Chen 2016

This article has been accepted for publication in a future issue of this journal, but has not been
fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/TPEL.2015.2489922, IEEE Transactions on Power Electronics
A Direct AC-DC and DC-DC Cross-Source Energy

Harvesting Circuit with Analog Iterating-based MPPT
Technique with 72.5% Conversion Efficiency and
94.6% Tracking Efficiency
Shin-Hao Chen1, Tzu-Chi Huang1, Student Member, IEEE, Shao Siang Ng1, Kuei-Liang Lin1,
Ming-Jhe Du1, Yu-Chai Kang1, Ke-Horng Chen1, Chin-Long Wey1, Ying-Hsi Lin2, Chao-Cheng
Lee2, Jian-Ru Lin2, and Tsung-Yen Tsai2
1
Institute of Electrical Control Engineering, National Chiao Tung University, Hsinchu, Taiwan
2
Realtek Semiconductor Corp., Hsinchu, Taiwan
Abstract—The proposed cross-source energy (CSE) harvesting system accepts universal energy
sources including AC and DC sources. The buck–boost conversion of CSE automatically converts
AC or DC input into DC output without being limited by the type of universal energy source. The
CSE technique simultaneously supplies the regulated output for the system loading and the battery
charging output. A test chip fabricated in the VIS BCD process can optimally derive harvest
energy with 72.5% power efficiency when one solar input source is used. A backup converter is
designed to complement the CSE technique to guarantee the regulation of output voltage. The
proposed analog iterating-based (AIB) maximum power point tracking (MPPT) technique
achieves 94.6% tracking efficiency without complex data calculation and storage, which are better
results compared with those of previous research.
Keywords—cross-source energy (CSE) harvesting system, universal energy sources, analog
iterating-based (AIB) maximum power point tracking (MPPT) technique
Ke-Horng Chen (corresponding author) is affiliated with the Institute of Electrical Control Engineering,
National Chiao Tung University, 1001 Ta-Hsueh Road, Hsinchu, Taiwan. (Telephone: +886-35712121-
54390, E-mail: khchen@cn.nctu.edu.tw, khchen@faculty.nctu.edu.tw).
0885-8993 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
I. INTRODUCTION
Energy harvesting techniques have been widely discussed on wireless sensor networks or portable
applications [1]–[4]. The most commonly discussed applications include wireless sensor nodes for health
care, embedded or implanted sensor nodes for medical applications, tire pressure monitoring in
automobiles, battery recharging of long-sustainability systems, security or guard systems in
housekeeping, and environment change monitor systems. In particular, the Internet of Things (IoT)
requires energy-harvesting techniques to obtain sufficient energy for extended battery usage time.
As depicted in Fig. 1, energy sources can be divided into two basic types: direct current (DC) and
alternating current (AC) energy sources. DC energy sources include thermal energy and solar energy.
By contrast, there are many types of AC energy sources, such as electromagnetic radiation, wind power,
RF power, and kinetic energy [5]–[14]. In this paper, the input power range of the proposed cross source
energy (CSE)-harvesting circuit is 200 μW to 50 mW. Input voltage can range from 0.06 V to 5 V.
According to Fig. 1, the energy sources that can be utilized are solar cell, wind power generator, magnetic
coil, and piezoelectric kinetic energy. However, the output power of the harvested RF and thermal energy
is lower than the required energy of some complex systems. Moreover, the energy supply of a harvesting
source may not satisfy the high-power consumption of an application. Battery capacity and power
delivery path require proper settings to maintain proper function.
Batteries are the most commonly used power source in conventional portable electronics.
Considering the use of an energy-harvesting system, the overall performance needs to outperform a
battery solution in energy density, power density, and/or cost. Typically, the niche for energy harvesting
is in long-lived applications where energy density is critical and the location of the devices may not be
reachable [7].
To overcome the disadvantages of both energy-harvested and battery-supplied systems, a hybrid
mode power supply system is recommended. Recharging a battery is preferred when using a harvested-
energy system. The battery supplies high power (up to several mWs) during a short period of time (as
the receiving, transmitting, and polling mode), whereas the energy harvester charges the battery with a
trickle current during the remainder of the time. Fig. 2 shows the system diagram of a conventional
cascaded harvesting system. AC-DC and DC-DC converter stages deal with the input source to generate
charging current to the storage device. The DC-DC converter stage regulates the system. This structure
conforms to the preferred scenario, and it is commonly used in many applications. However, the
conversion of the input energy uses the two-step approach, which results in low efficiency. To further
optimize conversion efficiency and reduce energy loss, the additional conversion loss induced by two-
stage structure can be lowered if the energy is directly transferred to the output load.
The system diagram of a parallel structure harvesting system is shown in Fig. 3. The parallel
structure can improve conversion efficiency and maintain charging function. The parallel structure takes
a primary path to directly convert energy to the load using a single conversion stage to avoid additional
conversion power loss. The secondary path transfers redundant energy to the storage device during
charging when the energy source generates more energy than the required load. A DC-DC converter is
placed after the battery to supply the primary path once the energy source becomes insufficient. This
structure brings more design challenges compared with the conventional cascaded structure. First, the
primary path of an AC-DC/DC-DC converter must regulate output voltage. Second, an energy
distribution scheme is required to accurately allot input energy to primary and secondary paths because
the two paths share the same energy source. Finally, the overhead cost and area caused by extra circuits
must be considered.
Many harvesting circuits are similar to conventional power converters. However, primary
differences exist between energy-harvesting circuit designs and conventional DC-DC or AC-DC
converter designs. These differences are listed as follows:
1. Energy quantity from the source is limited; hence loading condition of the system has to
meet the characteristics of the energy-harvesting source.
2. The energy source is not a fixed voltage or current source; load or environment also varies.
3. Input voltage may be very low or vary over a wide range.
4. Operation voltage and current have to be defined according to the MPP.
This paper is organized as follows. The proposed CSE harvesting system and analog iterating-based
maximum power point tracking (MPPT) are shown in Sections II and III, respectively. Section IV
discusses the experimental results. Finally, conclusions are presented in Section V.
II. PROPOSED CSE HARVESTING SYSTEM
A. AC Source Energy Harvesting Circuits
In traditional designs, an energy harvesting circuit with an AC source has several targets. The AC
input must be rectified to the DC value; otherwise energy is hardly used or stored. The magnitude of an
AC source in the harvesting system varies according to environmental conditions. Thus, setting a voltage
limit for some AC applications is required to protect the circuit from over-voltage damage. A two-stage
approach, as depicted in Fig. 2, is mainstream in AC source harvesting, and this has been presented in
previous literature. According to the input voltage and current range, the AC-DC structure is required to
convert the input voltage to a proper voltage level for back-end usage. Fig. 4 shows the summary and
function comparison of the harvesting structures. The rectifiers and converters are listed from high to
low power loss and classified according to the different function requirements. One diode has the forward
voltage drop VD (approximately 0.7 V) between two terminals, whereas the voltage drop of one MOSFET
is significantly smaller than VD and inversely proportional to its aspect ratio. The active rectifier structure
achieves minimum power loss compared with other AC-DC structures. Harvesting efficiency is defined
as output power PO divided by available input power PIN [15]. Power loss PLOSS in a power delivery path
determines overall efficiency ηH, as illustrated in (1).
PO PIN  PLOSS PIN  PC  PSW

