Highly efficient integrated rectifier and voltage boosting circuits for

advertisement
Adv. Radio Sci., 6, 219–225, 2008
www.adv-radio-sci.net/6/219/2008/
© Author(s) 2008. This work is distributed under
the Creative Commons Attribution 3.0 License.
Advances in
Radio Science
Highly efficient integrated rectifier and voltage boosting circuits for
energy harvesting applications
D. Maurath, C. Peters, T. Hehn, M. Ortmanns, and Y. Manoli
Chair of Microelectronics, Department of Microsystems Engineering (IMTEK), University of Freiburg,
Georges-Koehler-Allee 102, 79110 Freiburg, Germany
Abstract. This paper presents novel circuit concepts for integrated rectifiers and voltage converting interfaces for energy harvesting micro-generators. In the context of energy harvesting, usually only small voltages are supplied
by vibration-driven generators. Therefore, rectification with
minimum voltage losses and low reverse currents is an important issue. This is realized by novel integrated rectifiers
which were fabricated and are presented in this article. Additionally, there is a crucial need for dynamic load adaptation as well as voltage up-conversion. A circuit concept is
presented, which is able to obtain both requirements. This
generator interface adapts its input impedance for an optimal energy transfer efficiency. Furthermore, this generator
interface provides implicit voltage up-conversion, whereas
the generator output energy is stored on a buffer, which is
connected to the output of the voltage converting interface.
As simulations express, this fully integrated converter is able
to boost ac-voltages greater than |0.35 V| to an output dcvoltage of 2.0 V–2.5 V. Thereby, high harvesting efficiencies
above 80% are possible within the entire operational range.
1
Introduction
Especially electronic devices participate the development towards miniaturization and high performance with least power
consumption only. Nowadays, very low power applications
emerge increasingly in the field of stand-alone and wireless
acting sensor/actuator-systems. A further rise of the applicability of such systems as fully embedded nodes suffers from
the inconvenience with batteries. These electrochemical energy storages have limited lifespan, are very temperature sensitive and need to be replaced repetitively. In case of hardly
accessible locations, there is little chance for having battery
power supply based cost-efficient wireless embedded systems. Thus, the idea of power supplies, which are based on
energy harvesting principles, is a helpful contribution in order to realize cost-efficient fully embedded sensor/actuatorsystems.
A very space-efficient and robust kind of energy harvesting is possible with vibration-driven inductive microgenerators. However, such micro-generators supply only
small open-circuit output voltage in the range of 2–3 Vpp .
Beside this, a relatively high internal resistive impedance
causes only a small load optimum, were the maximum possible output power is obtained. These circumstances enforce
developments of novel kinds of rectifiers, since common fullwave rectifiers cause relatively high voltage losses.
In order to circumvent these crucial limitations, the second
chapter shows enhanced rectifier circuitry with active diodes,
which were designed, implemented and fabricated. These active diodes consist of an active circuit instead of passive devices only. Additionally, due to this low voltages and high internal resistance an adaptive generator-interface is required.
This interface gives the opportunity of most appropriate loading of the generator. Since only a simple capacitive energy
buffer connected to the output of the rectifier can not provide
adequate load matching. Due to the realization of this interface with switched-capacitors, the functionality of the rectifier is improved with respect to minimum input voltages.
Conclusively, the forth chapter presents simulated as well as
measured results. So, the active diodes of the proposed rectifier without the generator interface can operate above 1.75 V
generator output voltage amplitude. Together with the interface, generator output amplitudes of less than 1 V are sufficient for operation. Whereas the rectification, matching and
conversion efficiencies can be kept above 80% over the entire
operating range.
Correspondence to: D. Maurath
(maurath@imtek.de)
Published by Copernicus Publications on behalf of the URSI Landesausschuss in der Bundesrepublik Deutschland e.V.
tion independent of the generator excitation, two concerns
are necessary. First, if the generator is excited most possible
The first stage of
vert the negative h
220
D. Maurath et al.: Highly efficient integrated rectifier and voltage boosting circuits for energy harvestingpositive
applicationsones. Th
dard CMOS trans
e.g. isolated trans
left side is realiz
livers the highest
consists of NMOS
at point B. No add
are necessary, bec
Thus, the bulk of
nected to point A
Fig. 1. Common single stage power conversion and generator interfacing approach (a). The generator can only supply power within
Fig. 1. Common single stage power conversion and generator internumber of require
conduction degree α (b).
ctifier and Voltage Boosting Circuits for Energy Harvesting Applications
Dominic
Maurath:
Highly
Efficient
Integrated
Rectifier
Voltage Boos
facing approach (a). The generator can only supply power within
Theandability
to c
conduction degree α (b).
the input to the o
mparaith this
re usuhis apcase of
ch may
gh and
Schotd 0.6V,
se curegrated
ecause
tion in
must alNMOS
A dysistors,
to this,
se four
s used,
ces the
esented
nected
s, only
onable
ery, as
considr is expplicancerns
ossible
Fig. 4. B
an additi
version c
Fig. 2. Schematic of a negative voltage
voltage converter
converter (a),
(a),the
thebulk
bulkregregulation configuration
configuration (b),
(b), and
and the
the active
activediode
diodeprinciple
principle(c).
