Download High-Frequency AC-Link PV Inverter

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 61, NO. 1, JANUARY 2014
281
High-Frequency AC-Link PV Inverter
Mahshid Amirabadi, Student Member, IEEE, Anand Balakrishnan,
Hamid A. Toliyat, Fellow, IEEE, and William C. Alexander, Member, IEEE
Abstract—In this paper, a high-frequency ac-link photovoltaic
(PV) inverter is proposed. The proposed inverter overcomes most
of the problems associated with currently available PV inverters.
In this inverter, a single-stage power-conversion unit fulfills all the
system requirements, i.e., inverting dc voltage to proper ac, stepping up or down the input voltage, maximum power point tracking, generating low-harmonic ac at the output, and input/output
isolation. This inverter is, in fact, a partial resonant ac-link converter in which the link is formed by a parallel inductor/capacitor
(LC) pair having alternating current and voltage. Among the
significant merits of the proposed inverter are the zero-voltage
turn-on and soft turn-off of the switches which result in negligible
switching losses and minimum voltage stress on the switches.
Hence, the frequency of the link can be as high as permitted by the
switches and the processor. The high frequency of operation makes
the proposed inverter very compact. The other significant advantage of the proposed inverter is that no bulky electrolytic capacitor
exists at the link. Electrolytic capacitors are cited as the most
unreliable component in PV inverters, and they are responsible for
most of the inverters’ failures, particularly at high temperature.
Therefore, substituting dc electrolytic capacitors with ac LC pairs
will significantly increase the reliability of PV inverters. A 30-kW
prototype was fabricated and tested. The principle of operation
and detailed design procedure of the proposed inverter along
with the simulation and experimental results are included in this
paper. To evaluate the long-term performance of the proposed
inverter, three of these inverters were installed at three different
commercial facilities in Texas, USA, to support the PV systems.
These inverters have been working for several months now.
Index Terms—Inverters, photovoltaic (PV) systems, zerovoltage switching.
I. I NTRODUCTION
D
ISTRIBUTED energy (DE) systems have been receiving
wide acceptance during the last decade. Among the different types of distributed energy resources, photovoltaics (PV)
and wind are more popular.
Power electronics is an integral part of these DE systems as
they convert generated electricity into utility-compatible forms.
However, the addition of power electronics usually adds costs
as well as certain reliability issues [1], [2].
Manuscript received May 10, 2012; revised August 12, 2012 and
October 24, 2012; accepted December 10, 2012. Date of publication
February 6, 2013; date of current version July 18, 2013.
M. Amirabadi and H. A. Toliyat are with Texas A&M University, College
Station, TX 77843 USA (e-mail: [email protected]; [email protected]).
A. Balakrishnan was with Texas A&M University, College Station,
TX 77843 USA. He is now with Freescale Semiconductor, Inc., Tempe,
AZ 85284 USA (e-mail: [email protected]).
W. C. Alexander is with Ideal Power Converters, Inc., Austin, TX 78731
USA (e-mail: [email protected]).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TIE.2013.2245616
Fig. 1. Centralized converter-based PV system.
Fig. 2. Multiple-stage conversion systems used in PV applications.
Referring to a report by Sandia National Laboratories [3],
inverters are responsible for most of the PV system incidents
triggered in the field. They are costly and complex, and their
current mean time to first failure is unacceptable. Inverter
failures contribute to unreliable PV systems, which may result
in loss of confidence in renewable technology. Therefore, to
achieve long-term success in the PV industry, new power converters with higher reliability and longer lifetimes are required
[3], [4].
In past designs, a centralized converter-based PV system was
the most commonly used type of PV system. As shown in
Fig. 1, in this system, PV modules are connected to a threephase voltage-source inverter. The output of each phase of the
inverter is connected to an LC filter to limit the harmonics. A
three-phase transformer, which steps up the voltage and provides galvanic isolation, connects the inverter to the utility [1].
Low-frequency transformers are considered poor components mainly due to their large size and low efficiency. To avoid
low-frequency transformers, multiple-stage conversion systems
are widely used in PV systems [1], [5], [6]. The most common
topology, represented in Fig. 2, includes a dc–ac voltage source
inverter and a dc–dc converter. Commonly, the dc–dc converter
contains a high-frequency transformer [1]. Despite offering a
high boosting capability and galvanic isolation, this converter
consists of multiple power-processing stages which lower the
efficiency of the overall system. Moreover, bulky electrolytic
capacitors are required for the dc link. Electrolytic capacitors,
which are very sensitive to temperature, may cause severe
0278-0046/$31.00 © 2013 IEEE
282
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 61, NO. 1, JANUARY 2014
Fig. 3. Proposed PV inverter.
reliability problems for inverters, and an increase of even 10 ◦ C
can halve their lifetime [7]. Therefore, PV inverters containing
electrolytic capacitors are not expected to provide the same
lifetime as PV modules. Consequently, the actual cost of the
PV system involves the periodic replacement of the inverter,
which increases the levelized cost of energy extracted from the
PV system [8].