H    (1)
PIN PIN PIN ,
where PC  iL , RMS 2  RS  iIN , AVG 2
.
Conduction loss PC is low because resistance RS of each switch is slight. Moreover, PC decreases
quadratically with input power PIN. Hence a simple instruction is provided to determine the structure and
design of the harvesting circuit.
B. DC Source Energy-Harvesting Circuits
In energy harvesting, the number of AC sources is higher than that of DC sources. The most widely
discussed DC harvesting sources include thermal electric generator and solar cell. The DC-harvesting
circuit aims to convert input voltage to a proper and regulated voltage level for battery charging or system
supply. The energy conversion of DC source-harvesting circuits is simpler than that of AC source-
harvesting circuits because it is similar to conventional DC-DC converters. DC-DC harvesting circuits
can use DC-DC converters to perform voltage regulation in the system diagram shown in Fig. 2 and Fig.
3.
DC-DC harvesting circuits and conventional DC-DC converters have the same design goals.
However, the quiescent power is a more important issue in harvesting circuits than in conventional DC-
DC converters. DC-DC converters may require reference voltage or bias current that some sub-circuits
can provide to support the DC-DC converter regularly. Some sub-circuits such as bandgap voltage
reference, comparators, and amplifiers consume a considerable amount of power. Power loss may reach
up to hundreds of μWs. If DC-DC converters convert large output power or require fast transient
response, high-power consumption caused by the controller is necessary. However, for low-input power
applications, such as harvesting circuits, the cost of power consumption may not be affordable. Moreover,
the input power achieved may be less than the input power that can be possibly derived. Thus, one of the
challenges of DC-DC harvesting circuits is designing an ultralow power operation scheme and the
corresponding low-power control circuit.
In previous designs, each harvesting circuit is designed for AC or DC sources. Although custom-
made harvesting circuits achieve better efficiency or have a smaller chip area, these circuits have many
disadvantages. Developing a custom-made harvesting circuit for any energy source is problematic and
not cost-effective to develop custom-made harvesting circuit for any energy source. The difference
between each energy source is significant; hence, the output power range, output current range, and
output voltage stability are difficult to control. Moreover, the functions of these circuits are limited to a
narrow input range or can only receive a specific AC frequency. Therefore, a harvesting circuit that is
compatible with a global energy source is the perfect solution. In a wireless sensor node (WSN) design,
the circuit with a wide input range can simultaneously harvest all energy sources, such as solar cell (DC
source), wind power (AC source), and RF power (AC source with different frequencies). As a result, the
received energy as well as the energy collecting speed can be higher than those of other single-sourced
harvesting structures.
The difficulties encountered when one harvesting circuit is applied to universal energy sources,
including AC and DC sources, should be addressed. First, the power stage should be used if AC or DC
input is received. Low power loss in conversion is a demand. Large input voltage range forces the
harvesting circuit to be capable of up or down conversion of the input voltage to the regulated output
voltage. Consequently, an active rectifier becomes the suitable option. By synchronously controlling the
four power MOSFETs, wide input range and AC-DC or DC-DC converter can be achieved
simultaneously. Second, MPPT, which increases harvesting power, is needed. However, the MPPT
method must be applied to AC and DC sources at the same time [2], [4]–[6]. Conventional perturb and
observe (P&O) method [4]–[6] samples power information periodically. Comparison of the power
information of the two continuous samples indicates that the controller adjusts the operation duty or
oscillating frequency to maximize input harvesting power [1]–[6]. However, direct AC MPPT must set
a sampling period because the operating frequency varies with different sources. The frequency of AC
sources that varies from 2.5 Hz to 1 kHz makes it difficult to derive power information without any
complex data calculation and storage [1]. Although the AC input source have been used [1] [3], the
proposed MPPT has been shown to work only in rectifying DC input voltage instead of directly
connecting to an AC source. Conflicting design demand has been observed in previous studies. To
overcome the difficulties stated earlier, the proposed CSE harvesting system in Fig. 5 for both AC and
DC energy sources should have a direct AC and DC MPPT scheme capable of attaining high efficiency.
The schematic diagram in Fig. 5 is described in detail in the succeeding part.
C. Proposed CSE-Harvesting Circuits
The proposed CSE harvesting circuit provides dual outputs, as shown in Fig. 6. One regulated
output VOut can supply system loading, and the other output can charge battery VBat. For the input energy
source, regulated VOut has higher priority than battery VBat to maintain the stable operation of the system.
The battery can be charged if the input energy is larger than the system loading requirement, as shown
in Figs. 6(a) and (b). Both PCharge and POut can be delivered. The MPPT control monitors the harvesting
energy source at its maximum level; thus, an additional battery works as a buffer to store or provide
energy if loading at VOut cannot be completely provided by the input energy. As illustrated in Fig. 5, the
backup converter connected between the battery and VOut ensures the regulation at VOut if input harvesting
energy source is not sufficient, as shown in Fig. 6(c). The battery cannot be charged by PCharge and PLoad
needs to drain current from the battery. In the feedback loop, VrefH regulates VOut. Furthermore, two active
diodes that determine the power delivery path are used to avoid power loss caused by reverse current
where VrefL is set for the correct operation of the two active diodes. When an active diode is used in a
converter design, no reverse power is allowed to flow back to the converter because of its diode-like
characteristic.
To accommodate the wide input voltage range from different harvesting sources, the CSE power
stage is designed as a buck–boost configuration that is composed of six power MOSFETs. A comparator
is connected to Vin+ and Vin- to recognize the polarity of the input source. The AIB-MPPT circuit
