IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 20, NO.
4, JULY 2005
An Integrated Inverter With Maximum Power
Tracking for Grid-Connected PV Systems
Billy M. T. Ho, Student Member, IEEE, and Henry Shu-Hung Chung, Senior Member, IEEE
AbstractAn inverter for grid-connected photovoltaic systems
is presented in this paper. It can globally locate the maximum
power point of the panel over wide insolation and feed the solar
energy to the grid. Its structure mainly integrates a previously
developed maximum point tracking method and output current
shaping function into a buck-boost-derived converter and then
inverts the shaped current through a grid frequency bridge to
the grid. Instead of having a storage capacitor connecting in
parallel with the converter output, series connection is used, so
that the required capacitor voltage rating is lower than that in
classical inverters. Most importantly, the inverter output current
harmonics are less sensitive to the capacitor value. A 30-W labo-
ratory prototype has been built. The tracking capability, inversion
efficiency, and large-signal responses at different insolations have
been investigated. Detailed analysis on the inverter performance
has been performed. The theoretical predictions are verified with
the experimental results.
Index TermsDC/AC power conversion, inverters, photovoltaic
systems.
I. INTRODUCTION
P HOTOVOLTAIC (PV) technology has developed rapidly
over the last two decades from a small scale, specialist
industry supplying the United States space program to a
broadly based global activity [1]. The advancement of power
electronics and semiconductor technologies, the declining cost
of solar panels, and the favorable incentives in a number of
countries had profound impact on the commercial acceptance
of grid-connected PV systemswhich have been used in peak
shaving, demand reduction, and supply of remote loads [2],
[3]. Apart from the solar panels, the core technology associated
with these systems is a power-conditioning unit (inverter) that
converts the solar output electrically compatible with the utility
grid [4].
Most inverters in the mid 1990s consisted of a central in-
verter of dc power rating above 1 kW. They connect several
solar panel strings in parallel via a dc bus. However, the con-
cept has the drawbacks of causing a complete loss of gener-
ation during inverter outage and losses due to the mismatch
of strings [5]. Later, string inverters, which are designed for a
system of one string of panels, were used to lessen the problems
and have become popular nowadays. With further system de-
centralization, concept of AC-module was introduced. Every
solar panel has a module-integrated inverter of power rating
Manuscript received July 17, 2003; revised June 8, 2004. This work was sup-
ported by the Research Grants Council, Hong Kong Special Administrative Re-
gion, China, under Project CityU 1221/03E.
The authors are with the Department of Electronic Engineering, City Univer-
sity of Hong Kong, Hong Kong (e-mail: eeshc@cityu.edu.hk).
Digital Object Identifier 10.1109/TPEL.2005.850906
0885-8993/$20.00 2005 IEEE
953
Fig. 1. Typical structures of grid-connected PV systems. (a) With voltage-fed,
self-commutated inverter switching at high frequency. (b) With current-fed,
grid-commutated inverter switching at the grid frequency.
below 500 W mounted on the backside [6][10]. This panel-
inverter integration allows a direct connection to the grid and
provides the highest system flexibility and expandability. It also
offers the possibilities to overcome problems with respect to
high dc voltage level connection, safety, cable losses, and risk
of dc arcs, and to achieve high-energy yield in case of system
suffering from shading effect, due to the lack of mutual influ-
ence among modules operating points.
Typical structures of the AC-module consist of several power
conversion stages (Fig. 1) [9], [10]. The front stage has a max-
imum power point (MPP) tracker for maximizing the output
power of the panel, because the maximum power drawn from the
panel varies with temperature and insolation [11]. The grid-con-
nected stage uses a full-bridge inverter toward the grid, either
self-commutated with a high switching frequency [Fig. 1(a)], or
grid-commutated at the grid frequency [Fig. 1(b)]. In the former
structure [Fig. 1(a)], the panel voltage is firstly boosted to the grid
level together with the tracker. The dc/ac conversion stage, which
is usually a pulse-width-modulated (PWM) voltage-source
inverter, shapes and inverts the output current. A high-frequency
filter is used to eliminate the high-frequency component at the
inverter output. In the latter structure [Fig. 1(b)], the tracker,
voltage boost, and output current shaping are performed in the
front stage. The full bridge is switched at the grid frequency for
inverting the shaped output current [7], [9], [10].
As the inverter generates grid-conformed ac power, some en-
ergy storage elements, such as the dc-link capacitor and the
panel input capacitor , are used to compensate the difference
between the dc power from the panel and the time-varying in-
stantaneous power absorbed by the grid. They, particularly ,
determine the harmonic quality of the output current. For in-
verters without MPP tracking, is made large for keeping
the panel voltage constant and preventing the 100-Hz or 120-Hz
grid disturbance injecting into the panel. However, although the
terminal voltage is stable, the panel output current may be time
954 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 20, NO. 4, JULY 2005

