2. 898 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 26, NO. 3, MARCH 2011
tipartial cloud capability, and the fact that any single failure of
an MIC will not impact any other part of the system. As a result,
MICs in a parallel configuration have higher fault tolerance and
reliability that make them more promising for PV application in
a FREEDM system. However, the high gain requirement usually
compromises its efficiency.
The topologies suitable for this application can be categorized
into two groups: nonisolated topologies and isolated topologies.
For nonisolated topologies, boost, buck–boost, zeta, cuk, or their
Fig. 2. Two types of dc MIC structure: (a) parallel connection and (b) series derivatives [23]–[32] are commonly used. Isolated topologies
connection. mainly include flyback [33]–[39], current-fed push–pull [40],
[41], and resonant converters [42], [43]. The typical maximum
Therefore, it is very possible to reduce the system cost for end efficiency of these converters is around 80–97% [10]–[12], [19].
users. At present, significant research effort has been made to Among these topologies, the half-bridge LLC resonant con-
improve the performance of PV converters [7]–[9]; PV module- verter is a good candidate due to its several unique advan-
integrated converters (MICs) are gaining increasing amounts of tages [44]–[46]. However, it is difficult for an LLC resonant
attention due to their distinctive features [10]–[20]. converter to maintain high efficiency for a wide input range un-
1) The MIC is an integrated part of the PV panel. MICs re- der different load conditions. In this paper, a new resonant dc/dc
move losses due to the mismatch between panels and sup- converter with dual operation modes is proposed. By chang-
port panel level maximum power point tracking (MPPT). ing operation modes adaptively according to VPV and PPV , the
For a string inverter or a centralized inverter, a string or converter’s efficiency is improved.
multistring of PV panels shares a single MPPT controller,
but the mismatch loss is serious in partial shading condi-
tions [21]. Considering the mismatch loss together with III. OPERATION PRINCIPLE OF THE NEW
the dc/ac conversion loss contributing to the whole PV RESONANT CONVERTER
system loss, string/centralized inverters may have lower
system efficiency than MICs due to higher mismatch loss Fig. 3 shows a circuit diagram of the proposed resonant con-
although they usually have higher dc/ac conversion effi- verter. S1 and S2 are two power MOSFETs; DS 1 , CS 1 and
ciency than MICs. DS 2 , CS 2 are the body diodes and parasitic capacitances of S1
2) Panel level hot-spot risk is removed [11] and panel life- and S2 , respectively. Cr is the resonant capacitor; Lr and Lm
time can be improved. Hot spot takes place when a shaded are the magnetizing inductance of transformers Tx2 and Tx1 ,
cell within a partially shaded panel becomes reverse bi- respectively. Llkg is the sum of the leakage inductance of Tx1
ased and dissipates power in the form of heat [22]. For and Tx2 . D1 , D2 and Co1 , Co2 form a voltage doubler at the
series connected PV panels used with a string/centralized secondary side of Tx1 . A half-wave rectifier (HWR) formed by
inverter, a by-pass diode is added to each panel in practice. D3 , S3 , D4 , and CO 3 is added to the secondary side of trans-
For the MIC solution, the by-pass diode is not necessary former Tx2 . Diode D3 blocks the conductive path of the body
because each panel has its own MIC, leading to no direct diode of S3 . Thus, D3 and S3 form a unidirectional switch to en-
connection between PV panels. able or disable the HWR. When the HWR is enabled, the HWR
3) Its “plug and play” feature simplifies system installation. and voltage doubler will support the 400-V dc bus with their
In summary, the MIC solution allows for more flexible PV summed outputs. Table II summarizes the operation modes for
project planning and multifacet PV panel installation. the proposed converter and Vth is a predefined threshold volt-
age that is usually equal to the nominal voltage Vnom . For the
first three operation conditions listed in Table II, the HWR is
II. COMPARISON OF MICS IN SERIES AND
disabled by turning off switch S3 . As a result, the converter
PARALLEL CONNECTIONS behaves like a traditional LLC resonant converter with a voltage
Both dc MICs and ac MICs are available in the market. Only doubler [46]: an equivalent resonant inductor Lr , comprised of
dc MICs will be discussed in this paper, as they are suitable for Lr and Llkg , participates in the resonant circuit formed by Lm
the FREEDM system. As shown in Fig. 2, dc MICs have two and Cr . Diode D4 is conducting to provide a path for the load
kinds of connection structures. Fig. 2(a) shows a type I dc MIC current. Once VPV is smaller than Vth and PPV is lower than
configuration, consisting of multiple parallel connected MICs 50% of the rated power (Prated ), the PV panel is working under
directly interfaced with a dc bus. Type II dc MICs, shown in condition #4 and the converter will operate in Mode II.