Considering the aforementioned problems, it is essential to
support the design of alternative inverter topologies to reduce
the cost of inverters while increasing their reliability [7].
Several solutions have been proposed to partially overcome
these problems. Reference [5] introduces a transformerless
topology in which the ground leakage current is minimized;
however, the voltage cannot change over a wide range.
Reference [9] proposes an integrated solution for a PV/FCbased hybrid distributed generation system to eliminate the
requirement of high-voltage buffer capacitors for inverters. This
type of inverter, nonetheless, does not provide isolation. In
[10], large electrolytic capacitors are replaced by small film
capacitors; however, the proposed solution is merely applicable
in low-power PV systems.
A number of resonant PV inverters have been proposed
as well [11], [12]. A high-frequency link PV power conditioning system which includes a high-frequency resonant inverter, a rectifier, and a line-commutated inverter, operating
near unity power factor, has been proposed in [11]. Three
other resonant PV inverters are introduced in [12]: highfrequency resonant inverter cycloconverter, high-frequency resonant inverter–rectifier pulsewidth-modulated voltage source
inverter, and high-frequency resonant inverter–rectifier lineconnected inverter. All of these resonant PV inverters contain
multiple stages. The first and fourth inverters require a large
inductor for dc current link, and the third configuration needs a
large dc-link capacitor.
A high-frequency ac-link PV inverter which overcomes most
of the problems associated with existing inverters is proposed in
this paper. The proposed inverter is a partial resonant converter,
i.e., only a small time interval is allocated to resonance in each
cycle. Hence, while the resonance facilitates the zero-voltage
turn-on of the switches, the LC link has low reactive ratings and
low power dissipation. Due to the soft switching, the switching
losses in this inverter are negligible and the frequency of the
link can be very high, resulting in a compact link inductor.
Moreover, the zero-voltage turn-on results in reduced voltage
stress on the switches.
The other merit of the proposed inverter is the elimination
of the dc link and the replacement of the bulky electrolytic
capacitors employed in the multiple-stage conversion systems
with an LC pair having alternating current and voltage.
The present authors have introduced the partial resonant PV
inverter in [13], without including experimental verification.
Other applications of the proposed inverter were studied in
[14]–[18]. This paper presents the results of the simulation and
experimental test of a 30-kW prototype, along with a detailed
design procedure.
II. P ROPOSED C ONFIGURATION AND P RINCIPLES OF
O PERATION
Fig. 3 represents the proposed PV inverter. Four unidirectional switches forming the PV switch bridge interface the PV
modules to the link, and six bidirectional switches forming the
ac-side switch bridge connect the same link to the load. Since
PV cells cannot absorb electrical energy, the associated switch
bridge is formed merely by unidirectional switches. As will
be discussed in Section IV, the unidirectional switches at the
PV side would need to be replaced by bidirectional switches
if the inverter is required to inject reactive power during the
grid fault.
The converter transfers power entirely through the link inductor. This inverter is, in essence, a dc–ac buck–boost converter in which the link is charged through the PV and then
discharged into the output phases. The charging and discharging take place alternately. The frequency of charge/discharge is
called the link frequency and is typically much higher than the
output line frequency.
Complimentary switches on each leg facilitate the charging
and discharging of the link in a reverse direction, leading to an
alternating current in the link. The alternating current of the link
results in better utilization of the inductor and capacitor.
Fig. 4 depicts the link current in a soft-switching ac-link
buck–boost inverter. As represented in this figure, between each
charging and discharging, there is a resonating mode during
which none of the switches conduct and the LC link resonates
to facilitate the zero-voltage turn-on of the switches.
The input and output current pulses, resulting from alternating charge and discharge, have to be precisely modulated
to meet the references. The link will be charged through the
PV modules; however, there are two output phase pairs to be
charged through the link. In order to have better control on the
AMIRABADI et al.: HIGH-FREQUENCY AC-LINK PV INVERTER
283
turn-on of S0 and S3. The simple equivalent circuit shown
in Fig. 8(a) can be used to analyze the circuit during
mode 1. The link current (iLink ) during mode 1 can be
calculated using the following:
diLink (t)
dt
t
1
VPV × t
iLink (t) =
+ iLink (0).
VPV dt =
L
L
VPV = L
(1)
(2)
0
Fig. 4.
