solar cell.
is led to one of the inputs of the
The operating voltage
ultralow-power comparator LTC1440 [30], which compares to
derived by the pilot cell using a voltage
the reference
for the
divider and generates the PWM control signal
buck converter. By adding hysteresis, one threshold voltage is
replaced by a lower and an upper threshold. The hysteresis
prevents unwanted switching that would occur because of the
noise of the solar cell and pilot cell. In this way, the actual
operating point is not a fixed value but oscillates around the
MPP. Narrowing the hysteresis around the estimated MPP
means operating at higher switching frequencies, with higher
conversion efficiency, because the solar cell is confined to a
smaller voltage range.
We intensively simulated the circuit before implementation,
and a simulation plot is shown in Fig. 8. This specific simulation
depicts the startup of the MPP tracker. The curve
represents
the current going through the inductor in the direction of the
supercapacitor, while
represents the current leaving the
solar cell. The small negative dent of
at the beginning of the
simulation is caused by the fact that the comparator still needed
to start up. For a moment, the switch was closed, which caused
the current to flow from the slightly precharged supercapacitor
. Moreover, the figure clearly shows
to the input capacitor
how the PV voltage oscillates around the estimated MPP.
Different from the microscale solar scavengers proposed in
the literature, the tracker is still able to operate, even when the
energy reservoir is empty. Indeed, the MPP power-supply unit,
with two low-threshold diodes, leads two possible power supplies to the harvester comparator, and only one source can be
employed at the same time. With the supercapacitor being empty
and the dc/dc converter being turned off, the tracker is still able
to work if the solar cell supplies a high-enough voltage. This
way of supplying power to the harvester tries to ensure proper

Fig. 8. Simulation of the startup of the MPP tracker.

Fig. 9. Embedded platform powered by the solar harvester.

operation of the MPP tracker, even when the supercapacitor is


completely discharged and permits one to have the highest control signal
to switch off the PMOS transistor, suppressing
any drain current that could decrease the performance of the
power converter.
The proposed solar harvester does not need any specific microcontroller or computational platform. It is very flexible and
can supply a large variety of low-power embedded systems. In
particular, we use a 50-F ultracapacitor as ESD, while the output
dc/dc regulator is an LTC3401 step-up voltage regulator [31],
which provides a stable 3.3-V voltage with up to 97% efficiency.
In Fig. 9, we show a photograph where we power a microcontroller-based embedded system equipped with several sensor devices and radio communication capability. It is a commercial
sensor platform for WSN applications called Tmotesky [32]. It
is equipped with a 16-bit RISC CPU architecture with 8-kB of
on-chip flash memory and 256 B of RAM. It accommodates at
least seven different sensors (e.g., temperature, humidity, etc.)
and can transmit radio information to other sensor nodes or directly to a PC using a low-power radio within a range of about
50 m. The peak power consumed by our test application is 90
mW when the radio is active. Adjusting the communication rate,
it is possible to guarantee an average energy intake that is higher
than the energy required by the Tmotesky platform and to sustain the application.

BRUNELLI et al.: DESIGN OF A SOLAR-HARVESTING CIRCUIT FOR BATTERYLESS EMBEDDED SYSTEMS

2525

Fig. 11. Efficiency of power conversion.


Fig. 10. Power consumption of the comparator in the MPP tracker during
charging and idle phases.

V. EXPERIMENTAL RESULTS
The core of this paper is the design of an autonomous energy
harvester that can be used as simple plug-and-play enhancement
to embedded systems and wireless sensor nodes. A test setup has
PV module, with an
been implemented exploiting a 112irradiation that forces the solar cell to produce about 50 mW. We
evaluated the performance by measuring the power consumed
by the MPPT system and the efficiency of the PV harvester.
A. Power Consumption
The average power consumed by the whole circuit is below
1 mW, and the main contributions are given by the comparator
of the MPP tracker and by the switching activity of the MPP
regulator. The power consumed by the comparator is shown in
Fig. 10. The measurement was performed using a 50-F supercapacitor. After a complete recharging, the PV panel was covered
in order to emulate the absence of environmental energy and to
power the target system using only the energy in the reservoir.
We repeated several cycles and measured the power consumption also during idle periods. The figure shows that a maximum
of 700
is used at the beginning of charging, during which
the frequency of the MPP tracker is the highest. In the charging
interval, the frequency and the average comparator power decrease. Right before the end of charging, only 150
is required. When the switch of the MPP regulator is open, the circuit consumes less than 0.5 mW
- . Peaks of about
1 mW have been measured when the comparator switches the
MOS transistor on
- , but the actual value depends on
.
the voltage of the supercapacitor
B. Harvester Efficiency
We could define the harvesting process efficiency as follows:
(4)
where

