Energy Harvesting State of the Art - Harb 2010

8/18/2019 Energy Harvesting State of the Art - Harb 2010

1/14

Energy harvesting: State-of-the-art

Adnan Harb*

Department of Electrical Engineering, UAE University, 17555 Al Ain, United Arab Emirates

a r t i c l e i n f o

Article history:

Received 5 April 2010Accepted 9 June 2010

Available online 10 July 2010

Keywords:

Energy harvestingEnergy sources

Power management

Power electronicsMEMS

CMOS

a b s t r a c t

This paper presents a brief history of energy harvesting for low-power systems followed by a review of the state-of-the-art of energy harvesting techniques, power conversion, power management, and battery

charging. The advances in energy harvesting from vibration, thermal, and RF sources are reviewed as wellas power management techniques. Examples of discrete form implementation and integrated formimplementation using microelectromechanical systems (MEMS) and CMOS microelectronic processes

are also given. The comparison between the reviewed works concludes this paper. 2010 Elsevier Ltd. All rights reserved.

1. Introduction

Energy (orpower) harvestingor (scavenging) is withoutany doubt

a very attractive technique for a wide variety of self-powered micro-systems. Examples of such systems are wireless sensors, biomedicalimplants, military monitoring devices, structure-embedded instru-mentation, remote weather station, calculators, watches, Bluetooth

headsets. Recently, Nokia announced it is developing a mobileprototype that could harvest energy from ambient radio wavesemitted from mobile antennas, TV masts and other sources [1e6].

Energy harvesting has become of a growing interest in the last

few years and research report number has kept increasing for thelast decade. The scope of this paper is to provide the researchcommunity with an update of the state-of-the-art of energy har-vesting from vibration, thermal, and RF sources. The principle of

energy harvesting approaches can be found in Ref. [7].In the following, we list the energy harvesting sources (Section 2),

a brief history of energy scavenging (Section 3), state-of-the-artbased on the review of several recently published papers (Section 4)

and Conclusion.

2. Energy harvesting sources

Even though macro-energy harvesting has been around forcenturies in the form of windmills, watermills and passive solar

power systems, etc., they are not game changers for electronic

designers whose mission in life is to snip the wires e includingpower cords and even battery powered systems where theperpetual device is the ultimate design goal [8]. Progress in ultra-low-power microelectronic technology with the advance in micro-

energy Harvesting makes the number of battery charging cycles themain limit towards the perpetual self-powered device.

Towards this mission, and to meet the design community ’s longmarch to ultra-low-power technology, we can identify several

micro-energy harvesting sources:

Motion, vibration or mechanical energy: oors, stairs, object’smovement, transfer energy from the engine to the battery

during braking, etc. The electromechanical transducer canbe electromagnetic [9], electrostatic [10], or piezoelectric[11,12].

Electromagnetic (RF): Base stations, wireless internet, satel-

lite communication, radio, TV, digital multimedia broad-casting, etc. One must not confuse between electromagneticenergy source and electromagnetic transducer. In some arti-cles, electromagnetic generator is used for electromagnetic

transducer.

Thermal.

Momentum generated by radioactive reactions into electrical

energy.

Pressure gradients.

Micro water ow (e.g. faucet).

Solar and light.

Biological.* Tel.: þ971 3 713 606; fax: þ971 3 762 3156.

E-mail address: [email protected].

Contents lists available at ScienceDirect

Renewable Energy

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m/ l o c a t e / r e n e n e

0960-1481/$ e see front matter 2010 Elsevier Ltd. All rights reserved.

doi:10.1016/j.renene.2010.06.014

Renewable Energy 36 (2011) 2641e2654

mailto:[email protected]://www.sciencedirect.com/science/journal/09601481http://www.elsevier.com/locate/renenehttp://dx.doi.org/10.1016/j.renene.2010.06.014http://dx.doi.org/10.1016/j.renene.2010.06.014http://dx.doi.org/10.1016/j.renene.2010.06.014http://dx.doi.org/10.1016/j.renene.2010.06.014http://dx.doi.org/10.1016/j.renene.2010.06.014http://dx.doi.org/10.1016/j.renene.2010.06.014http://www.elsevier.com/locate/renenehttp://www.sciencedirect.com/science/journal/09601481mailto:[email protected]


2/14

Another classication scheme may consider who or what

provides the energy for conversion: the rst kind is called the

human energy source. The energy is provided by the activity of human beings or animals. The second kind is the energy harvestingsource that gets its energy from the environment [13,14].

The power transferred to a load is limited by the availability of the raw energy, and the ef ciency of the transducer and theconversion circuit.

The discontinuous nature of energy harvesting has conse-

quences on the way the electronic devices powered by energy

harvesting are operated. In principle, we can distinguish two

situations:

1. The power consumption of the device is lower than the averageharvested power. In this case, the electronic device may oper-

ate continuously.2. The power consumption of the device is greater than the

average harvested power. The operation must be discontin-uous, and the time between operations depends on the stored

energy of the device [13].

3. History

The rst observation of harvesting energy in form of current

from natural source was in 1826. Thomas Johann Seebeck foundthat a current would ow in a closed circuit made of two dissimilarmetals when they are maintained at different temperatures [15,16].For the following three decades, the basic thermoelectric effects

were explored and understood macroscopically, and their

Fig. 1. (a) Piezoelectric generator voltage versus time after rectication for a single

impact applied to the generator, (b) Schematic diagram of the connections between

the piezoelectric lm and the electrostatic generators. The diodes are Schottky type

with forward voltage drop near 0.33 V [24].