(c).
energy
has to be harvested.
most of that maximum
2 Conventional
harvestingSecond,
interfaces
possible energy need to be stored at some energy storage element,
likegenerator
a buffer capacity.
Additionally,
in order
to harvest
Common
interfacing
configurations
are comparacontinuously
the
maximum
possible
energy,
permanent
ble to the configuration shown in Fig. 1a. Together with optithis
mal
generator
loading
for maximum
generator
output
simple
structure
also conventional
circuit
principles
arepower
usuas
as efficient
voltage
conversion
are
allywell
applied.
In caserectification,
of batteries and
as power
sources,
this aprequired
(Shengwen,
2005).
proach can
be very useful
and efficient. However, in case of
energy harvesting with micro-generators this approach may
exhibit crucial drawbacks.
3 Proposed Harvesting Circuitry
2.1 generator
Conventional
rectifier
circuits
The
interface
was
especially designed to incorporate with inductive micro-generators (Spreemann, 2006;
Conventional
full-wave
rectifiers
sufferresistance
a relatively
Kulkarni,
2007).
The internal
source
ofhigh
theseand
inunappreciated
voltages
loss
of
about
1.4
V.
In
case
of
Schotductive generators is between Ri = 1 − 5kΩ. The generated
tky diodes, that
voltage
loss could
beVreduced
to around 0.6 V,
open-circuit
output
voltage
has 2.8
pp at most and frequenhowever,
those
diodes
have
intrinsically
higher
reverse curcies are not higher then fgen = 500 Hz.
rent, which is also not acceptable. Concerning integrated
rectifiers,
theactive
diodes
are realized with MOSFETs, because
3.1
Novel
rectifier
pn junctions are often not specified in forward direction in
CMOS
processes.
Thenovel
bulksrectifier
of PMOS
must
The
main
goals of this
aretransistors
the reduction
of althe
ways
be
connected
to
the
highest
potential
and
of
NMOS
to
voltage drop over the MOSFETs and achievement of a high
the
lowest
to
avoid
leakage
currents
and
latch-up.
A
dynamic
efficiency. This rectifier can be separated into two stages: (i)
bulknegative
regulation
(Fig. 2b),
consisting
of two
the
voltage
converter
and (ii)
the transistors,
active diodeavoids
part.
this and must be added to every transistor. Due to this, the
numberFirst
of transistors
and leakage
increases.
3.1.1
stage: negative
voltage
converter (CON)
The first stage of the proposed rectifier circuit is used to conAdv. Radio Sci., 6, 219–225, 2008
vert
the negative half waves of the input sinusoidal wave into
positive ones. This conversion is done with only four standard CMOS transistors (see Fig. 2a). No special technology,
Fig. 3.
3. Schematic
Schematic of
of the
the comparator
comparator circuit with differential
Fig.
differential stage,
stage,
common-source output
output amplifier
amplifier and inverter stage.
common-source
Similarly, conventional full wave (F W) rectifiers use four
The voltageconnected
drop withas
this
rectifierEvery
is lesshalf
thanwave
10 mV
(PeMOSFETs
diodes.
is used,
ters,the
2007)
and the
negative
halfVwave
is
now
converted
into
a
but
voltage
drop
is twice
and
strongly
reduces
the
t
positive one.
But this
circuit A
cansmart
not beimprovement
used to chargeisa storavailable
output
voltage.
preage capacitor,
becauseand
the Najafi,
current2004),
direction
is not
controlled
sented
in (Ghovanloo
which
replaces
two
and current backtransistors
flow occurs.
Due
to cross-coupled
this a second stage
is
diode-connected
with
two
transisnecessary
to
realize
a
useful
rectifier.
tors. Thus, only one V voltage drop is lost.
t
3.1.2 Common
Second conversion
stage: activeconcepts
diode current barrier (AD)
2.2
The common
main function
of theassecond
rectifier
stage
to control
The
approach,
depicted
in Fig.
1, isisreasonable
thecase
current
direction.
couldelements,
be in thelike
simplest
case as
a
in
of some
energyThis
storage
a battery,
diode-connected
MOSFET. Using are
this,consida full
aconventional
power source.
However, if micro-generators
wavepower
rectifier
only one only
voltage
drop
trant over the is
ered,
canwith
be gathered
when
theVgenerator
exsistorThat
results
(CONinD).