Fig. 2. Circuit schematics of the proposed inverter.

varying or even pulsating that cannot stabilize the panel oper- is used to smooth the panel output current. The MPP tracking
ating point. It is thus important to keep both the panel terminal method in [11] is used. Its technique is to perturb the switching
voltage and output current constant at the MPP [12]. frequency of and compare the ac component and the average
Majority works usually emphasize on researching new value of the panel voltage to control the duty cycle of . The
MPP tracker [11] and refining individual converter stage. MPP can be globally located over wide insolation.
However, the system complexity is still governed by the mul- When is on, and are off. stores energy. When
tistage structure, as depicted in Fig. 1. This paper proposes a is off, both and are on. The energy stored in and
buck-boost-derived converter that integrates the functions of the will transfer to the output. will be switched off when the
MPP tracker and current shaper in Fig. 1(b). Low component integrated value of is larger than that of the required output
count and the use of low-voltage-rating capacitor are the key current in a switching period .
advantages of this converter. The output current harmonics acts as a buffer that temporarily stores panel energy and
are also less sensitive to the storage capacitor value. Tracking transfers it to the output in a switching cycle. If the panel power
of the MPP is based on the switching frequency modulation output is higher than the power absorbed by the grid in one
scheme (SFMS) in [11]. Shaping of the output current is based cycle, the energy remaining in after switching off will
on controlling the converters output pulse duration such that its transfer to through and . Since the average voltage
average value is proportional to the required current reference across will be shown to be equal to half of the peak output
in each switching cycle [13]. A 30-W laboratory prototype ac voltage , it is used to regulate , and so . Thus,
has been built. The tracking capability, inversion efficiency, the series connection of and in transferring energy to
and large-signal responses at different insolation have been the grid provides the advantages of shaping the current and re-
investigated. Detailed analysis on the inverter performance has ducing the voltage stress on .
been performed. The theoretical predictions are verified with The second stage is a grid frequency inverter formed by
the experimental results. switches , , , and . and are switched on
in the positive half cycle of the grid voltage, whilst and
II. OPERATING PRINCIPLES OF THE INVERTER are switched on in the negative half cycle of the grid
voltage. For the output filter section, is used to absorb the
A. Description of the Inverters Operations current difference between and the currents through
Fig. 2 shows the circuit schematics of the inverter consisting and when and are on. and have
of two stages. The first stage is a buck-boost-derived converter the resonant frequency at the switching frequency of for
operating in discontinuous conduction mode. It is formed by keeping the line-frequency and attenuating the high-frequency
, , , and for tracking the MPP of the panel and voltage components across and . is used to filter the
shaping the output current . The input filter formed by and high frequency current components going out to the grid. The
HO AND CHUNG: INTEGRATED INVERTER 955

If , will increase, and vice versa. As shown in

Fig. 3

(7)
and

(8)
Consider the time interval , in-
creases from to . Thus, for

(9)

By substituting and into (9), it

can be shown that
(10)