Fig. 2. Energy harvesting and battery-charging system proposed by Torres et al. [1].

Fig. 3. Three steps to harvest energy: (a) Battery pre-charges the capacitor, (b)

vibrations cause capacitance value to decrease and energy is harvested into the battery,

and (c) reset [1].

Fig. 4. A complete harvesting cycle [1].

A. Harb / Renewable Energy 36 (2011) 2641e 26542642


3/14

applicability to thermometry, power generation, and refrigerationwas recognized [17].

In 1839, as he was experimenting with an electrolytic cell

composed of two metal electrodes, Edmund Becquerel discovered

the photovoltaic effect [18]. The rst large area solar cell was con-

structed in 1894 by Charles Fritts who coated a layer of seleniumwitha thin layer of gold [19]. While the photovoltaic effect wasrstobserved by Edmund Becquerel, it became fully comprehensibleonly after developing the quantum theory of light and solid statephysics in the early 1900s [18].

Joseph Henry and Michael Faraday independently discoveredthe principle of producing electricity from magnetism, known aselectromagnetic induction, in 1831 [20]. In October of the sameyear, Faraday invented the rst direct-current generator consisting

of a copper plate rotating between magnetic poles [21].The rst observation of harvesting energy in form of charge was

in 1880. Pierre and Jacques Curie successfully predicted and provedexperimentally that certain crystals would exhibit a surface charge

when subject to mechanical stress. This phenomenon was given the

name piezoelectricity [15].

Fig. 5. (a) Capacitor C par added as a hybrid alternative to voltage-constrained and

charge-constrained energy harvesting systems. (b) Timing waveforms [28].

Fig. 6. A plan view of the variable capacitor implemented using MEMS technology

[28].

Fig. 7. A dual polarity boost converter [29].

Fig. 8. Spice simulation of the circuit of Fig. 7, positive half cycle, for a source

displacement of 25 mm at 322 Hz. Top: Accumulated energy extracted from the coil,

output to the reservoir, and dissipated in the three main loss mechanisms. The three

loss mechanism lines are of similar magnitude. Bottom Boost inductor instantaneous

current [29].

A. Harb / Renewable Energy 36 (2011) 2641e 2654 2643


4/14

4. State-of-the-art review

4.1. Harvesting from vibrations

There has been much recent interest in using MEMS (Micro-

electromechanical Systems) to scavenge energy from ambientvibration and transfer it to electrical load. Such device is mechan-ically modeled with the base excitation of an elastically mounted

seismic mass moving past a coil [22]. A mathematical model of the

transferred energy is developed in Ref. [22]. It takes into consid-eration the seismic mass amplitude, the magnitude and frequencyof the excitation, the electrical analoge of the mechanical damping

coef cient. It is found that the matching condition to transfermaximum power to the load is different from the well knownsimple electrical matching and is given by:

Rload ¼ Rint þ K 2=c m

Where Rload is the load resistance, Rint is the coil internal resistance,

K is the electromechanical coupling (transducer) coef cient, and c mis the mechanical damping coef cient.

In order to nd optimal architectures for maximal power

generation under the different operating constraints, analysis andverication by simulation of three classes of MEMS-based vibra-tion-driven microgenerator architectures are presented in Ref. [23].

The three classes are velocity-damped resonant generators(VDRGs), Coulomb-damped resonant generators (CDRGs), and

Fig. 9. Modied buck converter [29].

Fig. 10. Ef ciencies of an electrostatic micro-generator. The maximum ef ciency of

a buck converter is reached for 10 cells of the unity MOSFET considered in simulation

[29]. Percentage scale was not in the original

gure.

Fig. 11. Magnetic spring generator structure using single moving magnet and 2 xed

magnets [30].

Fig.12. Measured no-load voltage during walking and slow running for the generator

with higher ef

ciency [30].



5/14

Coulomb-force parametric generator (CFPG). The later is a newclass that doesn’t operate in a resonant matter. The authorsconclude that each class is superior in power generation in different

conditions related to the resonant frequency and Z l/Y 0, where Z l isthe maximum possible of the amplitude of the mass-to-frame

displacement and Y 0 is the source motion amplitude. For idealimplementations, the CFPG produces the most power where ( Z l/

Y 0) < 0.1, the VDRG is superior above the resonant frequency when( Z l/Y 0) > 0.1, and the CDRG is superior below but near the resonantfrequency when ( Z l/Y 0) > 0.1.

Rocha et al. [24] described a system integrated in footwear to

harvest energy from vibration when people walk. Piezoelectric andelectrostatic generators were used to convert pressure variations toenergy because they provide the highest density of powercomparing to electromagnetic [25,26]. When the person steps with

the foot on theoor, the capacitance (two metallic plates separatedby a exible dielectric material) of the electrostatic generatorincreases about two times, which means that the voltage decreasesto one half. At this time, the piezoelectric generator (polymeric

material coated in both sides by a conducting material, which formthe electrodes) voltage is higher than the one at the electrostaticgenerator terminals (Fig. 1a), so, its capacitance will be charged.When the person raises his/her foot, the capacitance of the elec-

trostatic generator decreases and its voltage increases. In this case,the load that is an energy storage device (thin lm lithium batteryof 3 V) will have its voltage lower than the electrostatic generator,

the charge is transferred from the generator to the battery and thediode bridge is blocked (Fig. 1b). The power generated ranges fromtens to hundreds of milliwatts and the average energy generated in1 h, by a running person when the generator is coupled to a resis-

tive load doesn’t exceed 51 mJ.