Thistocircuit
is comparable
the full
ited.
means,
order
be able
to supply antoapplicawaveindependent
rectifier with
(FW S),
but it is two
less concerns
prone to
tion
of switches
the generator
excitation,
latch-ups
and uses
7 compared
12 transistors.
are
necessary.
First,only
if the
generator to
is excited
most possible
In order
to achieve
very high
output
voltages
and
high efenergy
has to
be harvested.
Second,
most
of that
maximum
ficienciesenergy
the diode
be replaced
another
element,
possible
needshould
to be stored
at somebyenergy
storage
elethe active
activeAdditionally,
diode works in
nearly
ideal
ment,
like adiode.
bufferThis
capacity.
orderastoan
harvest
diode, with current
flowing possible
in only one
direction
and nearly
continuously
the maximum
energy,
permanent
optino voltage
drop.
The main
differencegenerator
to an idealoutput
diodepower
is the
mal
generator
loading
for maximum
current
consumption
of the device.
complexconversion
realizationare
of
as
well as
efficient rectification,
andA voltage
an active(Shengwen
diode is presented
in (Lehmann, 2005). A control
et al., 2005).
required
circuit is used to determine the conducting and non conducting time of a MOSFET used as a switch. (Lam, 2005) needs
www.adv-radio-sci.net/6/219/2008/
two active diodes to realize
a full wave rectifier. The active
diode presented in this paper is simpler and based on a fast
and low power comparator circuit (see Fig. 2c). MP is the
switch which is controlled by the comparator. If the voltage
can be
back flo
imum v
simply b
amplifie
the cath
compar
input w
less than
voltage
the nega
tive dio
3.2
Ca
In order
the pow
Fig. 4)
be optim
the gen
first stag
ciently
stage is
ficient e
the buff
the com
slowly c
ond stag
The h
a voltag
tation,
convers
the pow
ed Rectifier and Voltage Boosting Circuits for Energy Harvesting Applications 3
D. Maurath et al.: Highly efficient integrated rectifier and voltage boosting circuits for energy harvesting applications
221
Fig. 4. Building blocks of the two stage approach, which is includes an additional load-adaptive generator interface into the power conversion
chain.
Fig. 4. Building blocks of the two stage approach, which is includes
3
an additional load-adaptive generator interface into the power conProposed harvesting
circuitry
3.1.2 Second stage: active diode current barrier (AD)
version
chain.
The main function of the second rectifier stage is to control
The generator interface was especially designed to incorthe current direction. This could be in the simplest case a
porate with inductive micro-generators (Spreemann, 2006;
conventional
diode-connected current
MOSFET. Using this, a full
Kulkarni
et al., 2007).
Thebe
internal
source resistance
of these
h differential
stage,
can
transferred
to the
capacitor
and additionally
wave
rectifier
with
only
one
voltage
drop Vt over the traninductive generators is between Ri =1−5 k. The generated
ge.
back
flow
occurs
after
the
input
voltage
has
passed
the
maxsistor
results
(CON
D).
This
circuit
is
comparable to the full
open-circuit output voltage has 2.8 Vpp at most and frequenwave
rectifier
with
switches
(FW
S),
imum
value.
is it is less prone to
cies are not higher then
fgen =500
Hz. The implemented comparator (see Fig. 3) but
latch-ups and uses only 7 compared to 12 transistors.
simply based on a differential stage,
a common-source output
In order to achieve very high output voltages and high ef3.1 Novel active rectifier
than 10 mV (Peamplifier and an inverter. The input
vin,DC
connected
ficiencies
the diodeis
should
be replacedtoby another element,
the
active
diode.
This
active
diode works nearly as an ideal
w converted
a of this
thenovel
cathode
(Fig.
2c)ofand
The into
main goals
rectifier (C)
are the
reduction
the is also the power supply of the
diode, with current flowing in only one direction and nearly
voltage
drop over the
MOSFETs and v
achievement
of a high
d to charge
a storcomparator.
the alternating
voltage
drop. The main sinusoidal
difference to an ideal diode is the
in,AC is connectednoto
efficiency. This rectifier can be separated into two stages: (i)
current
consumption
of
the
device. A complex
realization of
is not controlled
wave
The overall current consumption
is
the negative voltageinput
converter
and (ii)(anode
the active(A)).
diode part.
an active diode is presented in Lehmann and Moghe (2005).
a second stage is
less than 1µA for an input voltage
of ± circuit
2.75Vis used
. The
operation
A control
to determine
the conducting and non
3.1.1 First stage: negative voltage converter (CON)
conducting
time
of
a
MOSFET
used
voltage range is between 1.6 V and 3.75 V . Combinationasofa switch. Lam et al.