Fig. 2(b), need to form a series connection to obtain a voltage For one switching period, the operation of the converter in
high enough for interfacing with the dc bus. Generally, the power Mode II can be divided into nine stages. The equivalent circuit
rating of both types of dc MICs is around 200 W–300 W. for each stage is shown in Fig. 4 and its key waveforms are
The two system structures have different features. Table I depicted in Fig. 5. For the description of circuit operation (and
summarizes the comparison results of the two MIC structures: for the subsequent dc gain derivation in the next section), the
the parallel connection is more flexible due to its stronger an- following assumptions are made.
3. LIANG et al.: HIGH-EFFICIENCY PV MODULE-INTEGRATED DC/DC CONVERTER FOR PV ENERGY HARVEST IN FREEDM SYSTEMS 899
TABLE I
COMPARISON OF TWO TYPES OF DC MIC STRUCTURE
Fig. 3. Circuit diagram of the proposed resonant converter.
TABLE II 3) The turn ratio NT X 2 (Npri : Nsec ) of transformer TX 2 is
SUMMARY OF OPERATION MODES FOR THE PROPOSED RESONANT CONVERTER
the half of NT X 1 . Define NT X 2 = 1/2 NT X 1 = N .
The operation processes of Mode II are specified as follows.
Stage 1 (t0 –t1 ): When S2 is turned off at t = t0 , stage 1 be-
gins. Since Ipri is negative, capacitor Cs2 (Cs1 ) will be charged
(discharged) and the switching node voltage Vsw will increase
accordingly. Inductors Lm , Lr , and Llkg are all in resonance
with Cr . Vcr continues to decrease and no current flows through
the secondary side of either transformer. The output capacitors
1) All the components are ideal. The body diodes and par- Co1 , Co2 together with Co3 supply the load current and VC o1 –
asitic capacitance of S1 and S2 have been taken into ac- VC o3 all decrease in this period.
count. The output capacitors have equal values (Co1 = Stage 2 (t1 –t2 ): At time t = t1 , Vsw reaches Vpv . Ds1 is
Co2 = Co3 ). forward biased and starts to conduct a current Ipri . Ipri starts
2) Inductor Llkg includes the leakage inductance of TX 1 and to decrease. Once Ipri becomes smaller than the magnetizing
TX 2 ; it also includes the wire parasitic inductance. currents IL r and IL m , the resonance of [Lm , Lr , Llkg ] and Cr
4. 900 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 26, NO. 3, MARCH 2011
Fig. 4. Equivalent circuits for each operation stage (Mode II operation).
is stopped. Lr and Lm will be out of the resonance following changes its direction at t = t3 . The leakage inductor Llkg still
this. The difference between Ipri and IL m will flow in the sec- resonates with Cr , and Ipri keeps increasing. The magnetizing
ondary side of Tx1 . Similarly, the secondary side of Tx2 will currents IL r and IL m continue to increase with the same slope
conduct the current difference between Ipri and IL r . Thus, the as in Mode 2. The rectifier diodes D1 and D3 conduct current
voltage across the primary side of Tx1 and Tx2 is clamped by and power is delivered to the load. This stage ends when Ipri is
Vout . IL r and IL m start to decrease linearly. equal to IL m .
Stage 3 (t2 –t4 ): This stage begins when S1 is turned on at t = Stage 4 (t4 –t5 ): At t = t4 , Ipri and IL m are equal. The output
t2 . At this moment, the primary-side current Ipri is negative and current of the transformer Tx1 reaches zero. Transformer Tx1 ’s
flows through the body diode of S1 . Thus, ZVS turn on of S1 secondary voltage is lower than the output voltage. The output
can be achieved at t2 . The current Ipri continues to decrease and is separated from transformer Tx1 . Meanwhile, since Ipri is still
5. LIANG et al.: HIGH-EFFICIENCY PV MODULE-INTEGRATED DC/DC CONVERTER FOR PV ENERGY HARVEST IN FREEDM SYSTEMS 901
balance of the transformers Tx1 and Tx2 has still been preserved.