is the power at the MPP and


is the average power transferred to the energy buffer.
Using this definition, only scavengers that always work at the
correct MPP can achieve efficiencies that are close to 100%,

and losses are only caused by power dissipation of system components. In our tests, has also been evaluated considering
the dc/dc converter that affects the result with its own intrinsic
losses. Fig. 11 shows how efficient the proposed method is in
replenishing the supercapacitor over elapsed time. The continuous curve represents the efficiency of the charging behavior in
the case of a direct connection between the PV panel and the
storage device. The dashed curve shows the ideal trend: The supercapacitor is constantly refilled with the maximum available
. The charging behavior using the proposed energy
power
scavenger is plotted as the curve with dots.
The curve with triangles has been obtained, excluding from
the scavenging platform the MPP power-supply unit and using
the dc/dc as a unique source. This configuration is implemented
by several scavenging solutions, such as [21]. It is evident that
they cannot perform better since their tracking circuit does not
operate with an empty energy reservoir. Nevertheless, the harvester is still able to charge the ESD because the comparator
output is low, switching the PMOS transistor on and guaranteeing a conductive path to the supercapacitor. In the interval
between
, the dc/dc attempts sporadically to start up, providing temporary supply to the tracking
circuit, which causes a slight increment of the average efficiency of the system. Only when the voltage level of the
supercapacitor is high enough to turn the dc/dc on steadily
that the scavenger can work properly
and increase the efficiency.
When the dc/dc regulator is active, it introduces an overhead due to its own power consumption, which decreases the
charging efficiency. In Fig. 12, we try to estimate this phenomenon, illustrating the charging behavior of a direct connection
and of our harvester with and without an output dc/dc converter
(in the case of our harvester, the tracking circuit was powered
by an external source). As is known, a direct connection between the PV module and the supercapacitor is characterized
by a linear charging shape. Meanwhile, our scavenger presents
higher rate of conversion efficiency from the beginning. Moreover, we observe that the presence of an output dc/dc converter
decreases the charging slope as soon as the voltage of the enreaches
, turning the converter
ergy buffer
on. The oscillations of the actual operating point of the solar

2526

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMSI: REGULAR PAPERS, VOL. 56, NO. 11, NOVEMBER 2009

Fig. 12. Overhead of the dc/dc converter in the charging process.

Fig. 14. Efficiency of the tracking system.

system is plotted in the last photograph. Since the duty cycle


of the switching activity is very different from the ideal one
), the PV module will never operate at
(in this case,
the MPP.
Finally, we also evaluated how the tracking system influences
the energy conversion and the charging process. Fig. 14 shows
the efficiency defined in Section V-B using different irradiance
levels from the previous plots. Also, here, the lowest continuous curve depicts the direct connection between the PV module
and the supercapacitor, while the dashed curve is the ideal case
using a constant
as source. The curve with dots is obtained with the proposed circuit and can achieve conversion efficiency up to 80%. As is shown, it approaches closely the ideal
curve, with a maximum error that is less than 10%. Excluding
the MPP tracker, the efficiency curve varies widely with the irradiance condition, ranging from the dashed curve to below the
continuous one, representing the direct connection. The figure
shows a case in which decreases to around 50% (curve with
squares).