Torres and Rincón-Mora [1,27] proposed a voltage-constrainedenergy harvesting system, i.e. the voltage of the harvested energy

storage battery limits the maximum voltage of the capacitor (Fig. 2).It works as follows: The pre-charge control block lets the batterypre-charge the capacitor C VAR to V BAT through inductance L (Energy

loss) when its capacitance is maximum. It energizes L by switchingS 1 and S 3 ON while S 2 and S 4 are OFF, then it charges C VAR to V BAT by

turning S 2 and S 4 ON while S 1 and S 3 are OFF. S 1 through S 4 are thenall turned OFF leaving the capacitor under charge-constrained

Q C-VAR (open-circuit).

C VAR is designed in such a way ambient vibrations cause its

capacitance to decrease, converting mechanical energy to electrical(Energy harvesting): V C-VAR increases when C VAR decreases as

Q C-VAR ¼ constant and Q VAR ¼ C VAR $V C-VAR ¼ constant. When V C-VAR is high enough, the energy harvested in the capacitor will charge

the battery through diode D (current I HARV ) and V C-VAR will remainlimited by V BAT þ V D. The capacitor is then reset to its maximumvalue before getting pre-charged again by the battery. Figs. 3 and 4show the three harvesting steps, and the associated C VAR , V C-VAR ,

I Harv and E Harv respectively.The energy net gain depends on the capacitance value excursion

and the battery voltage. Thus, other than storing energy, the batteryrole is to hold constant the voltage across the capacitor. Otherwise

a 1e200 pF variation of the capacitor value, amplies the initialvoltage across it by a factor of 200 which can surpass the break-down limits of most modern CMOS technologies. The experimental

results showed a net energy harvesting gain of 6.9 nJ/cycle at200 Hz.

The harvester circuit was fabricated with the AMI semi-conductor 0.5 mm CMOS technology except for the inductor, the

Fig. 13. The electromagnetic generator (left) and its cross section (right) [32].

Fig. 14. Output voltages for different resonator shapes at 1 MU load resistance [32].

Fig. 15. Output power for different resonator shapes at resonance [32].



6/14

state detectors and part of the timer that were kept off chip forexperimental exibility. A 3 V supply emulated a moderately

charged Li-Ion battery (V BAT).Voltage-constrained energy harvesting system provides more

energy than the charge-constrained case. However, the majorobstacle for this approach is that it would require an additional

voltage source to hold the voltage. A hybrid alternative is proposedin Ref. [28] where a second capacitor of constant value C par is addedin parallel to the variable energy harvesting capacitor (C MEMS inFig. 5a). Analysis shows that as C par approaches innity, the charge

available approaches that available through voltage-constrainedsystem. The variable capacitor is implemented using MEMS tech-nology (Fig. 6). It consists of three basic parts: a oating mass,a folded spring (one per side), and two sets of interdigitated combs,

one per side. Each spring consists of four spring bars, a free rigid

beam, and a rigid anchor. The spring bars are connected to both theanchor and the free beam, limiting the motion of the mass to onedimension (perpendicular to the gure plan). The interdigitated

combs form two variable capacitors by connecting one terminal to

the moving mass at the anchors and the others to each of thestationary combs. This device’s wafer will be supported by anidentical silicon handle wafer. The two wafers and the separationoxide form the parasitic capacitor C par. An analysis of the capaci-

tance with the comb structure fully closed yields a capacitormaximum value of 260 pF and minimum value of 2 pF with thecomb structure fully open. The analysis and design of the control

and power electronics is based in these two values witha maximum gap voltage of 8 V. The systemworks as follows (Fig. 5aand b): At startup, the capacitor combination of C par and C MEMS hasno voltage across it, so V C ¼ V DD. At the beginning of t 1, a controller

trigger the conversion process. During t 1, SW2 is on, SW1 is off, andthe inductor current increases. At t 2, SW2 is off, SW1 is on, and the

inductor transfers energy to the capacitor. During t 3, both switchesare off and the variable capacitor plates move. This time constant is

near 400 ms while the resonant on time of the switches is

approximately 600 ns. It is therefore a reasonable approximation tosay that the MEMS capacitor value is constant during t 1, t 2, t 4 and t 5.During t 3 the plates move from their minimum separation(C max þ C par) to their maximum separation (C min þ C par). The

mechanical energy has moved the plates apart and caused thevoltage across the capacitor combination to reach a maximum, andthe energy harvesting is performed. During t 4 SW1 is on, SW2 is off,and the capacitor combination transfers energy to the inductor.

Note that this LC time constant is smaller than t 2 because theoverall capacitor value has decreased. Once the capacitor voltagereaches zero, corresponding to one-quarter of the resonant period

of the LC, SW1 is turned off, SW2 is turned on, and the energy putinto the inductor is transferred to the reservoir during t 5. This

process repeats at the frequency of the mechanical vibration, whichcorresponds to variations in C MEMS. The system controller has beenfabricated in a 0.6 mm CMOS process.

In order to achieve the highest possible power density from an

inertial energy scavenger, the authors of Ref. [29] found it isnecessary to optimize the damping force under a given operation

Table 1

Experimental results for three different resonators [32].