Link current in an ac-link buck–boost inverter.
Fig. 5.
Link current in the proposed inverter.
output currents and minimized input and output harmonics, a
link discharging mode is split into two modes. This concept is
presented in Fig. 5.
The general control strategy is to turn on each switch
such that its turn-on occurs at zero voltage and to turn off
the switches when the input or output currents meet their
references.
The negligible switching losses of this inverter allows the use
of slower switches with higher current density, possibly with
soft-switched optimized structures or, alternatively, higher link
frequencies.
Similar to a dc–dc buck–boost converter, the input and output
of this inverter appear as voltage sources. Therefore, filter
capacitors are placed across the input and output terminals
(as represented in Fig. 3).
The basic operating modes and relevant waveforms of this
inverter are represented in Figs. 6 and 7, respectively.
The link cycle is divided into 12 modes, with six power
transfer modes and six resonating modes. The link is energized
from the input during modes 1 and 7 and is de-energized to the
outputs during modes 3, 5, 9, and 11. Modes 2, 4, 6, 8, 10, and
12 are devoted to resonance.
A detailed description of the modes of operation is as
follows.
1) Mode 1 (Energizing): Before the start of mode 1, proper
switches on the PV switch bridge, which are supposed
to conduct during mode 1, are activated (S0 and S3 in
Figs. 6 and 7); however, they do not immediately conduct
since they are reverse biased. Once the link voltage,
which is resonating before mode 1, becomes equal to the
PV voltage, the proper switches (S0 and S3) are forward
biased, initiating mode 1. This results in the zero-voltage
In (1) and (2), L and VPV are the link inductance and PV voltage, respectively. During this mode,
the link voltage is equal to VPV , as shown in Fig. 7.
The link is charged until the PV current, averaged over
a cycle time, meets its reference value. The switches
located on the PV switch bridge are then turned off.
As mentioned earlier, the link capacitor acts as a buffer
across the switches during their turn-off, which results in
low turn-off losses.
2) Mode 2 (Partial resonance): During mode 2, none of the
switches conduct and the link resonates, causing the link
voltage to decrease. In this mode, the circuit behaves as a
simple LC circuit, as depicted in Fig. 8(b). The following
equation describes the behavior of the LC circuit (ic (t),
VLink (t), and C are the capacitor current, link voltage,
and link capacitance, respectively)
ic (t) = C
dVLink (t)
.
dt
(3)
As shown in Fig. 8(b), the current passing through
the capacitor is equal to “−iLink (t)” and the inductor
current is positive, resulting in negative dVLink (t)/dt.
This implies that the link voltage is decreasing. Once the
link voltage becomes negative, the controller determines
which output phase pairs have to be charged through the
link during modes 3 and 5. The sum of the output reference currents at any instant is zero. One of them is the
highest in magnitude and of certain polarity while the two
lower ones are of the opposite polarity. Although there are
three phase pairs in a three-phase system, considering the
polarity of the current in each phase, only two of these
phase pairs can provide a path of current when connected
to the link. Therefore, the charged link transfers power
to the output by discharging into two phase pairs. The
two phase pairs are the one formed by the phase having
the highest and second highest reference currents and the
one formed by the phase having the highest and lowest
reference currents, where the reference currents are sorted
as highest, second highest, and lowest in terms of magnitude alone. For example, if Ia_o = −10 A, Ib_o = 7 A,
and Ic_o = 3 are the three output reference currents, then
phase pairs ab and ac are chosen to transfer power to the
output. The phase pair with smaller line-to-line voltage
is the first one the link is discharged to. For example, if
Vab_o = 300 V and Vac_o = 400 V, the link will be first
discharged into phase pair ab.
284
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 61, NO. 1, JANUARY 2014
Fig. 6. Circuit behavior in different modes. (a) Mode 1. (b) Modes 2, 4, and 6. (c) Mode 3. (d) Mode 5.
In Figs. 6 and 7, it is assumed that the link will
be discharged into output phase pairs ab and ac during
modes 3 and 5, respectively.
During mode 2, the output switches which are supposed to conduct during modes 3 and 5 are turned on
(S18, S22, and S23 in Figs. 6 and 7); however, being
reverse biased, they do not conduct immediately. Once
the link voltage reaches Vab−o (assuming Vab−o is smaller
than Vac−o ), switches S18 and S22 will be forward biased
and will start to conduct, initiating mode 3.
3) Mode 3 (De-energizing): The output switches (S18 and
S22) are turned on at zero voltage to allow the link to
be discharged into the chosen phase pair until the current
of phase b at the output side averaged over a cycle time
meets its reference. At this point, S22 will be turned off,
initiating another resonating mode.