VI. COST ANALYSIS

Fig. 13. Comparison of the operating point with and without a tracking circuit.

cell simulated in Fig. 8 were measured directly on the implemented system. In Fig. 13, the ideal situation, shown in the
first photograph, is obtained when the MPP regulator is driven
directly by a signal generator tuned at 100 kHz. A duty ratio
of the switching activity of the MOS transistor is defined as
, where
is the switch-on time of the MOS tranis computed in order to match
sistor. The duty ratio
of the solar cell at the given irradiaccurately the actual
ance. Under the same condition, our solar harvester oscillates
around
(as shown in the middle photograph) because it
. Fiautomatically generates a control signal with
nally, the operating point of the circuit without the tracking

The proposed solar harvester can be very economical, and the


total cost is currently below 50 Euros. Furthermore, using (surface mounting device) SMD technology in the PV panel itself
or directly onto the target system, it also occupies a small area.
However, the cost of the solar harvester is an important consideration that certainly influences the choice between components
(i.e., high-speed versus ultralow-power comparators, supercapacitors versus other ESDs such as TFBs, etc.), considering also
that high volumes of manufacturing will decrease the component cost.
Of course, the contribution of the harvester to the global costs
may vary widely, depending on the supplied embedded system.
As simple analysis, Fig. 15 shows the cost distribution of the
system used for experimental results and shown in Fig. 9. The
pie chart shows that the PV module and the supercapacitor take
up to 84% of the total harvester cost, demonstrating that a harvester, which does not need any digital controller, increases the
cost efficiency.

BRUNELLI et al.: DESIGN OF A SOLAR-HARVESTING CIRCUIT FOR BATTERYLESS EMBEDDED SYSTEMS

2527

Fig. 15. Cost distribution of the batteryless solar-powered sensor node.

VII. CONCLUSION
An integrated cost-effective PV harvester for low-power and
environmental embedded systems has been proposed. The adoption of an MPP circuit has led to several benefits, such as the
possibility to shrink the size of PV modules or to reduce the capacity of the energy reservoir. The presented circuit performs
a high-efficiency conversion through an ultralow-power circuit
that requires less than 1 mW. The estimation of the peak power
point is done automatically, using a small PV module as reference, whereby sensing operation does not require additional
power. The scavenger can be used with any kind of embedded
system. Experimental results have shown that the global efficiency diverges from the ideal situation by less than 10%.
REFERENCES
[1] B. C. Yen and J. H. Lang, A variable-capacitance vibration-to-electric
energy harvester, IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 53,
no. 2, pp. 288295, Feb. 2006.
[2] G. K. Ottman, H. F. Hofmann, A. C. Bhatt, and G. A. Lesieutre, Adaptive piezoelectric energy harvesting circuit for wireless remote power
supply, IEEE Trans. Power Electron., vol. 17, no. 5, pp. 669676, Sep.
2002.
[3] C. Sauer, M. Stanacevic, G. Cauwenberghs, and N. Thakor, Power
harvesting and telemetry in CMOS for implanted devices, IEEE Trans.
Circuits Syst. I, Reg. Papers, vol. 52, no. 12, pp. 26052613, Dec. 2005.
[4] P. Li and R. Bashirullah, A wireless power interface for rechargeable
battery operated medical implants, IEEE Trans. Circuits Syst. II, Exp.
Briefs, vol. 54, no. 10, pp. 912916, Oct. 2007.
[5] C. Alippi and C. Galperti, An adaptive system for optimal solar energy harvesting in wireless sensor network nodes, IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 55, no. 6, pp. 17421750, Jul. 2008.
[6] F. Simjee and P. H. Chou, Everlast: Long-life, supercapacitor-operated wireless sensor node, in Proc. ISLPED, 2006, pp. 197202.
[7] T. Noguchi, S. Togashi, and R. Nakamoto, Short-current pulse-based
maximum-power-point tracking method for multiple photovoltaic-andconverter module system, IEEE Trans. Ind. Electron., vol. 49, no. 1,
pp. 217223, Feb. 2002.
[8] R.-J. Wai and W.-H. Wang, Grid-connected photovoltaic generation
system, IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 55, no. 3, pp.
953964, Apr. 2008.
[9] W. Xiao, M. Lind, W. Dunford, and A. Capel, Real-time identification of optimal operating points in photovoltaic power systems, IEEE
Trans. Ind. Electron., vol. 53, no. 4, pp. 10171026, Aug. 2006.
[10] D.-Y. Lee, H.-J. Noh, D.-S. Hyun, and I. Choy, An improved MPPT
converter using current compensation method for small scaled PV-applications, in Proc. 18th Annu. IEEE APEC, Feb. 913, 2003, vol. 1,
pp. 540545.
[11] J. Enslin, M. Wolf, D. Snyman, and W. Swiegers, Integrated photovoltaic maximum power point tracking converter, IEEE Trans. Ind.
Electron., vol. 44, no. 6, pp. 769773, Dec. 1997.
[12] T. Esram and P. Chapman, Comparison of photovoltaic array maximum power point tracking techniques, IEEE Trans. Energy Convers.,
vol. 22, no. 2, pp. 439449, Jun. 2007.
[13] C. Hua, J. Lin, and C. Shen, Implementation of a DSP-controlled photovoltaic system with peak power tracking, IEEE Trans. Ind. Electron.,
vol. 45, no. 1, pp. 99107, Feb. 1998.