Resonat or Resistance[U]

Electricalresonance [kHz]

Load resistance[U]

Mechanicalresonance [Hz]

Qualityfactor Q

Power atresonance [mW]

Voltage atresonance [mV]

Shape A 14.1 4910 65 127 31.7 265 185.7

Shape B 27.1 2420 76 102 20.4 290 183.2

Shape C 9.8 13950 47 98.2 27.2 229 148.5

Fig. 16. Step-down dcedc conversion circuitry [33].

Fig. 17. Theoretical and experimental optimal duty cycles for step-down converter as

a function of excitation [33]. Fig. 18. Maximum RMS power against displacement [34].



7/14

condition (size of the relative motion between the mass and theframe). In an electromagnetic generator (i.e. magnetic transducer)

for which the source vibrates with an amplitude Y o and drivesa VDRG with an internal displacement limit of Z l, the optimaldamping will be that which just allows the maximum internal

displacement ( Z l). The optimal damping factor is given by:

z ¼1

2

Y 0 Z l

As per the power processing circuit, it is advantageous to operate itat relatively high voltage (above 1 V) to reduce power loss inswitches. The authors proposed and veried by simulationa compromise between the inductance’s number of turns, its size,

the resistive loss, the volume of the permanent magnet, and theoutput voltage as the main challenge in such circuit is the lowoutput voltage of the transducer (w195 mV). The used dual polarityboost circuit separately process the positive and negative half

cycles of the generated voltage (Fig. 7). It is proposed to use it in

discontinuous synchronous conduction mode to reduce power lossin the switches (MOSFETs). An optimal ef ciency of about 50% has

been reached in several operating conditions for an output powerof 50 mW at 1.65 V (Fig. 8).

The authors used the converter depicted in Fig. 9 for constant-charge electrostatic microgenerators [29]. The trade off in opti-mizing the overall ef ciency depends on several parameters:

because of the high voltage produced by the transducer, a consid-erable amount of energy is stored in the depletion layer parasiticcapacitance of the blocking junction of the high-side MOSFET. Thisenergy is lost when the MOSFET is turned on. Decreasing the size of the MOSFET will, from one side decrease the parasitic capacitor and

hence the losses, and from the other side degrade the ef ciency of the converter (bad switch). The simulation shows a possibility tolocate an optimum switch size that maximizes the ef ciency of theconverter. The authors found that a multiple of 10 of a unity tran-

sistor provides the best ef ciency for the modied buck converter(Fig. 10).

A magnetic spring generator is described in Ref. [30]. A free to

move permanent magnet is placed inside a tube, and two othermagnets are xed at both ends of the tube in such a way that the

facing surfaces of the xed and moving magnets have the samepolarization. Finally, a coil is wrapped around the outside of the

Fig. 19. (a) the rectifying circuit. (b) The rectied output [35].

Fig. 20. Illustration of the generator that realizes the frequency upconversion: (upper left) isometric, (upper right) side, and (lower) schematic views [38].



8/14

tube (Fig.11). When the tube moves, the middle magnet vibrates upand down and a voltage will be induced in the coil. Two different

versions were prototyped and tested while placed inside a rucksackof walking/slow running person. For the higher ef ciency version

(one xed and one moving magnets), the average measuredmaximum load powers of the generators were 0.95 and 2.46 mWduring walking and slow running condition, respectively. Theoutput voltage is presented in Fig. 12. Using a simple diode

capacitor rectier, the prototype was able to transfer 3.54 J toa rechargeable battery in 1 h. This energy level is enough to powerlow-power sensor modules without battery [31].

An electromechanical power generator to convert vibrations toelectrical energy with electromagnetic transducer is proposed inRef. [32]. The transducer is shown in Fig. 13. The resonator,

a moveable planar inductor that can move with vibration, is xedby its external edge at the cylindrical case and it is placed

symmetrically between two sets of magnets (Fig. 13, left). In orderto improve the distribution of the magnetic eld density in the area

between the magnets, the two central magnets have oppositemagnetization vectors, while the two external magnets haveopposite magnetization vectors, but inverted than the other two(Fig. 13, right). Output voltages at resonance in the range of

150e185 mV (Fig. 14) and an output power in the range of 230e290 mW (Fig. 15) have been experimentally obtained for threedifferent resonators A, B, and C (Table 1).

An optimized method of harvesting vibrational energy witha piezoelectric element using a step-down dcedc converter ispresented in Ref. [33]. Analysis of the converter in discontinuous

Fig. 21. Simulation and measurement results for a single cantilever of the FupC design [38].

Fig. 22. (a) Parametric frequency increased generator (PFIG), (b) Theory of operation e

the generator is depicted at three instances of time during an incident displacement [41].



9/14

current conduction mode results in an expression for the step-down converter duty cycle-power relationship. For a vibrating

piezoelectric element modeled as a sinusoidal current source I psi-n(ut ) in parallel with its electrode capacitance C p (Fig. 16), it wasfound that the optimal duty cycle for maximum harvested poweris:

Dopt z

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi4uLC p f s

p

r

Where L is the inductance and f s is the switching frequency. Thecircuit has been implemented with the accompanying controlcircuitry, i.e. switch gate drive, pulse generator, duty cycle gener-ator and threshold control. Fig. 17 shows the theoretical optimal

duty cycle and the experimentally determined optimal duty cycleas a function of the mechanical excitation (characterized by thepiezoelectric element’s unloaded or open-circuit voltage). Over therange of excitation considered, up to 100 V, the two curves follow

a similar trend, both becoming nearly constant above excitations of 45 V.