generates duty control signal VDt to the fixed frequency pulse width modulation (PWM) controller that
can generate gate control signals VS1–VS4, as shown in Fig. 5.
Figs. 7 and 8 show the AC-DC and DC-DC conversion power delivery paths, respectively. CSE
works in a discontinuous conduction mode (DCM). In AC-DC conversion, switches S1 and S4 are turned
on to charge the inductor current during a positive half charging path, as shown in Fig. 7(b). By contrast,
switches S2 and S3 are turned on to charge the inductor current during a negative half charging path, as
illustrated in Fig. 7(c). The inductor discharges energy through the discharging path formed by shorting
S5 to the ground, as shown in Fig. 7(d). The negative part of the AC source will be forced to the chip’s
ground by shorting switch S6. The ground-connected structure releases the harvesting energy produced
by engaging negative voltage and reduces the risk of body diode leakage. The rectified inductor current
can be always positive. As a result, the harvesting energy source is not necessary an AC or DC source;
CSE can harvest any arbitrary type of current source without being limited to periodic signals. Moreover,
in DC-DC mode, switches S1 and S4 are turned on to charge the inductor current, as depicted in Fig. 8(a).
Inductor L discharges power by shorting S5 to the ground, as shown in Fig. 8(b). CSE operates simply
as a buck–boost converter [5].
Fig. 9(a) shows the circuit implementation of the backup converter. VBat is always higher than
VOut, hence the backup converter is a buck converter working in discontinuous conduction mode (DCM)
with zero current detection (ZCD) circuit for high efficiency. The active diode circuit is depicted in Fig.
9(b) as the comparison between Vin and Vout to determine the conduction of the active diode. When Vin
is larger than Vout, back-to-back configuration will be conducted. On the contrary, when Vin is smaller
than Vout, the back-to-back configuration will be disconnected. Voltage Vlock is connected to VP1 and
VP2 to determine the power delivery path in Fig. 5. Resistor Rlimit can limit the current consumption of
the active diode.
Fig. 10(a) shows the waveform of the proposed CSE-harvesting circuit operating at a fixed
frequency of 10 KHz if the input energy source has an AC frequency of 500 Hz. The zoom-in
waveforms of Fig. 10(b) show that once the output voltage is 30 mV below the target voltage due to
insufficient harvesting energy source or load current changes in case of light-to-heavy loading current
variation, the backup converter will be triggered. VP1 has 4.2 V rail-to-rail pulses, and the battery
charges back-up inductor LB to 100 mA current.
III. ANALOG ITERATING-BASED MPPT
Various harvesting sources, such as photovoltaic, thermal electric, vibration, or magnetic coil, have
their own electrical and mechanical properties. Output voltage, current, and power of each source are
influenced by the loading effect. For each source in a steady state, a specific output voltage and current
condition that delivers maximum output power can be derived. The point is selected as MPP, and the
equivalent load on the MPP is defined and tracked through the MPPT control [1], [7], [16].
Fig. 11(a) shows the equivalent model of solar cells. First, current IL generated by solar cells is
assumed to be constant in a steady irradiation condition. If the load at output voltage VPV is low, VPV
increases, and leakage current on the parasitic diode DS and resistance RSH is induced. If the load is high,
VPV drops. Most of the power is wasted on series resistance RS. Thus, it is important to determine MPP
to enable the extraction of maximum power under the same environment condition. Many harvesting
systems have proposed useful MPPT circuits to enhance efficiency and derive maximum output power
[1], [7], [12], [17]–[19]. Some issues that need to be considered in a harvesting system with the MPPT
function are as follows:
1. The characteristic of energy sources must be determined before using the MPPT method.
2. Power consumption of the MPPT circuit is critical for many low-power applications.
3. MPPT control will force the input source attain its maximum output energy.
4. Storage device is necessary if the harvesting system intends to have MPPT function.
5. To overcome environmental variations, continuous tracking is necessary.
MPP varies under different environmental conditions. Harvesting circuits need to control the power
delivery condition to maintain the operation of the energy-harvesting system at its MPP. The MPPT
scheme is a method that tracks input power. If a harvesting system directly provides supply voltage to
the load system, the MPPT function disables the harvesting system to regulate the output voltage at the
same time; this will occur unless the input power is always the same as the load requirement, which is
not possible. Thus, the following general ideas and methods for implementing the MPPT control are
introduced.
A. Impedance Matching
Many harvesting sources have a complex behaviors or models of their internal equivalent circuit.
Previous studies [1], [12], [20], and [21] have presented many optimized designs for different
characteristics of harvesting sources. These designs have outstanding output power performance because
of their special operation schemes. However, these circuits can only be applied to specific sources.
Impedance matching method, which is the most popular and important approach, is a general purpose
MPPT scheme that can be applied to different sources. Fig. 11(b) shows the Thevenin equivalent circuit
of an energy-harvesting source that is modeled as ideal voltage source VEQ with series resistance RS.
Generally, all sources can be modeled as Fig. 11(b) with different Thevenin equivalent impedances even
if the impedance is not pure resistance and inductance and capacitance are included. However, matching
the load impedance with inner impedance is still valid and provides tracking instructions when the MPPT
is desired.
Power PLoad on output loading RLoad is shown as follows:
2
 VEQ 
PLoad  I 2
Load  RLoad    RLoad (2)
 RS  RLoad  .
Impedance matching theory implies that a system achieves maximum power transfer when the
loading impedance is equal to the inner impedance. Maximum output power PLoad,max that is transferred
to the load is expressed as Eq. (). Matching efficiency ηm is defined as the ratio of power on load PLoad
to PLoad,max, as shown in Eq. ().
2
VEQ
PLoad ,max  (3)
4RLoad .
PLoad 4
m  
PLoad ,max R R (4)
1  2 S  ( S )2
RLoad RLoad
.
Fig. 12 shows the matching efficiency under different impedance errors. Impedance error represents the
mismatch percentage as compared with the inner impedance of the harvesting source. When the load