Fig. 3. Relationships among v , i , v , p , and p . Moreover

(11)
transformer is used to adjust the voltage level between the
inverter output and the grid voltage.
C. Operation of the Buck-Boost-Derived Converter
B. Relationship Between the Panel Power and Grid Power Fig. 4(a) shows the equivalent circuit of the front stage
The power input from the panel is constant under a con- buck-boost-derived converter operating in discontinuous con-
stant insolation and the power output to the grid is time- duction modethe inductor current begins with zero in
varying at twice of the grid frequency. Mathematically every switching cycle. The converter input is represented by a
voltage source , while its output is represented by a rectified
(1) sinusoidal voltage source . The components , , ,
(2) and form a classical buck-boost converter with its output
connecting in series with for supplying to . Fig. 4(b)(e)
(3) show the topological sequence of operations. Fig. 5 shows the
theoretical waveforms of the gate signal to , , , and
where , , and are the peak voltage, peak current, and , and the states of , , , and . The switching period
angular frequency of the grid, respectively, and is . The operations are described as follows.
is the peak instantaneous grid power. 1) Topology 1 [Fig. 4(b)]: The switch
Fig. 3 illustrates the relationships between and . At is on and is off. The inductor is charged up. The
the steady state, the average value of is equal to . That is diodes and are off. Hence

(12)
(4)
Hence, the peak current of , , is given by

The capacitor acts as a buffer for absorbing the difference (13)

between and . Fig. 3 depicts the low-frequency profile of
the capacitor voltage varying between and Also, the average current from , , is
at a dc value of . Thus
(14)
(5)
(6) It is noted that is relatively fixed under a constant
insolation, implying that a constant solar energy will
where . be converted into the grid.
956 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 20, NO. 4, JULY 2005

Fig. 5. Theoretical waveforms. (a) !t 6= (=2) or 3=2. (b) !t = =2 and

3=2.

2) Topology 2 [Fig. 4(c)]:

The switch is off and is on. The diode starts
conducting and blocks. The stored energy in is
released to , together with . Thus

(15)

Fig. 4. Equivalent circuit of the front stage converter. (a) Equivalent circuit. The output current , which is equal to , as
(b) Topology 1. (c) Topology 2. (d) Topology 3. (e) Topology 4. shown in Fig. 2, is sensed and integrated. If the inte-
HO AND CHUNG: INTEGRATED INVERTER 957

grated value is equal to or larger than the integrated By comparing (19) and (20), it can be seen that at ,
value of the reference current , will be turned for equality, the determinant must be
off. Mathematically zero. That is

(21)
(16)
Hence, from (4), (13) and (14)
The current reference is derived by multiplying
the rectified ac voltage and the controller output
.
3) Topology 3
[Fig. 4(d)]: The energy remaining in the coupled in-
ductors after Topology 2 will transfer to in this
topology. Both and are off. blocks and
starts conducting. Mathematically

(17)

However, when the instantaneous power to the grid is Therefore

at maximum (i.e., and ), the converter
will go to Topology 4 directly. The term (22)
equals zero at those moments and is the critical
condition for determining the value of , which will
Thus, the average value of (i.e., ) is load independent.
be shown in Section II-D. The waveforms are shown
Thus an error amplifier EA1 is used to compare the
in Fig. 5(b).
and and regulate the magnitude of the output current.
4) Topology 4 [Fig. 4(e)]:
As shown in Fig. 2, if is larger than , will be
All switches are off, until is switched on again in
increased, and vice versa.
Topology 1.

D. Determination of E. Turns-Ratio of the Coupled Inductors

As shown in Fig. 2, is used to adjust the magnitude of The primary-to-secondary turns-ratio of the coupled induc-
in (16) for determining . Thus, considering the wave- tors is determined by ensuring that still blocks in Topology
forms in Topology 2 shown in Fig. 5 and (16) 2. Thus, by referring Fig. 4(c)

(18)

where , and is the instantaneous (23)

output grid current before the voltage transformation by .
By solving (18), it can be shown that Thus, the R.H.S. of (23) is maximum when is maximum
(i.e., ). By using (22), if
(19)
(24)
Referring to Fig. 3, at , and
But, the voltage stresses on the switches increases with the value
. Thus, and , where
of . It can be shown that
and represent the peak values of the sinusoidal voltage
and current output respectively. The value of is at maximum, (25a)
which is denoted by . Thus, consider the area under the
waveform of in Fig. 5(b) (25b)

(25c)

(25d)
and so
where , , , are the voltage
(20) stresses of , , , and , respectively.
958 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 20, NO. 4, JULY 2005

F. Maximum Power Point Tracking of the Panel

Tracking the maximum power point (MPP) of the solar panel
is based on the frequency modulation technique in [11] that the
PWM signal to is modulated with a low-frequency signal.
The panel is modeled by a Thevenins equivalent circuit, which
consists of a voltage source connected in series with an output
resistance around the MPP. Both and are subject to
the level of insolation and temperature. The input voltage and
equivalent input resistance of the converter are and , re-
spectively. Assuming 100% conversion efficiency

(26)
Fig. 6. Setup of the laboratory prototype.