A prototyping of an inertial linear electromagnetic millimeterscale generator is reported in Ref. [34]. Fig. 18 shows the experi-

mental output power at resonance against source vibration

amplitude. This device is capable of generating 0.3 m

W at an exci-tation frequency of 4 MHz. Another small scale vibration-inducedpower harvester with total volume of about 1 cm3 is presented in

Ref. [35]. With the quadrupler rectifying circuit of Fig.19, the loadedgenerator is able of producing 1.3 V DC from 4 V, 60e120 Hz inputcorresponding to w200 mm input vibration. The power output forthis system is w100 mW.

More focus on piezoelectric and electromagnetic transducerfabrication to harvest from kinetic energy can be found inRef. [36,37].

4.2. Harvesting from low-frequency vibrations

Astheef ciency of vibration-based harvesters is proportional to

excitation frequency, Sari et al. [38] proposed a generator thatconverts low-frequency environmental vibrations to a higherfrequency by employing the frequency upconversion (FupC) tech-nique. The generator is depicted in Fig. 20. It is composed of two

mechanical structures: a magnet attached to a diaphragm thatresonates with ambient frequency (in the range of 1e100 Hz), and

an array of cantilevers (resonance frequency in the range of 2e3 kHz) located right below the diaphragm. At the tip of each

cantilever, nickel is electroplated for interaction with the magnet.As the diaphragm resonates in response to external vibrations, themagnet catches the cantilevers at a certain instance of its move-ment, pulls them up, and releases them at another point. The

released cantilevers start resonating at their high damped naturalfrequency with the given initial condition, realizing the FupC. Themotion of the released cantilevers exponentially decays out, and

before it completely dies, the cycle starts again.The authors derived a mathematical model for the output

voltage and the harvested power. Fig. 21 shows the simulated andmeasured output voltages from a single cantilever. In thegure, the

catch and release points of the cantilevers and the peak voltageoutput are also indicated. Voltage generation is realized right afterthe release of the cantilevers where they oscillate with theirdamped natural frequencies. The time in which output voltage is

almost zero, depends on the ambient frequency (95 Hz in this case),and damping factor of the cantilevers oscillations.

Fig. 23. Full-wave rectier circuit used to store the charge provided by a pyroelectric

cell [43].

Fig. 24. Experimental values of PZT pyroelectric cell I , T and calculated dT /dt over the

time when switching on (temperature increases) and off (temperature decreases) an

air dryer [43].

Fig. 25. Experimental data using the circuit depicted in Fig. 23 when consecutive

heating/cooling cycles were applied to a PZT cell [43].

Fig. 26. A boost converter using digitally controlled FETs [45].



10/14

The experimental result shows that from a natural diaphragmfrequency of 113 Hz, the system is able to produce a peak power and

peak voltage of 0.25 nW and 0.57 mV respectively per cantilever.This is to compare with the popular traditional large mass coil(LMC)-type [39,40] electromagnetic energy harvester that, scaledfor 113 Hz, produces a peak power and peak voltage of 0.23 nW and

0.24 mV respectively.

A non-resonant inertial generator for harvesting low-frequencyvibrations is presented in Ref. [41] and its principle is shown in

Fig. 22. The Parametric Frequency Increased Generator (PFIG) isdesigned to accommodate the large amplitudes associated withlow-frequency vibrations [42]. Two Frequency Increased Genera-tors (FIGs) are placed on either side of an inertial mass, oriented to

face each other (Fig. 22a). Attached to the bottom of the FIG spring

Fig. 27. Left: The principle of the thermoelectric microsystem proposed by [48]. When the heat ows across the pn junction, an electrical power current is generated by the Seebeck

effect. Right: Schematic of the step-up circuit.

Fig. 28. Smart RF energy harvesting system [53].

Fig. 30. Power generated at variable distance from a 13 dBm RF source [53].Fig. 29. A 3-stage Schottky diode based Villard voltage multiplier circuit [53].



11/14

is a magnet for power generation, while on top, a smaller magnet(actuation magnet) is used to generate a magnetic force in order tolatch the FIG and the inertial mass together. The operation of thePFIG is outlined in Fig. 22b. The generator operates such that the

inertial mass snaps back and forth between the two FIG generators,attaching magnetically. As the inertial mass moves, it pulls the FIGspring along. When the inertial mass approaches the opposing FIG,the magnetic force of attraction begins to increase. As the forces on

the FIG/inertial mass system overwhelm the holding magneticforce, the inertial mass detaches and is pulled to the opposing FIG.The freed device now resonates at its high natural frequency con-

verting the stored mechanical energy in its spring, to electrical. Thisprocess is subsequently repeated in the opposite direction. Witha total volume of 3.74 cm3, an inertial mass of 9 g, and an inputfrequency of 10 Hz, the harvested energy is 288 mW.