Further, if a full-wave rectifier (FWR) is added instead of the
HWR, Ipri will become symmetrical and the other character-
istics of the converter will remain. The theoretical analysis of
the aforementioned Mode II operation has been verified by the
simulation with Simetrix. Fig. 6 shows the simulation results
of the proposed converter with following operation conditions:
Vpv = 22 V, Vout = 400 V, Pout = 120 W (50% of Prated ), fs =
83 kHz.
IV. DC GAIN ANALYSIS FOR THE PROPOSED CONVERTER
OPERATION IN MODE II
Understanding of the dc gain characteristic for a resonant con-
verter has equal importance as knowing its operation principle.
Since the dc gain characteristic for Mode I operation is the same
as LLC resonant converter, only Mode II operation requires a
new analysis to be developed. The fundamental harmonic analy-
sis (FHA) method is widely used for dc gain analysis of resonant
converters [47]–[50] and it is also valid for the analysis devel-
Fig. 5. Key waveforms of the proposed converter (Mode II operation).
oped in this paper. This approach is based on the assumption
that the power transfer from the source to the load through the
resonant tank is almost completely dependent on the fundamen-
larger than IL r , the output current of Tx2 is not zero and power tal harmonic of the Fourier expansion of the currents and the
is delivered to the load through Tx2 . During this stage, Lm voltage involved. The voltage at the input of the two rectifiers
participates into the resonance again and the resonance between Vosq (t) can be expressed as
[Llkg , Lm ] and Cr begins.
Stage 5 (t5 –t6 ): Switch S1 is turned off at t = t5 . The current Vosq (t) = Vab (t) + Vcd (t) (1)
Ipri is positive and switching node voltage will decrease due to where Vab (t) and Vcd (t) are the secondary-side terminal volt-
charging (discharging) of Cs1 (Cs2 ). ages of transformers TX 2 and TX 1 (see Fig. 3). Like the con-
Stage 6 (t6 –t7 ): At time t = t6 , Vsw drops to zero that causes ventional LLC resonant converter, the current in the secondary
the conduction of the body diode Ds2 . With the drop of Vsw , the side is quasi-sinusoidal and the voltage Vosq (t) reverses when
voltage applied to Lm (VL m ) decreases to zero and continues to the current becomes zero. Therefore, Vosq (t) is an alternative
become more negative. Once VL m is higher than a certain level, square wave in phase with the rectifier current. The Fourier
diode D2 on the secondary side of Tx1 will be forward biased. expression of Vosq (t) is
Thus, the voltage applied to Lm is clamped and IL m will drop
linearly. Lm is out of resonance with Cr . Instead, only Llkg 4 1
Vosq (t) = Vout sin(n2πfsw t). (2)
resonates with Cr and Ipri decreases steeply. This stage ends π n =1,3,5,...
n
when IL r is equal to Ipri .
For convenience, the phase angle of Vosq (t) is assumed to be
Stage 7 (t7 –t8 ): At time t = t7 , IL r is equal to Ipri ; no
zero in (2). Its fundamental component Vo FHA (t) is
more current will flow in the secondary side of Tx2 . The output
is separated from Tx2 . D3 is turned off with ZCS. The voltage 4
Vo FHA (t) =Vout sin(2πfsw t). (3)
applied to Lr is not clamped and Lr participates in the resonance π
again with Cr and Llkg . The current Ipri is positive and continues The rms amplitude of Vo FHA (t) is
to flow through Ds2 , which creates the ZVS condition for S2 if √
2 2
S2 is turned on at this moment. Vo FHA = Vout . (4)
Stage 8 (t8 –t10 ): At t = t8 , S2 is turned on with ZVS. The π
current Ipri continues to decrease due to the resonance between Define the fundamental part of the rectifier current to be
√
[Lr , Llkg ] and Cr . The transformer Tx1 delivers power to the irect (t) = 2Irect sin(2πfsw t). (5)
output. This stage ends when current Ipri = IL m .