The rate of change of with respect to and can be

shown to be and

(27) (34)

At the MPP, and . Hence At the MPP, . and

. The tracking method is based on com-
paring the magnitudes of and , an error term is
(28) defined as

(35)
where is the average input voltage. This equation gives the
required dynamic input characteristics of the converter, in which where and are, respectively, the peak value of
has small-signal variation of subject to a small-signal and the scaling factor for . At the MPP, the converter
change of . will match with the panel. The sign of gives the required ad-
If the switching frequency of is modulated with a small- justment direction of the duty cycle of
signal sinusoidal variation
(36a)
(29)
(36b)
(36c)
where is the nominal switching frequency, is the modu-
lating frequency and is much lower than , and is the max-
imum frequency deviation. The tracking method in Fig. 2 is explained as follows. is mod-
consists of two components, including an average resis- ulated with a small-signal sinusoidal variation. and are
tance and a small variation . That is sensed. is then scaled down by the factor of and is com-
pared with . is obtained by using a peak detector to extract
(30) the value of the ac component in . The switching frequency
component in is removed by using a low-pass (LP) filter.
The error amplifier EA2 controls the PWM modulator to ad-
As shown in [14], for the buck-boost converter just by moving to the required value. The technique will
keep track of the output characteristics of solar panels without
(31) approximating the voltage-current relationships [11].

III. EXPERIMENTAL VERIFICATION

By differentiating (31), it can be shown that
has an average value of
can be expressed as
A laboratory prototype has been built and investigated. The
input side consists of three 10-W panels (Siemens SM-10)
(32)
connecting in series. A stainless steel cover is put on the panels
to ensure uniform illumination on the panels. The setup is
and a small variation , which shown in Fig. 6. The surface temperature is kept at about 40 C
throughout the test. The output power to the output resistance
characteristics of the panel at different insolation
levels is shown in Fig. 7. varies from 38 to 122 when
(33)
the insolation level is changed from 1500 W/m to 350 W/m .
HO AND CHUNG: INTEGRATED INVERTER 959

Fig. 7. p 0r characteristics of the panel.

TABLE I
COMPONENT VALUES OF THE LABORATORY PROTOTYPE

The output power is also decreased from 34 W to 12 W. The

component values and parameters of the prototype inverter are
tabulated in Table I. The switching frequency of is 20 kHz,
with a maximum frequency deviation of 2 kHz (10%) using a
1 kHz modulating signal.
The system is simulated with PSIM 4.1 [15]. Fig. 8(a) shows
the simulated voltage and current waveforms of and at
the insolation level of 1500 W/m . Fig. 8(b) and (c) show the Fig. 8. Simulated voltage and current waveforms. (a) v , i . (b) v , i
waveforms of , , and at and , re- and i at !t = =4. (c) v , i and i at !t = =2.
spectively. Fig. 9 shows the corresponding experimental wave-
forms at the insolation levels of 1500 W/m . Figs. 10 and 11
show the waveforms of , , and under the insolations 220 F and compared with the IEEE Std. 929-2000 (IEEE Rec-
of 750 W/m and 350 W/m , respectively. It can be seen that ommended Practice for Utility Interface of Photovoltaic (PV)
the results under the insolation level of 1500 W/m are in close Systems) [16]. They are all within the harmonic limits. Table II
agreement with the simulated results and the ones in Fig. 5. The shows the tracking and conversion efficiencies of the system at
average value of the capacitor voltage is around 30 V, which different insolation levels. The tracking efficiency, which is the
is equal to , under the three insolation levels. This con- ratio of the tracked power to the maximum extractable power,
firms the expression in (22). Fig. 12 shows the harmonic con- can be maintained at over 96% over the insolation range. The
tents of the output current at the insolation level of 1500 W/m overall conversion efficiency from the panel to the grid is over
with different values of , including 1000 F, 470 F, and 83%.
960 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 20, NO. 4, JULY 2005

Fig. 10. Experimental voltage and current waveforms at the insolation level
of 750 W/m . v (Ch 3: 100 V/div), i (Ch 4: 0.1 A/div) and v (Ch 2:
10 V/div) (Timebase: 2 ms/div).