4.3. Harvesting from thermal sources

Energy harvesting from pyroelectric cells based on screen-printed PZT (lead zirconate titanate) and PVDF (polyvinylidene

uoride) lms and subject to temperature variation has beenreported in Ref. [43]. The pyroelectric cell is modeled with currentsource I p, capacitor C p and resistor Rp (Fig. 23), with

l p ¼ S ldT

dt

where S is the electrode surface of the cell, l is the pyroelectric

coef cient and T is the temperature. Fig. 24 shows the measuredcharacteristic of the generated current when temperature changes.The current changes direction during free cooling off (temperaturedecrease). As the current is alternative during heating up and

cooling down, a diode based full-wave rectier has been used to

store energy (Fig. 23).The experimental results of Fig. 23 for C L ¼ 1 mF, shows how the

voltage increases when consecutive heating/cooling cycles were

applied to a PZT cell (Fig. 25). As it can be seen from Fig. 25, circuitlosses increased with output voltage increase. The output voltage

reached a maximum of 21.5 V for the measured cell. Two parallelcells were able to produce a maximum voltage of 31 V and provideenergy of 0.5 mJ. With these results, it is possible to power a low-power RF transmitter [44]. PVDF cells produce less energy because

of their low pyroelectric coef cient.

Carlson et al. presented a switched mode DCe

DC boostconverter with digital control for thermal source energy harvester[45]. The boostconverter is shown in Fig. 26. The control is based on

turning the pFET off when the inductor current falls to zero. Thecontrol circuit that is based on the status of V D (high or low) isalmost free of analog circuitry and is thus free of static powerdissipation. The whole circuit except for an inductance and two

ltering capacitors was fabricated with a 0.13 mm CMOS process.The circuit was then tested with body heat from a human arm. Theconverter was allowed to reach thermal equilibrium over a period

of 10 min. At equilibrium, the generator produced 34 mV (unloadedvoltage) and the boost converter was able to deliver 34 mW at 1 V.

Using the Seebeck effect [15,16,46,47], a thermoelectric micro-converter for energy scavenging systems that can supply individual

electroencephalogram (EEG) modules was fabricated using thinlms of bismuth and antimony tellurides (Fig. 27 left). Witha simple step-up circuit (Fig. 27 right), a 1 cm2 thermoelectricmicroconverter provided a power of about 18 mW [48].

In analogy with electrical matching, Leonov and Fiorini devel-oped equations for thermal matching of thermoelectric energyharvester [49]. Thermal matching is required to maximize theharvested power and serves in the design of TEGs (thermoelectric

generators). More details on TEGs design, fabrication, and testingcan be found in Ref. [50,51]. In Ref. [50], D.T.S., a German companypresented a prototype of its product, the Low-Power Thermoelec-tric Generator (LPTG). Its manufacturing process is fully compatible

with standard microelectronic technologies with power output of a few 10 mW.

4.4. Harvesting from RF sources

An energy harvesting system performance using multiple RFsources was demonstrated in Ref. [52,53]. In Ref. [52], multiple

energy harvesting antennas in one area are proposed and studied toincrease the power/area ratio. It was shown that an increase of 83%in area resulted in 300% increase in power. Jabbar et al. proposedSchottky diode and CMOS process compatible based version of

a 2-antennas RF energy harvesting systems shown in Fig. 28 [53].The former uses a 3-stage Schottky diode Villard voltage multipliercircuit (Fig. 29). The system was tested with different RF frequenciesat different source powers and distances. C tune was tuned with

respect to each antenna to get the maximum output power. Fig. 30

Fig. 31. Improved Villard voltage multiplier circuit [53].

Table 2

Output voltage comparison of traditional work and circuit simulated in Ref. [53].

Input power (mW) Output power (mW) % Increase over traditional

Traditional This Work

316.23 8.43 12.39 47.06%

562.34 45.12 94.19 108.79%1000 133.85 347.03 159.26%

1778.28 282.58 495.75 75.44%

3162.28 495.75 743.63 50%

Fig. 32. System architecture [56].



12/14

shows the output power for a constant source power of 13 dBm anddistance between the source and circuit is varied from 1 to 12 cm.The 2.4 GHz circuit response was not in a straight line due to the

presence of wireless Access Points in thebuilding. The CMOS version

uses an improved version of the NMOS traditional Villard voltage

multiplier described in Ref. [54]. The improved version is depicted inFig. 31. It uses only one self-biasing circuit, consisting of R1, R2 and C 1for both the transistors as compared to two in Ref. [54]. Simulationusing the TSMC (Taiwan Semiconductor Manufacturing Company)

0.25 mm CMOS technology showed a higher output voltage withrespect to the traditional circuit. The output power increase rangefrom about 50% (for low and high input power) to about 160% formedium input power (Table 2). As increased input RF energy results

in higher output voltage, the transferred power becomes larger thanthe circuit loss and ef ciency increases. However, at high inputenergy, the diode connected MOSFET’s leakage current increases and

the ef ciency decreases hence creating an optimal value of theef ciency. This is a consequence of larger output voltage resulting in

larger bias voltage on the gate preventing the diode connectedMOSFET’s from ever fully turning off.

A study of reception and rectication of broad-band statisticallytime-varying low-power-density microwave radiation is presented

Fig. 33. The DC/DC converter architecture [56].

Fig. 34. RF converter architecture. Vbandgap is the same as the 570 mV of Fig. 33 [56].

Fig. 35. Micro-battery voltage monitor. When Act goes low, the comparison between

V BAT and 1.5 V is activated, and if V BAT<

1.5 V (at about 11,500 s), soc

ag goes low [56].

Fig. 36. Micro-battery charger test result. When V BAT reaches 2.75 V, soc becomes

“high

” at the next Act [56].



13/14

in Ref. [55]. A 64-element dual-circularly-polarized spiral rectenna

array is designed and characterized over a frequency range of 2e18 GHz with single-tone and multitone incident waves. Theintegrated design of the antenna and rectier eliminates matching

and ltering circuits, allowing for a compact element design. Therectied dc power and ef ciency is characterized as a function of dcload and dc circuit topology, RF frequency, polarization, and inci-dence angle for power densities between 105e101 mW/cm2.