4) Mode 4 (Partial resonance): The link is allowed to swing
to the voltage of the other output phase pair chosen during
mode 2. For the case shown in Figs. 6 and 7, the link
voltage swings from Vab−o to Vac−o .
AMIRABADI et al.: HIGH-FREQUENCY AC-LINK PV INVERTER
285
represented in Fig. 9. The principle of operation of the inverter
with galvanic isolation is similar to that of the original inverter.
Fig. 10 represents the system control block diagram.
III. D ESIGN P ROCEDURE AND A NALYSIS
In this part, a detailed design procedure will be presented.
For the sake of simplicity, the resonating time, which is much
shorter than the power transfer time at full power, will be
neglected. Moreover, de-energizing the link is assumed to take
place in one equivalent mode instead of two modes in each
power cycle (each power cycle is half the link cycle). For this,
the link is assumed to be discharged into a virtual output phase
with the output equivalent current and voltage. Considering the
fact that the phase carrying the maximum output current is involved in the de-energizing of the link during both modes 3 and
5, the output equivalent current can be calculated as follows:
Io−eq =
Fig. 7. Voltage and current waveforms showing the behavior of the circuit
during different modes of operation.
3Io,peak
π
(4)
where Io,peak is the output peak phase current. It can be shown
that the output equivalent voltage is equal to
Vo−eq =
πVo,peak
cos θo
2
(5)
where Vo,peak and cos(θo ) are the output peak phase voltage
and output power factor, respectively.
The following equations describe the behavior of the circuit
during the energizing (charging) and de-energizing (discharge)
of the link:
Fig. 8.
Equivalent circuit during (a) energizing and (b) resonating modes.
5) Mode 5 (De-energizing): During mode 5, the link discharges to the selected output phase pair until there is just
sufficient energy left in the link to swing to a predetermined voltage (Vmax ), which is slightly higher than the
maximum input and output line-to-line voltages. At the
end of mode 5, all the switches are turned off, allowing
the link to resonate during mode 6.
6) Mode 6 (Partial resonance): During mode 6, the link
voltage swings to Vmax , and then, its absolute value decreases. Proper switches, which are supposed to conduct
during mode 7, are turned on during this mode; however,
they do not conduct as they are reversed biased.
Modes 7 to 12 are similar to modes 1 to 6, except that the
link charges and discharges in the reverse direction. For this, the
complimentary switch on each leg is switched when compared
with the ones switched during modes 1 to 6.
As seen in Fig. 6, the input- and output-side switches never
conduct simultaneously, which implies that the input and output
are isolated. However, if galvanic isolation is required, a singlephase high-frequency transformer can be added to the link.
In this case, the magnetizing inductance of the transformer
can play the role of the link inductance. This configuration is
VPV × tcharge
L
Vo−eq × tdischarge
.
=
L
ILink, Peak =
(6)
ILink, Peak
(7)
In the aforementioned equations, ILink,peak , tcharge , and
tdischarge represent the peak of the link current, the total charging time during mode 1, and the total discharging time during
modes 3 and 5, respectively.
Equations (6) and (7) determine the relationship between the
charging and discharging times as follows:
tcharge =
Vo−eq × tdischarge
.
VPV
(8)
On the other hand, the average of the PV current, which is
equal to the PV power over PV voltage, can be determined
based on the peak of the link current as follows:
1
(9)
IPV = (ILink,Peak × tcharge ).
T
In the aforementioned equation, T is the period of the link
current. Using (6)–(9), the following can be derived:
VPV
P
ILink,Peak = 2
1+
.
(10)
VPV
Vo−eq
This expression determines the link peak current based on
the nominal power (P ) and input and output voltages. Equation
(10) can be rewritten as follows:
ILink,Peak = 2 × (IPV + Io−eq ).
(11)
286
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 61, NO. 1, JANUARY 2014
Fig. 9. Proposed inverter with galvanic isolation.
Fig. 10. Control block diagram of the proposed inverter.
This implies that the PV and output currents determine the
link peak current. The frequency of the link can be chosen based
on the power rating of the system and the characteristics of
the available switches. Once the link frequency is chosen, the
following determines the inductance of the link:
L=
1
VPV
ILink, Peak 2f 1 +
VPV
Vo−eq
.
TABLE I
PARAMETERS OF THE D ESIGNED I NVERTER
(S IMULATED AND FABRICATED )
(12)
Link capacitance (C) is chosen so as to keep the resonating
periods within 5% of the link cycle at full power. It should be
noted that the resonance time is not negligible at low power
levels.
In this converter, the main sources of power dissipation are
the conduction loss of the switches and the core and copper
losses of the inductor. The switching losses are negligible,
as the switches turn on at zero voltage and their turn-off is
capacitance buffered.