(2005) needs two active diodes to realize a full wave rectifier.
negative
converter
(CON)
2a) with
the
ac-is simpler and based
The first stage of thethe
proposed
rectifiervoltage
circuit is used
to conThe active(Fig.
diode presented
in this
paper
vert the negative half
waves
of the input
wave into
arrier (AD)
on a mentioned
fast and low power
comparator circuit (see Fig. 2c). MP
tive
diode
(Fig.sinusoidal
2c) overcomes
the
drawback.
positive ones. This conversion is done with only four stanis the switch which is controlled by the comparator. If the
dard CMOS transistors (see Fig. 2a). No special technology,
voltage at the anode is higher compared to the cathode the
tage is toe.g.control
isolated transistors,
to realize generator
this circuit. The
3.2 is needed
Capacitive
interface
output and
of theconverter
comparator is 0 V and MP is on. If the cathode
left
side
is
realized
with
PMOS
transistors
and
always
deis
higher
the
output
is high and MP is off. The supply voltage
e simplest case a
livers the highest potential at Vin to node A. The right side
of the comparator is taken from the storage capacitor. Using
Using this,
a full
In orderandtotheharvest
continuously
highest
consists
of NMOS transistors
low voltage
potential is
a the
PMOS
transistorpossible
as switch noenergy,
additional start-up circuit is
at
point
B.
No
additional
dynamic
bulk
regulation
transistors
necessary
for
this
active
diode.
Vt over the tranthe power electronic is again separated in two stages (see
are necessary, because this is inherently given in this circuit.
The most important part of the active diode is the comparable toThus,
thethe
full
Fig.
The can
firstbe stage
is a generator
interface
to consumption is
bulk of the
PMOS4).
transistors
directly conparator. A fast
comparatorand
with ahas
low power
point A and
NMOS to point
B. This
reduces the
t is less nected
proneto to
needed. If
the comparator
tooto
slow
not all available energy
betheoptimized
for
adaptive
impedance
matching
soisas
load
number of required transistors.
can
be
transferred
to
the
capacitor
and
ansistors.
the generator as ideal as possible.
In addition to that, thisadditionally current
back flow occurs after the input voltage has passed the maxThe ability to convert nearly the entire voltage applied at
ages andthe
high
firstis the
stage
has oftothis
transfer
harvested
energy
mostcomparator
effi- (see Fig. 3) is
imum
value. The
implemented
inputefto the output
mainalso
advantage
circuit. the
The
voltage
drop
with
this
rectifier
is
less
than
10
mV
(Pesimply
based
on
a
differential
stage,
a common-source outanother element,
ciently into a buffer. Thereby, the
output voltage of the first
ters et al., 2007) and the negative half wave is now converted
put amplifier and an inverter. The input vin,DC is connected to
nearly as into
an aideal
neither
regulated
by the
ef-power supply of the
positive one. stage
But this is
circuit
can notfixed
be used nor
to charge
the cathode
(C) converter.
(Fig. 2c) and isFor
also the
a
storage
capacitor,
because
the
current
direction
is
not
concomparator.
v
is
connected
to
the
alternating sinusoidal
ection and nearly
ficient energy transfer, the output voltage in,AC
is rather adapted to
trolled and current back flow occurs. Due to this a second
input wave (anode (A)). The overall current consumption is
ideal diode
buffer
of V. The operation
stageisis the
necessary tothe
realize
a usefulvoltage
rectifier. Vb (t). With appropriate
less than 1 µA fordimensioning
an input voltage of ±2.75
lex realization of
the components, the buffer voltage will vary only little and
Adv.the
Radio
Sci., 6, 219–225, 2008
2005). Awww.adv-radio-sci.net/6/219/2008/
control
slowly compared to the generator ac voltage. Thus,
secand non conductond stage is realized as a dc/dc converter.
Lam, 2005) needs
The here presented generator-interface works basically as
to obon ratio
Thus, if
adapted
1/fgen .
rticular
d cause
educed
urrents
pacitive
converly only
generanterface
edance
Pmax
capacitage of
cycle.
matching
situation
where Plosses
=to R
= Pmax
another one
causes intrinsic
series
gen (Ridue
load ) resistances
the input amplitude is shown in Fig. 10. The use of an active
Generally,
by application
of shows
the proposed
harvesting
2006).Therefore,
But by minimizing
the initial
voltage
beis(Gobbi,
achieved.
for optimal
operation
the gap
capacidiode in a half
wave rectifier
a strongenergy
increase
of the
circuit
concepts
more
energy
extracted
out of the
same
tween
Vstack and
the charge
are minimized.
tors
charging
levelVbshould
staytransfer
aroundlosses
a optimal
voltage of
efficiency
(HW vs.