4.5. Harvesting power management

Lhermet et al. [56] have fabricated a power generator unitincluding two micropower sources (radio frequency RF andThermal) and their management integrated circuit (IC) (Fig. 32).

The RF and thermal microgenerators, the micro-battery and the

L e

C power lter of the DC/DC converter are not integrated withinthe IC. The raw input power from the thermogenerator is processedby a power switch followed by the L eC power lter, yielding the

conditioned output power. The control system varies the switchduty cycle to cause the output voltage to follow a reference voltageof 570 mV from bandgap (Fig. 33).

The RF/DC converter is less complicated. It is composed of

a limiter, a rectier and a control loop to provide a stabilized DCoutput (Fig. 34).

As the micro-battery can be charged by either converter,therefore, the power supply manager manages priority between

the two sources when they are simultaneously present and acti-vates self-powered micro-battery protection in the case of externalpower source interruption.

The power manager compares the micro-battery voltage toa reference voltage every hour and half (an internal digital signalAct goes “low”) and sets an internal digital signal (soc) to “low”when micro-battery is discharged (


14/14

variable-capacitive resonator for power-free LSI. Proceedings of InternationalSymposium on Low Power Electronics and Design (ISLPED); 2003:193e8.

[11] Keawboonchuay C, Engel TG. Maximum power generation in a piezoelec-t ric pulse g ener at or . IEEE Transactions o n Plasma Scie nce2003;31(1):123e8.

[12] Yang Jiashi, Chen Ziguang, Hu Yuantai. An exact analysis of a rectangular platepiezoelectric generator. IEEE Transactions on Ultrasonics, Ferroelectrics andFrequency Control 2007;54(1):190e5.

[13] Mateu Loreto, Moll Francesc. Review of energy harvesting techniques andapplications for microelectronics. Proceedings of the SPIE Microtechnologies

for the New Millenium 2005;5837:359e

73.[14] Stevels A, Jansen AJ. Renewable energy in portable radios: an environmental

benchmarking study. The Journal of Sustainable Product Design; January,1998:51e5.

[15] Justin R. Farmer. A comparison of power harvesting techniques and relatedenergy storage issues M.S Thesis. Dept. Mech. Eng., Virginia PolytechnicInstitute and State Univ. Blacksburg, VA; 2007.

[16] Blatt FJ, Schroeder PA, Foiles CL, Greig D. Thermoelectric power of materials.New York: Plenum Press; 1976.

[17] Nolas GS, Sharp J, Goldsmid HJ. Thermoelectrics: basic principles and newmaterials developments. New York: Springer; 2001.

[18] Mah O. Fundamentals of photovoltaic materials. National Solar PowerResearch Institute, Inc.; 1998.

[19] Damaschke JM. Design of a low-input-voltage converter for thermoelectricgenerator. IEEE Transactions on Industry Applications 1997;33(5):1203e7.

[20] Myers RL. The basics of physics. Westport, Conn: Greenwood Press; 2006.[21] Elliot RS. Electromagnetics: history, theory, and applications. Piscataway: IEEE

Press; 1993.[22] Stephen NG. On energy harvesting from ambient vibration. Journal of Sound

and Vibration 2006;293:409e

25.[23] Mitcheson PD, Green TC, Yeatman EM, Holmes AS. Architectures for vibration-

driven micropower generators. Journal of Microelectromechanical Systems2004;13(3):429e40.

[24] Rocha JG, Goncalves LM, Rocha PF, Silva MP, Lanceros-Méndez S. Energyharvesting from piezoelectric materials fully integrated in footwear. IEEETransactions on Industrial Electronics 2010;57(3):813e9.

[25] Silva HR, Afonso JA, Morim PC, Oliveira PM, Correia JH, Rocha LA. Wirelesshydrotherapy smart-suit network for posture monitoring. Proceedings of IEEEInternational Symposium on Industrial Electronics, ISIE; 4e7 June,2007:2713e7.

[26] Roundy S, Wright PK, Rabaey J. Energy scavenging for wireless sensornetworks with special focus on vibrations. Kluwer Academic Press; 2003.

[27] Torres Erick O, Rincón-Mora Gabriel A. A 0.7-mm BiCMOS electrostatic energy-harvesting system IC. IEEE Journal of Solid State Circuits 2010;45(2):483e96.

[28] Meninger Scott, Mur-Miranda Jose Oscar, Amirtharajah Rajeevan,Chandrakasan Anantha P, Lang Jeffrey H. Vibration-to-electric energyconversion. IEEE Transactions on VLSI Systems 2001;9(1):64e76.

[29] Mitcheson PD, Green TC, Yeatman EM. Power processing circuits for elec-tromagnetic, electrostatic and piezoelectric inertial energy scavengers.Microsystem Technologies 2007;13:1629e35.

[30] Saha CR, O’Donnell T, Wang N, McCloskey P. Electromagnetic generator forharvesting energy from human motion. Sensors and Actuators A: Physical2008;47:248e53.

[31] Bharatula NB, Ossevoort S, Stager M, Troster G. Towards wearable autono-mous microsystems, pervasive computing. In: Proceeding of the 2nd Inter-national conference, Vienna, Austria; 2004. p. 225e37.