Fig. 11. Experimental voltage and current waveforms at the insolation level
of 350 W/m . v (Ch 3: 100 V/div), i (Ch 4: 0.1 A/div) and v (Ch 2:
10 V/div) (Timebase: 2 ms/div).

Fig. 9. Experimental voltage and current waveforms at the insolation level of

1500 W/m . (a) v (Ch 3: 100 V/div), i (Ch 4: 0.1 A/div) and v (Ch 2:
10 V/div) (Timebase: 2 ms/div). (b) !t = =4. v (Ch 1: 10 V/div), i (Ch
2: 5 A/div) and i (Ch 3: 5 A/div) (Timebase: 20 s=div ). (c) !t = =2.
v (Ch 1: 10 V/div), i (Ch 2: 5 A/div) and i (Ch 3: 5 A/div) (Timebase: Fig. 12. Harmonic contents of i at the insolation level of 1500 W/m with
20 s=div ). different values of C , including 1000 F, 470 F, and 220 F.

The transient response of the inverter system is investigated 1500 W/m . Fig. 13(a) shows the waveforms of , and .
by suddenly changing the insolation level from 350 W/m to Fig. 13(b) shows the response when the insolation is changed
HO AND CHUNG: INTEGRATED INVERTER
TABLE II
TRACKING AND INVERSION EFFICIENCIES OF THE PROTOTYPE
Fig. 13. Transient responses of the inverter system. (Ch 2: v , 10 V/div; Ch
4: i , 0.1 A/div), (Timebase: 100 ms/div). (a) From 350 W/m to 1500 W/m .
(b) From 1500 W/m to 350 W/m .
from 1500 W/m to 350 W/m . With the insolation level in-
creased, the output current cannot be increased instantaneously.
Thus, the capacitor voltage increases during the transient pe-
riod. Conversely, with the insolation level decreased, the output
current cannot be decreased instantaneously. Thus, the capacitor
voltage decreases during the transient period. Nevertheless, the
961
capacitor will revert back to 30 V in both cases after the tran-
sient period, confirming the control mechanism. The transient
period takes about 0.6 s.
IV. CONCLUSION
This paper has discussed the implementation and perfor-
mance of an integrated inverter for photovoltaic systems. The
inverter provides the function of tracking the MPP of the panels
and shaping the output current over wide insolation level.
A 30-W laboratory prototype has been built. The tracking
capability, inversion efficiency, and large-signal responses at
different insolations have been investigated. Theoretical pre-
dictions have been confirmed with the experimental results of
the prototype.
REFERENCES
[1] G. Foley, Photovoltaic applications in rural area of the developing
world, Tech. Rep., World Bank, New York, 1995.
[2] C. J. Hatiziadoniu, F. Chalkiadakis, and V. Feiste, A power conditioner
for a grid-connected photovoltaic generator based on the 3-level in-
verter, IEEE Trans. Energy Conv., vol. 14, no. 4, pp. 16051610, Dec.
1999.
[3] K. Y. Khouzam, Technical and economic assessment of utility inter-
active PV systems for domestic applications in South East Queens-
land, IEEE Trans. Energy Conv., vol. 14, no. 4, pp. 15441550,
Dec. 1999.
[4] T. Markvart, Solar Electricity. New York: Wiley, 1994.
[5] H. Becker, K. Kiefer, V. Hoffmann, U. Rindelhard, and G. Heilscher,
Five years of operational experience in the German 1000-Roofs-PV
programmeresults of monitoring and system inspection, in Proc. 14th
Eur. Photovoltaic Solar Energy Conf., Barcelona, Spain, Jul. 1997, pp.
16771680.
[6] H. Oldenkamp, I. De Jong, C. Baltus, S. Verhoeven, and S. Elstgeest,
Reliability and accelerated life tests of the ac module mounted OKE4
inverter, in Proc. 25th IEEE Photovoltaic System Conf., May 1996, pp.
13391342.
[7] A. Lohner, T. Meyer, and A. Nagel, A new panel-integratable inverter
concept for grid-connected photovoltaic systems, in Proc. IEEE Int.
Symp. Industrial Electronics, Jun. 1996, pp. 827831.