impedance is perfectly matched with the inner impedance, the error percentage is zero and the matching
efficiency is consistent. If 90% matching efficiency is desired, the endurable load impedance error ranges
from −48% to +93%. A wide endurable range indicates that even when a large impedance error
percentage occurs, the output power remains very close to the MPP.
B. Resistor Emulation
Regardless of type, whether switching or linear power converters, an equivalent resistance of the
converter can be obtained if the ratio of input voltage to the average input current is considered. This is
called the resistance emulation scheme [16] [22]. Equivalent resistance can be determined by calculating
the average input current. To illustrate, Figs. 13 (a) and (b) show the charging and discharging phases of
the buck–boost converter, respectively, which are derived when a buck ̶ boost converter with a constant
switching frequency PWM control operates in DCM. In the DCM operation, inductor current will be
released to zero in every clock cycle. Fig. 14 shows the timing diagram of the inductor current when
input voltage varies in the situation where the PWM clock and duty are fixed.
The converter’s average input current Iin,avg is shown as:
I peak  D VinT  D2
Iin,avg  
2T 2L ,
(5)
where IPeak is the peak current in the charging phase, D is the duty cycle, T is the time of a
switching period, and L is the inductor.
By dividing input voltage Vin by Iin,avg, equivalent resistance Req can be derived using Eq. (6), which
is related to L, D, and T.
Vin 2L
Req   (6)
I in , avg T  D2
.
Normally, the control factor can be D or T because L is a selected value and cannot be used as a
control factor of the equivalent resistance. As shown in Fig. 15(a), duty or switching frequency (controls
T) can be used to adjust the value of Req through PWM or pulse frequency modulation (PFM) control.
Thus, the converter can be regarded as a tunable resistance to match the inner impedance and to obtain
the maximum output power.
Several considerations in relation to the resistance emulation method should be noted. First, in the
selection of converter parameters, Req should fit the inner impedance of the harvesting source. The
coverage range of Req should be designed according to the target source. If a switching converter is used
as the emulated resistor, a low-switching frequency will result in a large switching current ripple. If the
energy source’s driving capability is not sufficient to sink current, the source’s terminal will reduce
significantly. Large voltage variation on some energy sources, such as a solar cell that has complex
equivalent internal models, will influence the output power condition. Even if the equivalent resistance
is the ratio of the average input voltage and current, large voltage variation may cause extra power loss
or may diverge the power condition from the MPP.
Under the resistor emulation approach, the MPPT controller adjusts the control factor of the resistor
emulation converter, such as the duty or the frequency. However, resistor emulation only adjusts the
operation point. Adjusting the operation mode changes the emulated loading to match the characteristic
of the harvesting source. However, determining the MPP requires the MPPT method, which is a
procedure that requires that the adjustment should be arranged. The output power of the energy source
should be sampled to define the power status. Based on the power status, the adjustment is meaningful.
By continuously sampling and comparing the power status through the adjustment, the harvesting system
can eventually operate at the MPP of the energy source. Different harvesting sources and operation
conditions require various tracking methods to obtain MPP.
For harvesting sources with complex models and behaviors, the general way of tracking the MPP
is hill climbing, which is also called the P&O method [1] [16]. This method has been used on solar cells
for decades because of the complex output power behavior of solar cells under various environmental
conditions. To illustrate, output power and output current versus output voltage of a solar cell with the
MPPT operation is shown in Fig. 15(b). At the start of adjusting the MPPT process, the output power is
measured after each adjustment. If power increases, further adjustment in the same direction is attempted
until power increase is prevented. If power decreases, the adjustment changes to the inverse direction.
This scheme is similar to the hill climbing method because it tracks the power curve. Once the adjustment
is beyond the MPP, the output power falls below the MPP. The adjustment turns into the inverse direction
and traces back to the MPP. The tracking operation goes back and forth around the MPP. To continue
tracking the output power, the MPPT keeps on operating to deal with the environmental variations. Thus,
if the environmental condition is stable, the adjustment will terminate in a limiting cycle oscillation.
C. Analog Iterative MPPT
The hill climbing method compares the power status after each sampling and adjusting. However,
this method is not suitable for simultaneous sampling and adjusting of AC or DC sources. The iterative-
based MPPT method diverts the comparison issue from the previous sampled power status to a dynamic
target power through the MPPT in the AC or DC source. First, the target power is allowed to trace the
present sampled power status. Next, the target power level is promoted and the converter is adjusted to
enable the output power to reach the target. The procedure is repeated, and the power target and sampled
power status iteratively track each other to reach the MPP.
Fig. 16 shows the tracking operation. A target power represented by voltage signal VTarget is
compared with the peak of the current sense signal VCS. At the beginning, VTarget is set to be higher than
VCS. A timeout period TO is designed as the settling time for the converter and the energy harvesting
source. If VTarget and VCS do not encounter each other within the timeout period, VTarget will be set to a
lower level. Through several comparisons, VTarget will approach VCS. Moreover, VTarget represents the
recent power status. VTarget is set to a slightly higher level as a new target for VCS. The MPPT controller
adjusts the PWM duty and checks the variation of VCS. The adjustment may cause output power
fluctuation. The increase of VCS to VTarget before the timeout indicates that output power rises and the
duty adjusting trend is correct. Subsequently, VTarget shifts to a higher level as the new target. The
adjusting trend continues at the same direction. However, if VCS cannot reach VTarget, an incorrect