The core and copper losses of the inductor are proportional
to the link frequency and the link rms current, respectively.
Therefore, by increasing the power level, the former decreases
whereas the latter increases.
IV. S IMULATION AND E XPERIMENTAL R ESULTS
A 30-kW PV inverter has been designed, and the parameters
are summarized in Table I. The link capacitance in the prototype
is 0.8 μF, which is much higher than the amount achieved
through the design procedure. Although the resonating time is
preferred to be as short as possible, it should be long enough
to allow the microcontroller or processor to turn on the proper
switches prior to the next energizing or de-energizing mode.
Due to the processor limitations, the link capacitance has been
chosen as 0.8 μF in the prototype.
The designed PV inverter has been simulated in PSim, and
Figs. 11–21 represent the simulation results. In Figs. 11–14
and 16, the link capacitance is 0.8 μF.
AMIRABADI et al.: HIGH-FREQUENCY AC-LINK PV INVERTER
287
Fig. 15. Link current and voltage at full power, using 0.1-μF link capacitance.
Fig. 11. PV current and voltage at full power.
Fig. 16. Link current and voltage at 15 kW.
Fig. 12. AC-side current and voltage at full power.
Fig. 17. AC-side current and voltage when the irradiance drops from 850 to
650 w/m2 .
Fig. 13. Link voltage at full power.
Fig. 18. AC-side current and voltage when the temperature changes from
25 ◦ C to 50 ◦ C.
Fig. 14. Link current at full power.
Fig. 11 represents the input voltage and current at full power.
The input voltage is 700 V, and the input current is 43 A. The
output current along with the output voltage are illustrated in
Fig. 12. In this case, the power factor is close to unity and the
frequency is 60 Hz.
Figs. 13 and 14 represent the link voltage and current, respectively. The link current and voltage are both alternating. Since
the frequency of the link (7200 Hz) is much higher than the
frequency of the line (60 Hz), the link voltage and current are
illustrated in these figures over a short time interval. As it can
be seen in these figures that the predetermined voltage to which
the link resonates during mode 6 is set at 850 V. Moreover,
the peak of the link current is 200 A, although during mode 1,
288
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 61, NO. 1, JANUARY 2014
Fig. 19. Proposed inverter configuration for providing LVRT.
Fig. 20. AC-side current and voltage when the AC-side voltage drops to 10%
of its nominal value (at t = 0.016 s).
Fig. 21. PV current and voltage when the AC-side voltage drops to 10% of its
nominal value (at t = 0.016 s).
the link current is increased to 180 A. During mode 2, as the
link resonates, its peak current reaches 200 A. As previously
mentioned, the higher the link capacitance is, the longer the
resonating time will be. Fig. 15 depicts the link current and
voltage at full power using 0.1-μF link capacitance. As seen
in this figure, the peak of the link current in this case is very
close to the calculated value [using (10)] which is 180 A.
At lower power levels, the peak of the link current will
be lower than that of the full power. Fig. 16 represents the
link voltage and current at 15 kW. As seen in this figure,
the predetermined voltage to which the link resonates during
mode 6 is still set at 850 V. Although the PV voltage has
dropped to 350 V, the maximum output line-to-line voltage is
still 679 V. The peak of the link current in this case is about
147 A. Again, the resonating time is noticeable due to the use
of 0.8-μF link capacitance.
As mentioned earlier, one of the merits of this inverter is
the possibility of both stepping up and down the voltage. In
Figs. 11–15, the peak of the line-to-line output voltage is
lower than the input voltage so the inverter is, in fact, stepping
down the voltage. Meanwhile, in Fig. 16, the input voltage is
lower than the peak of line-to-line output voltage, and thus, the
inverter is stepping up the voltage.
Maximum power point tracking (MPPT) is essential in PV
inverters. Similar to other PV inverters, the proposed inverter
can perform MPPT as it can follow any references within
its range. Fig. 17 represents the ac-side current/voltage when
the irradiance drops from 850 to 650 w/m2 (at t = 20 ms).
The temperature is assumed to be 25 ◦ C. Fig. 18 depicts the
ac-side current/voltage when the temperature increases from
25 ◦ C to 50 ◦ C (at t = 20 ms). Although a change in temperature usually takes place gradually, in order to see the effect of
this change on the behavior of the proposed inverter, a sharp
change is considered in this simulation. For the case shown in
Fig. 18, the irradiance is 850 w/m2 .