[14] N. Femia, G. Petrone, G. Spagnuolo, and M. Vitelli, Optimization of


perturb and observe maximum power point tracking method, IEEE
Trans. Power Electron., vol. 20, no. 4, pp. 963973, Jul. 2005.
[15] T.-Y. Kim, H.-G. Ahn, S. K. Park, and Y.-K. Lee, A novel maximum
power point tracking control for photovoltaic power system under
rapidly changing solar radiation, in Proc. IEEE ISIE, 2001, vol. 2,
pp. 10111014.
[16] Y.-C. Kuo, T.-J. Liang, and J.-F. Chen, Novel maximum-power-pointtracking controller for photovoltaic energy conversion system, IEEE
Trans. Ind. Electron., vol. 48, no. 3, pp. 594601, Jun. 2001.
[17] C. Rodriguez and G. A. J. Amaratunga, Analytic solution to the photovoltaic maximum power point problem, IEEE Trans. Circuits Syst.
I, Reg. Papers, vol. 54, no. 9, pp. 20542060, Sep. 2007.
[18] D. Brunelli, L. Benini, C. Moser, and L. Thiele, An efficient solar energy harvester for wireless sensor nodes, in Proc. Conf. DATE, 2008,
pp. 104109.
[19] A. Kansal, J. Hsu, M. Srivastava, and V. Raghunathan, Harvesting
aware power management for sensor networks, in Proc. 43rd ACM/
IEEE Des. Autom. Conf., Jul. 2428, 2006, pp. 651656.
[20] C. Moser, L. Thiele, D. Brunelli, and L. Benini, Adaptive power management in energy harvesting systems, in Proc. Conf. DATE, 2007, pp.
773778.
[21] C. Park and P. Chou, AmbiMax: Autonomous energy harvesting platform for multi-supply wireless sensor nodes, in Proc. SECON, 2006,
vol. 1, pp. 168177.
[22] M. Magno, D. Brunelli, P. Zappi, and L. Benini, A solar-powered
video sensor node for energy efficient multimodal surveillance, in
Proc. 11th Euromicro Conf. Digital Syst. Des., Parma, Italy, Sep. 2008,
pp. 512519.
[23] X. Jiang, J. Polastre, and D. E. Culler, Perpetual environmentally powered sensor networks, in Proc. 4th Int. Symp. IPSN, Apr. 2527, 2005,
pp. 463468.
[24] V. Raghunathan, A. Kansal, J. Hsu, J. Friedman, and M. B. Srivastava,
Design considerations for solar energy harvesting wireless embedded
systems, in Proc. IPSN, Apr. 2527, 2005, pp. 457462.
[25] J. M. Carrasco, L. G. Franquelo, J. T. Bialasiewicz, E. Galvan, R. C. P.
Guisado, M. A. M. Prats, J. I. Leon, and N. Moreno-Alfonso, Powerelectronic systems for the grid integration of renewable energy sources:
A survey, IEEE Trans. Ind. Electron., vol. 53, no. 4, pp. 10021016,
Jun. 2006.
[26] C. Moser, D. Brunelli, L. Thiele, and L. Benini, Real-time scheduling
for energy harvesting sensor nodes, Real-Time Syst., vol. 37, no. 3, pp.
233260, Dec. 2007.
[27] L. Zubieta and R. Bonert, Characterization of double-layer capacitors
for power electronics applications, IEEE Trans. Ind. Appl., vol. 36, no.
1, pp. 199205, Jan./Feb. 2000.
[28] T. Wei, X. Qi, and Z. Qi, An improved ultracapacitor equivalent circuit model for the design of energy storage power systems, in Proc.
ICEMS, Oct. 2007, pp. 6973.
[29] CPC1824 4 V Output Solar Cell, Clare, Inc., Beverly, MA, 2008,
datasheet no. DS-CPC1824-R01, Tech. Rep..
[30] LTC1440/LTC1441/LTC1442 Ultralow Power Single/Dual Comparator With Reference, Linear Technol. Corp., Milpitas, CA, 1996,
datasheet no. LT 0806 REV D, Tech. Rep..
[31] LTC3401 1 A, 3 MHz Micropower Synchronous Boost Converter,
Linear Technol. Corp., Milpitas, CA, 2001, datasheet no. LT 0607 REV
B, Tech. Rep..
[32] J. Polastre, R. Szewczyk, and D. Culler, Telos: Enabling ultra-low
power wireless research, in Proc. 4th Int. Conf. IPSN, Piscataway, NJ,
Apr. 2005, pp. 364369.