adjustment trend occurs. After the timeout, the MPPT controller lowers VTarget to determine VCS and to
change the adjusting trend. In conclusion, VTarget is always set to a slightly higher target value compared
with the recent input power condition.
The inductor peak current is proportional to the average input power because of the characteristics
of the fixed-frequency DCM and buck–boost operation. Based on this information for the MPPT, the
current sense circuit as shown in Fig. 17(a) is proposed to monitor the input power. Using AC input as
an example, Fig. 17(b) shows the signal timing diagram. The digital-to-analog converter (DAC)
generates tunable reference VTarget for comparison with current sense voltage VCS and to determine the
inductor peak current. The period of the input source is unknown; thus, the timeout circuit can determine
the comparison sampling timing. Fig. 18 shows the AIB-MPPT finite state machine (FSM). During AIB-
MPPT tracking, VTarget is supposed to be higher than VCS, and the AIB-MPPT sets VTarget to a lower level
because comparator output VComp is not triggered to a height within 200 ms. Through several comparisons,
VTarget reaches VCS to represent the recent input power level. VTarget is set to a slightly higher level as the
new target for VCS. The AIB-MPPT adjusts the PWM duty and checks the variation of VCS. The
adjustment may cause output power fluctuation. The increase of VCS to VTarget before the timeout indicates
that output power rises and the duty adjusting trend is correct. VTarget shifts to a higher level as the new
target. The adjusting trend continues in the same direction. However, if VCS cannot reach VTarget, an
incorrect adjusting trend occurs. After the timeout, the AIB-MPPT lowers VTarget to determine VCS and
to change the adjusting trend. In conclusion, VTarget should always be set to be a slightly higher target
compared with the recent input power condition. After repeating the procedure, VTarget and VCS will
iteratively track each other to attain MPP [5].
The sensing information will be converted by one R-2R DAC circuit (Fig. 19) and one 7-bit
successive approximation register (SAR) to form the binary search shown in Fig. 20(a). The 7-bit SAR
circuit is composed of five registers, a multiplexer (MUX), and a D flip-flop (DFF). The register in Fig.
20(b) operates through the truth table in Fig. 20(c). Ena denotes the enable signal. The overall operation
is expressed in the flow chart shown in Fig. 20(d). After each comparison, the current working register
will trigger the next register to set the control code ZA[n-1]=1 for the sequent comparison and will
subsequently receive the comparison result from the Comp to adjust output ZA[n] to one or zero.
After the seven-cycle comparison, the least significant bit is determined, and the last DFF increases
and locks out. The output of the 7-bit SAR will be used to derive the accurate off-time value. An
additional signal ‘Lock’ derived from the off-chip control can lock out the 7-bit SAR or allow the LSB
of the 7-bit SAR to work continuously. When the calibration is completed, the AIB-MPPT circuit is shut
down to save power. The tuning range of the sensing circuit has an input voltage tolerance from 1.2 V
to 2.5 V. The current mismatch can achieve 110% at the input voltage of 1.5 V.
IV. EXPERIMENTAL RESULTS
The proposed CSE-harvesting circuit is tested using four input sources: wind turbine generator,
current transformer (CT), solar cells, and piezoelectric transducer (PZT) [23]. The measured AC input
conversion with and without the backup converter is shown in Fig. 21. Measurement results indicate that
the CSE-harvesting circuit rectifies the AC input current and directly supplies VOut. The CSE selects a
negative terminal from the AC source and connects the negative terminal to the harvesting circuit’s
ground automatically. Thus, the circuit does not experience a voltage lower than that of the ground. VOut
drops periodically because of the energy dead band in the AC source. During this process, the system
can be supplied by the backup converter to obtain extra energy and to ensure high-quality and regulated
voltage. The voltage ripple can be less than 30 mV, as shown in Fig. 22(a). By contrast, Fig. 21 shows
the conversion without the backup converter. The waveform has a significant voltage drop at VOut during
the trough of AC waveforms. When the energy delivered to VOut is sufficient, the extra energy will be
used to charge the battery. The slope of IL is significantly steeper when energy is delivered to the battery,
as indicated in Fig. 22(b). Fig. 23 shows the output ripple and ZCD waveforms of the backup converter.
Fig. 24 shows that the AIB-MPPT can track the maximum input power and damp within a small range.
The CSE system reaches a peak power efficiency of 72.5% and tracking efficiency of 94.6%, as shown
in Fig. 25. Maximum power efficiency can be calculated based on Eqs. (1) and (4). Tracking efficiency
is based on Eqs. (3) and (6). Table I shows a comparison with the findings of previous studies [3], [4],
[6]. The proposed work harvests energy from either AC or DC sources. Input voltage range is not limited
by output voltage because of the buck–boost structure. By eliminating the complex data processing, the
proposed AIB-MPPT achieves direct AC source MPPT as well as the benefit of area and energy
efficiency. Fig. 26 shows the design specifications and chip micrograph with an active area of 1.69 mm2
in VIS 0.15 μm BCD process [5].
V. CONCLUSIONS
Most state-of-the-art harvesting circuits can only receive DC energy source or single frequency AC
energy source. Moreover, harvesting AC source energy requires a diode bridge rectifier, which leads to
more power losses. A direct AC-DC and DC-DC CSE- harvesting circuit with analog iterating-based
MPPT technique is proposed in this paper. The buck–boost conversion of CSE automatically converts
AC or DC input into DC output without being limited by the universal input voltage range. In AC-DC
conversion, the CSE harvesting circuit can accept wide range of AC sources including the non-periodic
AC source through the use of the AIB-MPPT technique. Furthermore, the proposed AIB-MPPT
technique achieves 94.6% tracking efficiency without complex data calculation and storage that have
not been achieved by previous techniques. A backup converter connecting VBat and VOut is designed to
guarantee output voltage stability.
ACKNOWLEDGEMENTS
This research was supported by the National Science Council, Taiwan, under Grant MST 103-2622-
E-009 -010 -CC2, MST 103-2923-E-009 -005 -MY3, MST 103-2221-E-009 -074 -MY3, and MST 103-