Another important issue to study is the behavior of the
proposed inverter during the grid fault. Although this inverter
does not have a dc link, it can inject reactive power into the grid
during voltage sags. To provide the low-voltage ride-through
(LVRT) feature, the PV-side switches should be replaced by
bidirectional switches, as illustrated in Fig. 19. The principle
of operation of the inverter during the grid fault is slightly
different from that of the normal operation. During mode 1,
the link will be charged through the PV up to a certain level.
Then, similarly to the normal operation, it will be discharged
into the output phases; however, no net energy is taken from
the link in this case. After the output-side switches are turned
off, the energy stored in the link is discharged into the input.
Once the link is completely discharged, it will be recharged
from the PV with current flowing in the opposite direction.
In this case, the PV-side filter capacitor absorbs the energy
discharged into the input. Therefore, if injecting reactive power
to the grid is necessary during the grid faults, this capacitor
needs to be chosen such that it meets the system requirements.
This method of control can also be used for normal operation.
In the case of the normal operation, the remnant energy of the
link after turning off the output-side switches, which needs to be
discharged into the PV-side capacitor, is much lower than that
of the grid-fault case. Fig. 20 depicts the ac-side current/voltage
when the output voltage drops to 10% of its nominal value (at
t = 0.016 s). As seen, the power factor is almost zero in this
case. The PV current and voltage are represented in Fig. 21.
The active power drawn from PV in this case is equal to the
losses of the inverter.
AMIRABADI et al.: HIGH-FREQUENCY AC-LINK PV INVERTER
289
Fig. 22. Thirty-kilowatt prototype.
Fig. 25.
power.
Link current (50 A/division) and voltage (400 Vdc/division) at full
Fig. 23. DC-side current (20 A/division) and voltage (400 Vdc/division).
Fig. 26. AC-side current (20 A/division) and voltage (200 V/division) of the
grid-tied 30-kW inverter running at 25 kW.
Fig. 24. AC-side voltage (200 Vdc/division) and current (30 A/division).
A 30-kW prototype, as shown in Fig. 22, has been designed,
fabricated, and tested. This inverter has been installed in a
commercial facility and has been connected to the grid for
the test procedure. The parameters of the fabricated prototype
are the same as those of the simulated inverter. The link
capacitance was chosen as 0.8 μF. Fig. 23 represents the
input voltage and current at full power. The input voltage at
full power is 700 V, and the input current is set at 45 A. The
difference between this current and the input current in the
simulation (43 A) is due to the power losses which could not be
modeled in the simulation. The output current and voltage are
depicted in Fig. 24. Similar to the simulation, the peak of the
output current is regulated at 51 A. Fig. 25 represents the link
voltage and current waveforms while the prototype is running
at full power. The peak of the current is almost 200 A, which is
the same as the simulation result.
Fig. 27. FFT of the line-to-neutral grid voltage.
Fig. 26 represents the ac-side current and voltage when the
inverter is operating at 25 kW. As seen in Figs. 24 and 26,
the ac-side current contains some low-frequency harmonics,
the source of which is the low-frequency components of the
grid voltage. Due to the existence of nonlinear loads, the grid
voltage is polluted by harmonics. Fig. 27 shows the fast Fourier
transform (FFT) of the line-to-neutral voltage when the inverter
is turned off. To better show the low-frequency harmonics,
the zoomed-in image of the FFT is represented. Regardless
of the type of inverter employed, these low-frequency voltage
harmonics generate low-frequency current harmonics. Fig. 28
depicts the FFT of the line current corresponding to Fig. 26. As
seen in this figure, the line current contains the same harmonics
as the grid voltage. Studying the causes of the power line
290
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 61, NO. 1, JANUARY 2014
Fig. 28. FFT of the line current when the inverter is operating at 25 kW.
Fig. 31.
Fig. 29.
15 kW.
PV current (5 A/division) and voltage (200 Vdc/division) at 6.2 kW.
Link current (50 A/division) and voltage (400 Vdc/division) at
Fig. 32. AC-side current (20 A/division) and side voltage (200 Vdc/division)
at 6.2 kW.
Fig. 30. Link current (50 A/division) and voltage (400 Vdc/division) at
3.75 kW.
distortion and the solutions to improve it is beyond the scope
of this paper.
The link current and voltage when the prototype is operating
at 15 kW and the PV voltage is 350 V are represented in Fig. 29.
As shown in this figure, the peak of the link current is 144 A,
which is very close to the simulation results. This inverter is
capable of operating at very low power levels as well. Fig. 30
represents the link current and voltage at 3.75 kW (12.5% of
the rated power). As seen in this figure, at this power level, the
resonating time is longer than the power transfer time. Fig. 31
depicts the PV current and voltage at almost 20% of the rated
power. The ac-side current and voltage when the inverter is
operating at this power level are shown in Fig. 32. Similar to the
full-power operation, the ac-side current contains 5th, 7th, 11th,
and 13th harmonics, which are generated by the low-frequency
Fig. 33.