[32] Marioli D, Sardini E, Serpelloni M. Electromagnetic generators employingplanar inductors for autonomous sensor applications. Procedia Chemistry2009;1:469e72.

[33] Ottman G, Hofmann H, Lesieutre G. Optimized piezoelectric energy harvestingcircuit using step-down converter in discontinuous conduction mode. IEEETransactions on Power Electronics 2003;18(2):696e703.

[34] Williams C, Shearwood C, Harradine M, Mellor P, Birch T, Yates R. Develop-ment of an electromagnetic micro-generator. IEE Proceedings Part G: Circuits,Devices and Systems 2001;148(6):337e42.

[35] Li W, Ho T, Chan G, Leong P, Wong HY. Infrared signal transmission by a laser-micromachined vibration-induced power generator. In: Proceedings of the 43rdIEEE Midwest symposium on circuits and systems, vol. 1; 2000. p. 236e9.

[36] Khaligh Alireza, Zeng Peng, Zheng Cong. Kinetic energy harvesting usingpiezoelectric and electromagnetic technologiesdstate of the art. IEEE Trans-actions on Industrial Electronics 2010;57(3):850e60.

[37] Swallow LM, Luo JK, Siores E, Patel I, Dodds D. A piezoelectric ber compositebased energy harvesting device for potential wearable applications. SmartMaterials and Structures 2008;17(2):025017.

[38] Sari Ibrahim, Balkan Tuna, Külah Haluk. An electromagnetic micro power

generator for low-frequency environmental vibrations based on thefrequency upconversion technique. IEEE Journal of MicroelectromechanicalSystems 2010;19(1):14e27.

[39] Williams CB, Yates RB. Analysis of a micro-electric generator for micro-systems. Physica A 1996;52(1/3):8e11.

[40] Williams CB, Woods RC, Yates RB. Feasibility study of a vibration poweredmicro-electric generator. Proceedings of IEE Colloquim on Compact PowerSources 1996;96/107:7/1e7/3.

[41] Galchev Tzeno, Kim Hanseup, Naja Khalil. A parametric frequency increasedpower generator for scavenging low frequency ambient vibrations,. ProcediaChemistry 2009;1:1439e42.

[42] Galchev T, Kim H, Naja K. Non-resonant bi-stable frequency increased powergenerator for low-frequency ambient vibration. IEEE Solid-State Sensors,Actuators and Microsystems Conference: Transducers; 2009:632e5.

[43] Cuadras A, Gasulla M, Ferrari V. Thermal energy harvesting through pyro-electricity. Sensors and Actuators A: Physical 2010;158:132e9.

[44] Ferrari M, Ferrari V, Marioli D, Taroni A. Modeling, fabrication and perfor-mance measurements of a piezoelectric energy converter for power har-vesting in autonomous microsystems. IEEE Transactions on Instrumentationand Measurement 2006;55(6):2096

e101.

[45] Carlson Eric J, Strunz Kai, Otis Brian P. A 20 mV input boost converter withef cient digital control for thermoelectric energy harvesting,. IEEE Journal of Solid State Circuits 2010;45(4):741e50.

[46] Vining C. Semiconductors are cool. Nature 2001;413(6856):577e8.[47] Bell L. Cooling, heating, generating power, and recovering waste heat with

thermoelectric systems. Science 2008;321(5895):1457e61.[48] Carmo João Paulo, Gonçalves Luis Miguel, Correia José Higino. Thermoelectric

microconverter for energy harvesting systems. IEEE Transactions on IndustrialElectronics 2010;57(3):861e7.

[49] Leono V, Fiorini. P. Thermal matching of a thermoelectric energy scavengerwith the ambient. In: Proceedings of 5th European Conference on thermo-electrics (ECT), Odessa, Ukraine; 2007.

[50] Jovanovi V, Ghamaty. S. Design, fabrication and testing of energy harvestingthermoelectric generator. In: Matsuzaki Y, editor. Smart structures andmaterials 2006: smart structures and integrated systems. Japan: Nagoya Univ;2006. p. 142e9.

[51] Stordeur M, Stark I. Low power thermoelectric generator-self-suf cient

energy supply for micro systems. In: I EEE 16th conference on thermoelectrics.Dresden; 1997. p. 575e7.[52] Mi M, Mickle MH, Capelli C, Swift H. RF energy harvesting with multiple

antennas in the same space. IEEE Antennas and Propagation Magazine2005;47(5):100e6.

[53] Jabbar Hamid, Song Young S, Jeong Taikyeong Ted. RF energy harvestingsystem and circuits for charging of mobile devices. IEEE TransactionsConsumer Electronics 2010;56(1):247e53.

[54] Salter TS. Low power smartdust receiver with novel applications andimprovements of an RF power harvesting circuit. PhD Thesis. Department of Electrical Engineering, University of Maryland, USA; 2009.

[55] Hagerty J, Helmbrecht F, McCalpin W, Zane R, Popovi&cacute Z. Recyclingambient microwave energy with broad-band rectenna arrays. I EEE Trans-actions on Microwave Theory and Techniques 2004;52(3):1014e24.

[56] Lhermet Hélène, Condemine Cyril, Plissonnier Marc, Salot Raphaël,Audebert Patrick, Rosset Marion. Ef cient power management circuit: fromthermal energy harvesting to above-IC microbattery energy storage. IEEE Journal of Solid State Circuits 2008;43(1):246e55.


Energy Harvesting State of the Art - Harb 2010

Documents

Transcript of Energy Harvesting State of the Art - Harb 2010