2220-E-009 -022.
REFERENCES
[1] Y.-K. Tan and S. K. Panda, “Optimized wind energy harvesting system using resistance emulator
and active rectifier for wireless sensor nodes,” IEEE Trans. Power Electron., vol. 26, no. 1, pp.
38–50, Jan. 2011.
[2] K. Kadirvel, Y. Ramadass, ;U. Lyles, J. Carpenter, V. Ivanov, V. McNeil, A. Chandrakasan and
Lum-Shue-Chan, “A 330nA Energy-Harvesting Charger with Battery Management for Solar and
Thermoelectric Energy Harvesting,” ISSCC Dig. Tech. Papers, pp. 106-108, Feb. 2013.
[3] S. Stanzione, C. van Liempd, R. van Schaijk, Y.Naito, R.F. Yazicioglu and C. Van Hoof, “A self-
biased 5-to-60V input voltage and 25-to-1600μW integrated DC-DC buck converter with fully
analog MPPT algorithm reaching up to 88% end-to-end efficiency,” ISSCC Dig. Tech. Papers,
pp. 74-75, Feb. 2013.
[4] J.-P. Im, S.-W. Wang, K.-H. Lee, Y.-J. Woo, Y.-S. Yuk, T.-H. Kong and S.-W. Hong, “A 40mV
Transformer-Reuse Self-Startup Boost Converter with MPPT Control for Thermoelectric Energy
Harvesting” ISSCC Dig. Tech. Papers, pp. 104-106, Feb. 2012.
[5] T.-C. Huang, M.-J. Du, K.-L. Lin, S. –S. Ng, K.-H. Chen and C.-L. Wey, “A Direct AC-DC and
DC-DC Cross-Source Energy Harvesting Circuit with Analog Iterating-based MPPT Technique
with 72.5% Conversion Efficiency and 94.6%” Symposium on VLSI Circuits Digest of Technical
Papers, 2014.
[6] R. Enne, M. Nikolic and H. Zimmermann, “A Maximum Power-Point Tracker without Digital
Signal Processing in 0.35μm CMOS for Automotive Applications,” ISSCC Dig. Tech. Papers,
pp. 102-104, Feb. 2012.
[7] S. Priya and D. J. Inman, Energy harvesting technologies. Springer, NY, 2009.
[8] Y. Yang, F. Lambert and D. Divan, “A survey on technologies for implementing sensor networks
for power delivery systems,” in Proc. IEEE Power Engineering Society General Meeting, pp.1–
8, Jun. 2007.
[9] S. J. Roundy, “Energy scavenging for wireless sensor nodes with a focus on vibration to electricity
conversion,” Ph. D thesis, University of California, Berkeley, 2003.
[10] J. A. Paradiso and T. Starner, “Energy scavenging for mobile and wireless electronics,” IEEE
Pervasive Computing, pp. 18–27, Jan. 2005.
[11] L. Mateu and F. Moll, “Review of energy harvesting techniques and applications for
microelectronics,” in Proc. of SPIE, 5837, pp. 359–373, 2005.
[12] Y. K. Ramadass and A. P. Chandrakasan, “A battery-less thermoelectric energy harvesting
interface circuit with 35 mv startup voltage,” IEEE J. Solid-State Circuits, vol. 46, no. 1, pp. 333–
341, Jan. 2011.
[13] E. J. Carlson, K. Strunz and B. P. Otis, “A 20 mV input boost converter with efficient digital
control for thermoelectric energy harvesting,” IEEE J. Solid-State Circuits, vol. 45, no. 4, pp.
741–750, Apr. 2009.
[14] “Thermoelectric generators,” [Online]. Available: http://www.tellurex.com/
[15] Niell Elvin and Alper Erturk, Advances in Energy Harvesting Methods. Springer, NY, 2013.
[16] S. Bandyopadhyay and A. P. Chandrakasan, “Platform architecture for solar, thermal, and
vibration energy combining with MPPT and single inductor,” IEEE J. Solid-State Circuits, vol.
47, no. 9, pp. 2199–2215, Sep. 2012.
[17] E. Koutroulis and K. Kalaitzakis, “Design of a maximum power tracking system for wind-energy-
conversion applications,” IEEE Trans. Industrial Electron., vol. 53, pp. 486–494, Apr. 2006.
[18] Y.-K. Tan and S. K. Panda, “Self-autonomous wireless sensor nodes with wind energy harvesting
for remote sensing of wind-driven wildﬁre spread,” IEEE Trans. Power Electron., vol. 26, no. 4,
pp. 1367–1377, Apr. 2011.
[19] T. Paing, E. Falkenstein, R. Zane, and Z. Popovic, “Custom IC for ultra-low power RF energy
harvesting,” IEEE Applied Power Electronics Conference and Exposition(APEC 2009), pp.
1239–1245, 2009.
[20] D. Kwon and G. A. Rincon-Mora, “A single-inductor AC–DC piezoelectric energy-
harvester/battery-charger IC converting ± (0.35 to 1.2V) to (2.7 to 4.5V),” ISSCC Dig. Tech.
Papers, pp. 494–495, Feb. 2010.
[21] T.-C. Huang, M.-J. Du, Y.-Y. Yang, Y.-H. Lee, Y.-C. Kang, R.-H. Peng and K.-H. Chen, “Non-
invasion power monitoring with 120% harvesting energy improvement by maximum power
extracting control for high sustainability power meter system,” in Proc. IEEE Custom Integrated
Circuits Conf. (CICC), pp. 1–4, Sep. 2012.
[22] T. Paing, J. Shin, R. Zane and Z. Popovic, “Resistor emulation approach to low-power RF energy
harvesting,” IEEE Trans. Power Electron., vol. 23, no. 3, pp. 1494–1501, May 2008.
[23] Midé Technology Corporation, V25W-ND, Piezoelectric energy harvesters, Jan. 2013.
TABLE CAPTIONS:
TABLE I: MEASUREMENT COMPARISONS OF PRIOR HARVESTING SYSTEM.
FIGURE CAPTIONS:
Fig. 1. Available energy in environment.
Fig. 2. System diagram of a conventional serial structure harvesting system.
Fig. 3. System diagram of a parallel structure harvesting system.
Fig. 4. Function selection and characteristics of different AC-DC structures.
Fig. 5. Proposed harvesting circuit and power deliver diagram.
Fig. 6. Energy distribution priority. (a) Input energy is used to charge the battery when loading is low.
(b) Input energy is used to supply the loading when loading increases. (c) Input energy mainly supplies
the loading and disconnects from the battery when input energy decreases.
Fig. 7. AC input current paths. (a) Waveforms. (b) Positive-half charging path. (c) Negative-half
charging path. (d) Discharging path.
Fig. 8. DC input current paths. (a) DC charging path. (b) Discharging path.
Fig. 9. Circuit implementation of (a) backup converter and (b) active diode.
Fig. 10. (a) Waveforms of CSE harvesting circuit. (b) Zoom-in waveforms.
Fig. 11. (a) Modeling of the solar cell and (b) Thevenin equivalent circuit of energy source.
Fig. 12. Matching efficiency versus impedance error.
Fig. 13. Buck ̶ boost converter operation in (a) charging and (b) discharging phases.
Fig. 14. Inductor current under DCM operation.
Fig. 15. (a) Using PWM duty to adjust the value of Req. (b) Output power and output current versus
output voltage of solar cells if the MPPT is used.
Fig. 16. The target power and the sampled power status iteratively track each other due to the proposed
AIB-MPPT.
Fig. 17. (a) AIB-MPPT circuit and (b) Timing diagram.
Fig. 18. AIB-MPPT FSM.
Fig. 19. R-2R DAC.
Fig. 20. (a) 7-bit SAR circuit. (b) Register. (c) Truth table. (d) Flow chart.
Fig. 21. Measured AC input conversion with and without backup converter.
Fig. 22. Measured (a) ILB and (b) IL of AC input conversion.
Fig. 23. Output ripple and ZCD of the backup converter.
Fig. 24. Measured AIB-MPPT tracking.
Fig. 25. Measured power and tracking efficiency.
Fig. 26. (a) Design specifications. (b) Chip micrograph.
TABLE I: MEASUREMENT COMPARISONS OF PRIOR HARVESTING SYSTEM.