HW
AD). can
Thebe
highest
efficiencies
could
222
D. optimal
Maurathconversion
et al.: Highly
integrated
voltage boosting circuits
for energy
harvesting
The power
and voltage
lossesapplications
are reduced
Hence, the
ratioefficient
ki is obtained
by rectifier andmicro-generator.
and the load matching condition is improved.
ki =
Vb
Vb
=
Vch,opt + ∆Vch
Vstop
(1)
The strategy for stacking is that a maximum quantity ni →
kmax of capacitors is used, whereas within each stack-stage
ni / ki capacitors are connected in parallel (see Fig. 5). Thus,
the maximum capacitance for a certain conversion ratio is
used, which reduces the conversion frequency fconv .
tion wit
example
4.2 In
Within a
cluded a
Because
are two
ηT x =
ηhvst =
4
Results
Generally, by application of the proposed energy harvesting
circuit concepts more energy can be extracted out of the same
micro-generator. The power and voltage losses are reduced
and the load matching condition is improved.
Fig.5.
5. Schematic
Schematic and
and state
state diagram
diagram of
Fig.
of aa conversion
conversion cycle
cycle cycle.
cycle.
Fig.7.7. This
This graph
graph shows
shows one
Fig.
one conversion
conversion cycle,
cycle,whereas
whereasVVarray
is
array is
thehighest
highestvoltage
voltagewithin
within the
the array.
array.
the
slowly compared to the generator ac voltage. Thus, the second stage is realized as a dc/dc converter.
Fig. 6. This plot shows the transient power transfer while an empty
The here presented generator-interface works basically as
capacitor is charged.
a voltage converter with adaptive input impedance adaptation, efficient energy transfer and implicit voltage upconversion. By applying power saving circuit techniques,
the power consumption of the control circuitry and switch
driving should be less then Ptotal ≤50 µW.
3.2.1
Fig. 6. This plot shows the transient power transfer while an empty
capacitor is charged.
voltage range is between 1.6 V and 3.75 V. Combination of
the negative voltage converter (CON) (Fig. 2a) with the active diode (Fig. 2c) overcomes the mentioned drawback.
3.2
Capacitive generator interface and converter
In order to harvest continuously the highest possible energy,
the power electronic is again separated in two stages (see
Fig. 4). The first stage is a generator interface and has to
be optimized for adaptive impedance matching so as to load
the generator as ideal as possible. In addition to that, this
first stage also has to transfer the harvested energy most efficiently into a buffer. Thereby, the output voltage of the first
stage is neither fixed nor regulated by the converter. For efficient energy transfer, the output voltage is rather adapted to
the buffer voltage Vb (t). With appropriate dimensioning of
the components, the buffer voltage will vary only little and
Adv. Radio Sci., 6, 219–225, 2008
Adaptive conversion principle
The proposed adaptive switching converter is based on two
capacitor arrays, which toggle complementary-phased with
frequency fconv between a charging and a transfer state, Sc
and St , respectively (see Fig. 5). Whereas, a conversioncycle is the single succession of Sc and St . In the considered
converter both arrays consist of kmax =6 capacitors, whereas
kmax directly corresponds to the maximum required voltage
conversion ratio. However, the number of array capacitors
ni , which are applied within a single conversion-cycle, is dynamically adapted according to the current required conversion ratio ki =Vb /Vgen (t). Hence, while one array is in charging state Sc , ni array capacitors out of the kmax available array capacitors are connected to the generator in parallel. At
this state, the capacitors are charged until the capacitor voltages Varray reach a certain voltage limit Vstop , as shown in
Fig. 7. Afterwards, the array transitions to transfer state St .
Therefore, the array capacitors have to be re-configured so
as to obtain a stacked voltage Vstack , which is higher than
the buffer voltage Vb of an external buffer. In order to obtain the dynamic adaptation of the optimal conversion ratio
ki , the converter works in an oversampling manner. Thus,
if fgen fconv . is given, the conversion ratio can be adapted
quite continuously within a generator voltage period 1/fgen .
3.2.2
Adaptive impedance matching
High internal generator resistance Ri requires a particular
converter design since an exceeding current load would cause
a dramatic voltage breakdown, which would cause reduced
www.adv-radio-sci.net/6/219/2008/
Wher
pares th
ated Rectifier
and Voltage Boosting Circuits for Energy Harvesting Applications 5
D. Maurath et al.: Highly efficient integrated rectifier and voltage boosting circuits for energy harvesting applications
223
ximum output voltncy ηrect , which is
input energy over
put voltage is espeh inductive micro-
implemented with
5 process from ausage and Fig.
efficiency
8. The layout of the active diode enhanced rectifier, whereas the dimensions are 75×40 µm.
Fig. 8. The layout of the active diode enhanced rectifier, whereas
ncy, storage capacthe dimensions are 75 × 40µm.
ansient behavior
of power. Due to the low output currents of 4 Results
generator output
harvesting generators (50−500 µA) capacitive voltions the energy
frequency
Generally, by application of the proposed energy harvesting
age conversion is preferred rather than inductive converd a loadsion.