CEC efficiency curve of the proposed inverter.
components of the grid voltage. In the case of low power, the
fundamental component of the current is smaller as compared
with that in the case of full power; however, the low-frequency
harmonics are the same in both cases. Hence, the total harmonic
distortion is higher in the case of low power.
Fig. 33 represents the measured California Energy Commission (CEC) efficiency of this inverter. As shown in this figure,
the efficiency of the proposed inverter varies between 94% and
96.5% throughout the whole power range.
V. C ONCLUSION
In this paper, a reliable and compact PV inverter has been
proposed. This inverter is a partial resonant ac-link converter
in which the link is formed by an LC pair having alternating
AMIRABADI et al.: HIGH-FREQUENCY AC-LINK PV INVERTER
current and voltage. The proposed converter guarantees the
isolation of the input and output. However, if galvanic isolation
is required, the link inductance can be replaced by a singlephase high-frequency transformer. The elimination of the dc
link and low-frequency transformer makes the proposed inverter more compact and reliable compared with other types
of PV inverters. In this paper, the principle of operation of the
proposed converter along with the detailed design procedure
has been presented. The performance of the proposed converter
has been evaluated through both simulation and experimental
results.
R EFERENCES
[1] S. Chakraborty, B. Kramer, and B. Kroposki, “A review of power
electronics interfaces for distributed energy systems towards achieving
low-cost modular design,” Renew. Sustain. Energy Rev., vol. 13, no. 9,
pp. 2323–2335, Dec. 2009.
[2] Y. Huang, F. Z. Peng, J. Wang, and D. W. Yoo, “Survey of the power
conditioning system for PV power generation,” in Proc. IEEE PESC,
Jun. 18–22, 2006, pp. 1–6.
[3] S. Atcitty, J. E. Granata, M. A. Quinta, and C. A. Tasca, Utility-scale gridtied PV inverter reliability workshop summary report, Sandia National
Labs., Albuquerque, NM, USA, SANDIA Rep. SAND2011-4778. [Online]. Available: http://energy.sandia.gov/wp/wp-content/gallery/uploads/
Inverter_Workshop_FINAL_072811.pdf
[4] Y. C. Qin, N. Mohan, R. West, and R. Bonn, Status and needs of
power electronics for photovoltaic inverters, Sandia National Labs.,
Albuquerque, NM, USA, SANDIA Rep. SAND2002-1535. [Online].
Available: www.prod.sandia.gov/techlib/access-control.cgi/2002/021535.
pdf
[5] T. Kerekes, R. Teodorescu, P. Rodríguez, G. Vázquez, and E. Aldabas, “A
new high-efficiency single-phase transformerless PV inverter topology,”
IEEE Trans. Ind. Electron., vol. 58, no. 1, pp. 184–191, Jan. 2011.
[6] G. Grandi, C. Rossi, D. Ostojic, and D. Casadei, “A new multilevel conversion structure for grid-connected PV applications,” IEEE Trans. Ind.
Electron., vol. 56, no. 11, pp. 4416–4426, Nov. 2009.
[7] R. Margolis, A review of PV inverter technology cost and performance
projections, Nat. Renew. Energy Lab., Golden, CO, USA, NREL/SR-62038771. [Online]. Available: http://www.nrel.gov/pv/pdfs/38771.pdf
[8] S. J. Castillo, R. S. Balog, and P. Enjeti, “Predicting capacitor reliability
in a module-integrated photovoltaic inverter using stress factors from an
environmental usage model,” in Proc. NAPS, 2010, pp. 1–6.
[9] S. Jain and V. Agarwal, “An integrated hybrid power supply for distributed
generation applications fed by nonconventional energy sources,” IEEE
Tran. Energy Convers., vol. 23, no. 2, pp. 622–631, Jun. 2008.
[10] C. Rodriguez and G. A. J. Amaratunga, “Long-lifetime power inverter
for photovoltaic AC modules,” IEEE Trans. Ind. Electron., vol. 55, no. 7,
pp. 2593–2601, Jul. 2008.
[11] A. K. S. Bhat and S. B. Dewan, “A novel utility interfaced high-frequency
link photovoltaic power conditioning system,” IEEE Trans. Ind. Electron.,
vol. 35, no. 1, pp. 153–159, Feb. 1988.
[12] A. K. S. Bhat and S. B. Dewan, “Resonant inverters for photovoltaic array
to utility interface,” IEEE Trans. Aerosp. Electron. Syst., vol. 24, no. 4,
pp. 377–386, Jul. 1988.