[8] M. Meinhardt et al., Miniaturized low profile module integrated con-
verter for photovoltaic applications with integrated magnetic compo-
nents, in Proc. IEEE Applications Power Electron. Conf. Expo., Mar.
1999, pp. 305311.
[9] M. Calais, J. Myrzik, T. Spooner, and V. Agelidis, Inverters for
single-phase grid connected photovoltaic systemsan overview,
in Proc. IEEE Power Electronic Specialists Conf., Jun. 2002, pp.
19952000.
[10] S. B. Kjaer, J. K. Pedersen, and F. Blaabjerg, Power inverter topologies
for photovoltaic modulesa review, in Proc. IEEE Industry Applica-
tions Conf., vol. 2, 2002, pp. 782788.
[11] K. K. Tse, M. T. Ho, H. Chung, and S. Y. R. Hui, A novel maximum
power point tracker for PV panels using switching frequency modula-
tion, IEEE Trans. Power Electron., vol. 17, no. 6, pp. 980989, Nov.
2002.
[12] Y. Chen, K. Smedley, F. Vacher, and J. Brouwer, A new maximum
power point tracking controller for photovoltaic power generation, in
Proc. IEEE Applications Power Electronic Conf. Expo, Mar. 2003, pp.
5862.
[13] K. Smedley and S. Ck, One-cycle control of switching converters,
IEEE Trans. Power Electron., vol. 10, no. 6, pp. 625633, Nov.
1995.
[14] D. Maksimovic and S. Ck, A unified analysis of PWM converters in
discontinuous modes, IEEE Trans. Power Electron., vol. 6, no. 3, pp.
476490, Jul. 1991.
[15] Users Manual of PSIM 4.1, Powersim Technologies Inc., 2000.
[16] IEEE Recommended Practice for Utility Interface of Photovoltaic (PV)
Systems, IEEE Std. 929-2000, Jan. 2000.
962
Billy M. T. Ho (S02) was born in Hong Kong in
1979. He received the B.Eng. (first class honors) de-
gree in electronic engineering from the City Univer-
sity of Hong Kong in 2001 where he is currently pur-
suing the Ph.D. degree in power electronics.
His research interests include computer-aided
simulation techniques, photovoltaic panels power
conversion, control methodologies, and inverter
applications.
Mr. Ho received the Simatelex Charitable Foun-
dation Scholarships for two consecutive years during
his undergraduate studies.
IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 20, NO. 4, JULY 2005
Henry Shu-Hung Chung (M95SM03) received
the B.Eng. and Ph.D. degrees in electrical engi-
neering from The Hong Kong Polytechnic University
in 1991 and 1994, respectivelty.
Since 1995, he has been with the City University
of Hong Kong (CityU). He is currently a Professor
in the Department of Electronic Engineering and
Chief Technical Officer of e.Energy Technology
Limited (an associated company of CityU). He
has authored five research book chapters, and over
200 technical papers including 90 refereed journal
papers in his research areas, and holds four US patents. His research interests
include time- and frequency-domain analysis of power electronic circuits,
switched-capacitor-based converters, random-switching techniques, control
methods, digital audio amplifiers, soft-switching converters, photovoltaic
systems, and electronic ballast design.
Dr. Chung received the Grand Applied Research Excellence Award from
the City University of Hong Kong in 2001. He was IEEE Student Branch
Counselor and Track Chair of the Technical Committee on Power Electronics
Circuits and Power Systems of the IEEE Circuits and Systems Society,
from 1997 to 1998. He was Associate Editor and Guest Editor of the IEEE
TRANSACTIONS ON CIRCUITS AND SYSTEMS, PART I: FUNDAMENTAL THEORY
AND APPLICATIONS, from 1999 to 2003. He is currently an Associate Editor of
the IEEE TRANSACTIONS ON POWER ELECTRONICS.