of However,
20 kΩ charging a capacitor allows generally only at circuit concepts more energy can be extracted out of the same
micro-generator. The power and voltage losses are reduced
topt ≈ ln(2)
maximum power transfer from the generator into
capacitance
value
and the load matching condition is improved.
the capacitor (see Fig. 6). The generator-interface matching
e. But this
value
is
situation at topt is equivalent to the impedance matching situperformance
of thePgen (Ri =Rload )=Pmax is achieved. Therefore, 4.1 Rectification efficiencies
ation where
for optimal operation the capacitors charging level should
stay around a optimal voltage of Vch,opt ≈Vgen (t), which corresponds to topt . This means, within Sc the generator charges
the capacitors up to Vstop =Vch,opt +1Vch . Then the capacion
tors are toggled to St and discharged onto an output buffer Cb
until the capacitor voltage reaches , 1Vch and 1Vdisch . The
ve form of
the dis-of keeping the caps only in a small charging
consequence
range
of 0.5 Vgen (t)±1V requires an increased number of
rcuit input voltage
capacitors. Thus, twice as many capacitors are implemented
verter in tocombinaboost the generator voltage up to a certain voltage.
arly the same voltVoltage conversion and array configuration
fiers with3.2.3
the active
arly the input voltDue to the fact, that charge transfer from one capacitor to
AD rectifier
is one
alsocauses intrinsic losses due to series resistances
another
et al., 2006). But by minimizing the initial voltage
d and the(Gobbi
capacitor
The main aspects of rectifiers are the maximum output voltage which can be reached and the efficiency ηrect , which is
simply the quotient of the output and the input energy over
one period. The maximum achievable output voltage is especially important for energy harvesting with inductive microgenerators.
The circuits have been simulated and implemented with
Cadence using a 0.35 µm HV CMOS C35 process from austriamicrosystems (see Fig. 8). The voltage and efficiency
values strongly depend on the used frequency, storage capacitor and ohmic load. Figure 9 shows the transient behavior of the discussed rectifiers. For these simulations the frequency is 125 kHz, a capacitance of 500 pF and a load of
20 k are used. For useful applications a larger capacitance
value should be used to reduce the voltage ripple. But this
value is more suitable for showing the excellent performance
of the proposed rectifiers.
gap between Vstack and Vb the charge transfer losses are minFig. 9.conversion
Simulation
of the output signal of the compared recimized. Hence, the optimal
ratioresults
ki is obtained
by
voltage
comparison
tifiers. The output voltage range of4.1.1
CONRectifier
D is inoutput
average
less
than
the output voltage range of CON AD.
Figure 9 shows the input and the output wave form of the discussed rectifiers. In this case the open circuit input voltage
(1)
was ±2.75 V. The negative voltage converter in combination
with a PMOS
diode (CON
D) has nearly the same voltage
be reached using the negative voltage
converter
in combinant rectifiers versus
The strategy for stacking is that a maximum quantity
characteristic as the FW S. Both rectifiers with the active
tion with the active diode (CONdiode
AD).
The efficiency in this
The use ofnian
→kactive
(HW AD and CON AD) reach nearly the input voltmax of capacitors is used, whereas within each stackexample
is always
stage nof
are connected
in parallellarger
(see Fig.than
5). 75%.
age. The minimum voltage of the CON AD rectifier is also
ong increase
i / kthe
i capacitors
Thus, the maximum capacitance for a certain conversion ravery high, because every half wave is used and the capacitor
t efficiencies
could
tio is used, which reduces
the
conversion
frequency
f
.
is charged upresults
twice at each period.
convand conversion
4.2 Interface matching
Vb
Vb
ki =
=
V
+
1V
V
stop
ch,optof
ch
fficiency is also
www.adv-radio-sci.net/6/219/2008/
Adv. Radio Sci., 6, 219–225, 2008
Within all simulations and calculations, the generator
was included as a sine wave source with an ohmic serial resistance.
Because energy harvesting applications are considered, there
d with
m ausciency
capacvior of
quency
20 kΩ
e value
alue is
of the
Therefore, Fig. 11b shows the overall harvesting efficiency
ηhvst over the generator voltage range.
Fig. 8. The layout of the active diode enhanced rectifier, whereas
224dimensions
D. Maurath
et 40µm.
al.: Highly efficient integrated rectifier and voltage boosting circuits for energy harvesting applications
the
are 75 ×
he disvoltage
mbinae voltactive
ut voltis also
pacitor
Fig. 9. Simulation
Simulation results
results of
of the
the output
outputsignal
signalofofthe
thecompared
comparedrecrecaverageless
lessthan
than
tifiers. The output
output voltage
voltage range
range of
of CON
CON DDisisininaverage
6theDominic
Maurath:
Integrated Rectifier and Voltage Boosting Circuits for Energy Harvesting Applications
output voltage
range
of
AD.
voltage
range Highly
of CON
CONEfficient
AD.
also of
versus
active
of the
s could
array
Fig.11.