[13] M. Amirabadi, A. Balakrishnan, H. A. Toliyat, and W. Alexander, “Soft
switched AC-link direct-connect photovoltaic inverter,” in Proc. IEEE Int.
Conf. Sustain. Energy Technol., 2008, pp. 116–120.
[14] A. Balakrishnan, H. A. Toliyat, and W. C. Alexander, “Soft switched AC
link buck boost converter,” in Proc. IEEE APEC, 2008, pp. 1334–1339.
[15] A. Balakrishnan, M. Amirabadi, H. Toliyat, and W. Alexander, “Soft
switched AC-link wind power converter,” in Proc. IEEE ICSET, 2008,
pp. 318–321.
[16] M. Amirabadi, H. A. Toliyat, and W. C. Alexander, “Battery-utility interface using soft switched AC link buck–boost converter,” in Proc. IEMDC,
2009, pp. 1299–1304.
[17] M. Amirabadi, H. A. Toliyat, and W. C. Alexander, “Battery-utility interface using soft switched AC link supporting low voltage ride through,” in
Proc. ECCE, 2009, pp. 2606–2613.
[18] A. K. Balakrishnan, “Soft switched high frequency AC-link converter,”
M.S. thesis, Elect. Comput. Eng. Dept., Texas A&M Univ., College
Station, TX, USA, 2008.
291
Mahshid Amirabadi (S’05) received the B.S. degree in electrical engineering from Shahid Beheshti
University, Tehran, Iran, in 2002 and the M.S. degree in electrical engineering from the University of
Tehran, Tehran, in 2006. She is currently working
toward the Ph.D. degree at Texas A&M University,
College Station, TX, USA.
Her main research interests and experience include the design, analysis, and control of power
converters, renewable energy systems, and variablespeed drives.
Anand Balakrishnan received the Bachelor of Engineering degree in electrical and electronics engineering from the College of Engineering, Anna
University, Tamil Nadu, India, in 2006 and the M.Sc.
degree in electrical engineering from Texas A&M
University, College Station, TX, USA, in December
2008.
He is presently a Systems and Applications Engineer with Freescale Semiconductor, Inc., Tempe,
AZ, USA. His interests include renewable energy
systems, power electronics, and embedded systems.
Hamid A. Toliyat (S’87–M’91–SM’96–F’08) received the B.S. degree in electrical engineering from
Sharif University of Technology, Tehran, Iran, in
1982, the M.S. degree in electrical engineering from
West Virginia University, Morgantown, WV, USA, in
1986, and the Ph.D. degree in electrical engineering
from the University of Wisconsin, Madison, WI,
USA, in 1991.
He is currently Raytheon Endowed Professor of
Electrical Engineering with Texas A&M University,
College Station, TX, USA. He has supervised more
than 50 graduate students and published over 385 technical papers. He is
the holder of 12 issued and pending U.S. patents. He is the author of DSPBased Electromechanical Motion Control (CRC Press, 2003), the coeditor of
Handbook of Electric Motors—2nd Edition (Marcel Dekker, 2004), and the
coauthor of Electric Machines—Modeling, Condition Monitoring, and Fault
Diagnosis (CRC Press, 2013).
Dr. Toliyat is the recipient of the Space Act Award from NASA and the Texas
A&M Select Young Investigator Award in 1999, the Eugene Webb Faculty
Fellow Award in 2000, the Schlumberger Foundation Technical Awards in 2000
and 2001, the E. D. Brockett Professorship Award in 2002, the Distinguished
Teaching Award in 2003, the prestigious Cyrill Veinott Award in Electromechanical Energy Conversion from the IEEE Power Engineering Society in 2004,
the Texas Engineering Experiment Station Faculty Fellow Award in 2006, and
the Patent and Innovation Award from Texas A&M University System Office
of Technology Commercialization in 2007. He is also the recipient of the IEEE
Power Engineering Society Prize Paper Awards in 1996 and 2006, the 2006
IEEE Industry Applications Society Transactions Third Prize Paper Award, and
the 2008 Industrial Electronics Society Electric Machines Committee Second
Best Paper Award.
William C. Alexander (M’95) received the B.S.
and M.S. degrees in engineering from the University of Texas, Austin, TX, USA, in 1976 and 1979,
respectively.
He is the Chief Technology Officer and Founder
of Ideal Power Converters, Inc., Austin, a company
formed to develop the soft-switched ac-link power
converter for motor-drive, wind-power, and solarinverter applications. He is a prolific inventor with
over 15 granted U.S. patents. He has over 25 years
of experience in engineering including mechanical,
electrical, electronic, and embedded-software and hardware description language engineering.
Mr. Alexander is a Registered Professional Engineer in the State of Texas.