2528

IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMSI: REGULAR PAPERS, VOL. 56, NO. 11, NOVEMBER 2009

Davide Brunelli was born in Italy on March 26,


1977. He received the M.S. (summa cum laude)
and Ph.D. degrees in electrical engineering from the
University of Bologna, Bologna, Italy, in 2002 and
2007, respectively.
From 2005 to 2007, he was an Academic Guest
with the Swiss Federal Institute of Technology
(ETH) Zurich, Zurich, Switzerland. He is currently
a holder of a postdoctoral position with the Department of Electronics, Computer Sciences and
Systems (DEIS), University of Bologna. His research
interests concern the development of new techniques of energy scavenging
for wireless sensor networks (WSNs) and embedded systems, optimization
of low-power and low-cost WSNs, interaction and design issues in embedded
personal and wearable devices, and pervasive and ubiquitous computing.

Clemens Moser received the B.Sc. and Dipl.Ing.


degrees in electrical engineering and information
technology from the Technical University of Munich,
Munich, Germany, in 2003 and 2004, respectively.
For his diploma thesis in 2004, he joined DoCoMo
Euro-Labs to work on the topology aspects of
wireless multihop networks. He is currently working
toward the Ph.D. degree in the Computer Engineering and Networks Laboratory, Swiss Federal
Institute of Technology (ETH) Zurich, Zurich,
Switzerland.
His research interests include the design, analysis, and optimization of energy-harvesting sensor networks.

Lothar Thiele (S83M85) was born in Aachen,


Germany, on April 7, 1957. He received the Dipl.Ing.
and Dr.Ing. degrees in electrical engineering from
the Technical University of Munich, Munich, Germany, in 1981 and 1985, respectively.
Since 1994, he has been a Full Professor of computer engineering with the Swiss Federal Institute
of Technology (ETH) Zurich, Zurich, Switzerland,
where he is currently leading the Computer Engineering and Networks Laboratory. His research
interests include models, methods, and software
tools for the design of embedded systems, embedded software, and bioinspired
optimization techniques.
Dr. Thiele was the recipient of the Dissertation Award from the Technical
University of Munich in 1986, the Outstanding Young Author Award from
the IEEE Circuits and Systems Society in 1987, the Browder J. Thompson
Memorial Award from IEEE in 1988, and the IBM Faculty Partnership Award
in 20002001. In 2004, he joined the German Academy of Natural Scientists
Leopoldina. In 2005, he was the recipient of the Honorary Blaise Pascal Chair
of University Leiden, The Netherlands.

Luca Benini (S94M97SM04F07) received


the Ph.D. degree in electrical engineering from
Stanford University, Stanford, CA, in 1997.
He is currently a Full Professor with the Department of Electronics, Computer Sciences and
Systems (DEIS), University of Bologna, Bologna,
Italy. He also is a holder of a visiting faculty position
with the Ecole Polytechnique Federale de Lausanne,
Lausanne, Switzerland. He has published more than
350 papers in peer-reviewed international journals
and conference proceedings, four books, and several
book chapters. His research interests are in the design of system-on-chip
platforms for embedded applications. He is also active in the areas of energy-efficient smart sensors and wireless sensor networks.
Dr. Benini is an Associate Editor for the IEEE TRANSACTIONS ON
COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS and the
ACM Journal on Emerging Technologies in Computing Systems.

Potrebbero piacerti anche