11. The
Thesimulated
simulated efficiencies
efficiencies in
in graph
graph (a)
(a) show
Fig.
show the
the ratio
ratio of
of
harvested
power
compared
to
maximum
possible
power,
and
harvested power compared to maximum possible power, and (b)
(b)
showsthe
theoverall
overallconversion
conversion efficiency.
efficiency.
shows
be reached using the negative voltage converter in combination with the active diode (CON AD). The efficiency in this
example is always larger than 75%.
4.2 Interface matching and conversion results
Within all simulations and calculations, the generator was included as a sine wave source with an ohmic serial resistance.
Because energy harvesting applications are considered, there
are two important efficiencies:
ηT x
1
=
(t2 − t1 )
R t2
t1
Vb Ich dt
V̂gen
Fig.
4the
Ricompared
Fig.10.
10. Efficiencies
Efficiencies of
of the
compared rectifiers
rectifiers
(2)
plotted versus the
input
inputvoltage.
voltage.
R t2
is
Vload Iload dt
ηhvst = ηT x × ηconv = ηT x Rt1t2
(3)
V I
dt
4.1.2 Enhanced rectification efficiency
t1 gen gen
converter
power. the
Theinput
higherefficiency
ηT x is, theηTbetter
achieved
Whereinput
(2) defines
x , which comBeside
the attainable
output
voltage the
efficiency
is also of
ispares
impedance
matching.
In
addition
to
this,
the
efficiency
the theoretical maximum generator power to the actual
The efficiency of the different rectifiers versus
ηimportance.
hvst in (3) gives a measure of the complete harvesting efthe inputThis
amplitude
is shown
in Fig.affected
10. Theby
useinternal
of an active
ficiency.
efficiency
is mainly
condiodelosses
in a half
rectifier
shows
a strong
increase
of 11a
the
verter
andwave
capacitor
charge
transfer
losses.
As Fig.
efficiency the
(HW
vs. HWinput
AD).power
The highest
could
illustrates,
interface
almostefficiencies
equals the maxibe reached
using
the power.
negative voltage converter in combinamum
possible
output
tion with the active diode (CON AD). The efficiency in this
example
is always
larger
Both graphs
prove
thatthan
this 75%.
novel interface approach and
conversion method operates as required. Generally, in order
Interface the
matching
and conversion
results
to4.2
characterize
efficiency
by interfacing
a generator, not
only ηconv can be considered, but also ηT x is of importance.
Within all simulations
and calculations,
the generator
was inTherefore,
Fig. 11b shows
the overall harvesting
efficiency
cluded
as
a
sine
wave
source
with
an
ohmic
serial
resistance.
ηhvst over the generator voltage range.
Because energy harvesting applications are considered, there
Adv. Radio Sci., 6, 219–225, 2008
are
important efficiencies:
5 two
Conclusions
R t2
Vb Ich dtconsisting of a voltage converter
1
t1concept
A
novel
rectifier
ηT x =
(2)
V̂gen presented. No additional bulk regu)
and an (t
active
was
2 − t1diode
Ri
lation transistors are4 needed
and nearly the entire generator
voltage appears at the rectifier
R t2 output. In combination with
a MOSFET diode the efficiency
is Iabout
load dt50%. In combinat Vload
ηtion
= ηTan
ηconvdiode
= ηT xrectifier
R1t2
hvst with
x ×
active
efficiencies
over 90%(3)
can
t1 Vgen Igen dt
be reached. The maximum output voltage is only a few mV
below the
voltage.
Where
Eq.input
(2) defines
the input efficiency ηT x , which comIt was
shown,maximum
that high generator
efficient power
and
pares
the also
theoretical
powerextraction
to the actual
voltage up-conversion
is feasible
harvesting
enviconverter
input power. The
higher ηinT xenergy
is, the better
achieved
ronments.
Especially
theInrealization
of this,
adaptive
impedance
is
impedance
matching.
addition to
the efficiency
and
maximum
generator
power
ηmatching
(3)almost
gives continuous
a measure of
the complete
harvesting
hvst in Eq.
efficiency.
Thisusage
efficiency
is mainly affected
by internal
conenhances the
of micro-generators
further.
These
adverter
losses
and
charge
losses. generators
As Fig. 11ato
vantages
give
thecapacitor
opportunity
of transfer
using smaller
illustrates,
the interface
input power
almost
equalsmight
the maxisupply autonomous
applications.
One
drawback
be the
mum
possible
power.
relatively
largeoutput
quantity
of capacitances, which need to be inBoth graphs
this area
novelisinterface
and
tegrated.
Thus,prove
a lot that
of chip
necessaryapproach
and parasitic
conversion
method
operates
as required.
Generally, in order
effects could
decrease
the overall
efficiency.
to characterize the efficiency by interfacing a generator, not
Acknowledgements.