ISSCC 2013 ISSCC 2012 ISSCC 2012
This Work
[3] [4] [6]
Technology 0.25 m 0.13 m 0.35 m 0.15 m
Energy source PZT Thermal Solar Magnetic, Solar, Wind, PZT
Input type AC DC DC AC or DC
Rectifier Full bridge N/A N/A Direct AC/DC
Converter type Buck Boost Boost Buck-Boost
Storage Li-ion battery N/A N/A Li-Ion battery
Open circuit
MPPT algorithm VS-P&O P&O AIB-MPPT
P&O
Regulated output No Yes No Yes
25 W ̶ 1.6 2.25 W ̶ 36
Input power 40 200 W ̶ 50 mW
mW W
Input voltage 5 V ̶ 60 V 40mV~0.3V <4.5 V 0.06 V ̶ 5 V
Regulated: 1.2 V ̶ 2.5 V
Output voltage 2V̶5V 2V 12 V
Charge: 2.6 V ̶ 4.2 V
Controller Power N/A N/A 3.68 mW 35 W
88.7% 61% 72.5%
Efficiency N/A
@VIN=30V @VIN=0.3V @VIN=3V Solar
Die size 9.52 mm2 0.09 mm2 0.441 mm2 1.69 mm2
Fig. 1. Available energy in the environment.
Fig. 2. System diagram of a conventional serial structure harvesting system.
Fig. 3. System diagram of a parallel structure harvesting system.
Fig. 4. Function selection and characteristics of different AC-DC structures.
Fig. 5. Proposed harvesting circuit and power deliver diagram.
(a)
PCharge
PIn
PLoad
POut
(b)
(c)
Fig. 6. Energy distribution priority. (a) Input energy is used to charge the battery when loading is low.
(b) Input energy is used to supply the loading when loading increases. (c) Input energy mainly supplies
the loading and disconnects from the battery when input energy decreases.
(a)
(b)
(c)
(d)
Fig. 7. AC input current paths. (a) Waveforms. (b) Positive-half charging path. (c) Negative-half
charging path. (d) Discharging path
Vin
D1
VBat
S1 S2
VDC CIn
L D2 CB
VOut
S5 S6 CO
S3 S4
(a)
(b)
Fig. 8. DC input current paths. (a) DC charging path. (b) Discharging path.
(a) (b)
Fig. 9. Circuit implementation of (a) backup converter and (b) active diode.
Inductor Zoom-in
Vin Current waveforms
ts ts t
Avg. Current
ton
(a)
(b)
Fig. 10. (a) Waveforms of CSE harvesting circuit. (b) Zoom-in waveforms.
Fig. 11. (a) Modeling of the solar cell. (b) Thevenin equivalent circuit of energy source.
Fig. 12. Matching efficiency versus impedance error.
(a) (b)
Fig. 13. Buck ̶ boost converter operation in (a) charging and (b) discharging phases.
Fig. 14. Inductor current under DCM operation.
D1
VBat
S1 S2
VAC CIn CB
L D2
VOut
Req
S5 S6 CO
S3 S4
Buck-Boost
PWM Duty
(a)
(b)
Fig. 15. (a) Using PWM duty to adjust the value of Req (b) Output power and output current versus
output voltage of solar cells if the MPPT is used.
Fig. 16. The target power and the sampled power status iteratively track each other due to the proposed
AIB-MPPT.
(a) (b)
Fig. 17. (a) AIB-MPPT circuit. (b) Timing diagram.
Fig. 18. AIB-MPPT FSM.
Fig. 19. R-2R DAC circuit.
(a)
(b) (c) (d)