This work but
is supported
theimportance.
German Reonly
ηconv can be considered,
also ηT x by
is of
search
Foundation
(Deutsche
Forschungsgemeinschaft
- DFG) unTherefore, Fig. 11b shows the overall harvesting efficiency
der
Grant
Number
GR1322
and
MA
2193/7-1.
ηhvst over the generator voltage range.
5References
Conclusions
M.novel
Ghovanloo,
K. Najafi,
integrated
A
rectifier
concept”Fully
consisting
of awideband
voltage high-current
converter
rectifier
for
inductively
powered
devices”,
IEEE
Journal
Solid
and an active diode was presented. No additional
bulk of
reguState Circuits, vol. 39, pp 1976-1984, 2004.
lation transistors are needed and nearly the entire generator
L. Gobbi, A. Cabrini, and G. Torelli, Impact of Parasitic Elements
voltage appears at the rectifier output. In combination with
on CMOS Charge Pumps: a Numerical Analysis, IEEE Internaa MOSFET
diode the
about 50%.
In combinational Symposium
on efficiency
Circuits andisSystems
(ISCAS),
May 2006.
tion
with
an
active
diode
rectifier
efficiencies
over
90% canS.
S. Kulkarni, E. Koukharenko, J. Tudor, S. Beeby, T. ODonnell,
Roy, Fabrication and Test of Integrated Micro-Scale Vibration
Based Electromagnetic Generator,
International Solid-State Senwww.adv-radio-sci.net/6/219/2008/
sors Actuators and Microsystems Conference, 2007.
Y.-H. Lam, W.-H. Ki, C.-Y. Tsui, ”Integrated Low-Loss CMOS Active Rectifier for Wirelessly Powered Devices”, IEEE Trans. on
L. Gobbi
on CM
tional
S. Kulka
Roy, F
Based
sors A
Y.-H. Lam
tive R
Circui
T. Lehm
biome
Circui
C. Peters
Integr
ternati
2415-2
X. Sheng
Conve
20th A
sition,
D. Spree
transd
roSens
D. Maurath et al.: Highly efficient integrated rectifier and voltage boosting circuits for energy harvesting applications
be reached. The maximum output voltage is only a few mV
below the input voltage.
It was also shown, that high efficient power extraction and
voltage up-conversion is feasible in energy harvesting environments. Especially the realization of adaptive impedance
matching and almost continuous maximum generator power
enhances the usage of micro-generators further. These advantages give the opportunity of using smaller generators to
supply autonomous applications. One drawback might be the
relatively large quantity of capacitances, which need to be integrated. Thus, a lot of chip area is necessary and parasitic
effects could decrease the overall efficiency.
Acknowledgements. This work is supported by the German Research Foundation (Deutsche Forschungsgemeinschaft – DFG) under Grant Number GR1322 and MA 2193/7-1.
References
Ghovanloo, M. and Najafi, K.: Fully integrated wideband highcurrent rectifier for inductively powered devices, IEEE J. Solid
State Circuits, 39, 1976–1984, 2004.
Gobbi, L., Cabrini, A., and Torelli, G.: Impact of Parasitic Elements
on CMOS Charge Pumps: a Numerical Analysis, IEEE International Symposium on Circuits and Systems (ISCAS), May 2006.
www.adv-radio-sci.net/6/219/2008/
225
Kulkarni, S., Koukharenko, E., Tudor, J., Beeby, S., O’Donnell,
T., and Roy, S.: Fabrication and Test of Integrated Micro-Scale
Vibration Based Electromagnetic Generator, International SolidState Sensors Actuators and Microsystems Conference, 2007.
Lam, Y.-H., Ki, W.-H., and Tsui, C.-Y.: Integrated Low-Loss
CMOS Active Rectifier for Wirelessly Powered Devices, IEEE
Trans. Circuits and Systems, 53(12), 1378–1382, 2006.
Lehmann, T. and Moghe, Y.: On-chip active power rectifiers for
biomedical applications, IEEE International Symposium on Circuits and Systems, 1, 732–735, 2005.
Peters, C., Kessling, O., Henrici, F., Ortmanns, M., and Manoli,
Y.: CMOS Integrated Highly Efficient Full Wave Rectifier, IEEE
IEEE International Symposium on Circuits and Systems (ISCAS), 2415–2418, New Orleans, 2007.
Shengwen, X., Ngo, K. D. T., Nishida, T., Gyo-Bum, Ch., and
Sharma, A.: Converter and controller for micro-power energy
harvesting, 20th Annual Applied Power Electronics Conference
and Exposition, 1, 226–230, March 2005.
Spreemann, D., Folkmer, B., Maurath, D., and Manoli, Y.: Tuneable transducer for low frequency vibrational energy harvesting,
EuroSensors 2006, September 2006.
Adv. Radio Sci., 6, 219–225, 2008
Download