Fig. 20. (a) 7-bit SAR circuit. (b) Register. (c) Truth table. (d) Flow chart.
Fig. 21. Measured AC input conversion with and without the backup converter.
VOut Regulated @ 1.7V 30mV
Backup converter
AC input low, voltage ripple
Backup converter is
ILB activated
IL
fan.gif
2ms
(a) (b)
Fig. 22. Measured (a) ILB and (b) IL of AC input conversion.
Fig. 23. Output ripple and ZCD of the backup converter.
Fig. 24. Measured AIB-MPPT tracking [5].
Fig. 25. Measured power and tracking efficiency [5].
Design Specifications
Technology VIS 0.15m
Energy source Magnetic, Solar, Wind, PZT
Input type AC or DC
Rectifier Direct AC/DC
Converter type Buck-Boost
Storage Li-ion battery
MPPT algorithm AIB-MPPT
Regulated output Yes
Input power 200W~50mW
Input voltage 0.06V~5V
Regulated: 1.2V~2.5V
Output voltage
Charge: 2.6V~4.2V
Controller Power 35W
72.5%
Efficiency
@VIN=3V Solar
Die size 1.69 mm2
(a) (b)
Fig. 26. (a) Design specifications. (b) Chip micrograph.

Chen 2016

Uploaded by

Copyright:

Available Formats

Chen 2016

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chen 2016

Uploaded by

Copyright:

Available Formats

This article has been accepted for publication in a future issue of this journal, but has not been

A Direct AC-DC and DC-DC Cross-Source Energy

Keywords—cross-source energy (CSE) harvesting system, universal energy sources, analog

iterating-based (AIB) maximum power point tracking (MPPT) technique

automobiles, battery recharging of long-sustainability systems, security or guard systems in

delivery path require proper settings to maintain proper function.

To overcome the disadvantages of both energy-harvested and battery-supplied systems, a hybrid

converter designs. These differences are listed as follows:

meet the characteristics of the energy-harvesting source.

3. Input voltage may be very low or vary over a wide range.

4. Operation voltage and current have to be defined according to the MPP.

discusses the experimental results. Finally, conclusions are presented in Section V.

II. PROPOSED CSE HARVESTING SYSTEM

A. AC Source Energy Harvesting Circuits

determines overall efficiency ηH, as illustrated in (1).

PO PIN  PLOSS PIN  PC  PSW

design of the harvesting circuit.

B. DC Source Energy-Harvesting Circuits

corresponding low-power control circuit.

The schematic diagram in Fig. 5 is described in detail in the succeeding part.

C. Proposed CSE-Harvesting Circuits

can generate gate control signals VS1–VS4, as shown in Fig. 5.

as a buck–boost converter [5].

the active diode.

charges back-up inductor LB to 100 mA current.

III. ANALOG ITERATING-BASED MPPT

function are as follows:

5. To overcome environmental variations, continuous tracking is necessary.

Power PLoad on output loading RLoad is shown as follows:

to PLoad,max, as shown in Eq. ().

The converter’s average input current Iin,avg is shown as:

the maximum output power.

or may diverge the power condition from the MPP.

conditions require various tracking methods to obtain MPP.

C. Analog Iterative MPPT

power status iteratively track each other to reach the MPP.

with the recent input power condition.

iteratively track each other to attain MPP [5].

IV. EXPERIMENTAL RESULTS

in VIS 0.15 μm BCD process [5].

guarantee output voltage stability.

38–50, Jan. 2011.

pp. 74-75, Feb. 2013.

Harvesting” ISSCC Dig. Tech. Papers, pp. 104-106, Feb. 2012.

pp. 102-104, Feb. 2012.

conversion,” Ph. D thesis, University of California, Berkeley, 2003.

Pervasive Computing, pp. 18–27, Jan. 2005.

microelectronics,” in Proc. of SPIE, 5837, pp. 359–373, 2005.

[12] Y. K. Ramadass and A. P. Chandrakasan, “A battery-less thermoelectric energy harvesting

341, Jan. 2011.

741–750, Apr. 2009.

[14] “Thermoelectric generators,” [Online]. Available: http://www.tellurex.com/

47, no. 9, pp. 2199–2215, Sep. 2012.

pp. 1367–1377, Apr. 2011.

[20] D. Kwon and G. A. Rincon-Mora, “A single-inductor AC–DC piezoelectric energy-

harvester/battery-charger IC converting ± (0.35 to 1.2V) to (2.7 to 4.5V),” ISSCC Dig. Tech.

Papers, pp. 494–495, Feb. 2010.

Circuits Conf. (CICC), pp. 1–4, Sep. 2012.

TABLE I: MEASUREMENT COMPARISONS OF PRIOR HARVESTING SYSTEM.