Stage 9 (t10 –t11 ): At t = t9 , Ipri = IL m . No more current will The phase angle of Irect is also zero since it is in phase with
flow in the secondary side of Tx1 . The voltage applied to Lm Vo FHA (t). Thus, the average value of Iout can be calculated as
is not clamped anymore and Lm participates in the resonance TSW √
2 2 2 2Irect
again with Lr , Llkg , and Cr . At t = t11 , S2 is turned off and a Iout = irect (t)dt = . (6)
TSW 0 π
new switching cycle begins.
From the aforementioned analysis, the energy transferred by Iout can be expressed as
Tx1 and Tx2 is different. The positive and negative parts of the Vout
current Ipri are not symmetrical. However, the voltage-second Iout = . (7)
Rout
6. 902 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 26, NO. 3, MARCH 2011
Fig. 6. Simulation results of the proposed converter operating in Mode II.
Fig. 7. Equivalent FHA resonant circuit model for the proposed converter operation in Mode II.
Equation (8) can be derived by combining (6) and (7) as substituted by an equivalent transformer Txe with turn ratio N .
follows: The resulting expression for the dc gain of the converter can be
√ derived through a circuit analysis based on the model in Fig. 7.
2πVout
Irect = . (8) Define the dc gain
4Rout
Insert (8) into (5) N Vo FHA
M= . (11)
√ Vi FHA
√ 2πVout πVout
irect (t) = 2 sin(2πfsw t) = sin(2πfsw t).
4Rout 2Rout Consider
(9)
Vdc 2 1
The equivalent ac output impedance Ro ac can be derived by VSW (t) = + Vdc sin(n2πfsw t). (12)
combining (4) and (8) as follows: 2 π n =1,3,5,...
n
Vo FHA 8Rout
Ro ac = = . (10) vi FHA (t) is the fundamental part of VSW (t)
Irect π2
The expression for Ro ac is the same as the one for a conven- 2
vi FHA (t) = Vdc sin(2πfsw t). (13)
tional LLC resonant converter. With the known Ro ac , the equiv- π
alent FHA resonant circuit model can be obtained, as shown in
Vi FHA can be derived as follows:
Fig. 7.
In this model, Vi FHA is the rms value of the fundamental √
2
component of the voltage at the switching node SW (VSW ). The Vi FHA (t) = Vdc . (14)
voltage VSW is generated by the controlled switches S1 and S2 . π
The output current Iout is produced from Irect after the rectifier Combining with (4), (11), and (14), the input-to-output volt-
network and filter capacitors. From a turn ratio perspective, the age conversion ratio is
conversion gain of a transformer with turn ratio 2N followed
by a voltage doubler is equal to a transformer with turn ratio N . Vout 1
Therefore, transformer Tx1 together with voltage doubler can be = |M | . (15)
Vdc 2N
7. LIANG et al.: HIGH-EFFICIENCY PV MODULE-INTEGRATED DC/DC CONVERTER FOR PV ENERGY HARVEST IN FREEDM SYSTEMS 903
From the FHA model, Zout is the impedance seen from the
primary side of the two transformers
N 2 Ro ac · Lm r · S
Zout = (16)
N 2 Ro ac + Lm r · S
where Lm r = Lm + Lr . The dc gain M can be derived as
follows:
Zout
M (S) = . (17)
(1/S · Cr ) + S · Llkg + Zout
By substituting S = j2πfSW , the amplitude of M (S) is, as
shown (18), at the bottom of this page.
For convenience, (18) can be rewritten as
1
M (fn ) = . (19)
(1 + λ − (λ/fn 2 ))2 + Q2 (f − (1/f ))2
n n
Fig. 8. Series of example of dc gain curves of a new resonant converter with
The parameters in (19) are defined as follows: different Q value (Mode II).
1
fr = (20)
2π Llkg · Cr
Z0
Q= (21)
N2 · Ro ac
Llkg
λ= (22)
Lm + Lr
Llkg
Z0 = (23)
Cr
f SW
fn = . (24)
fr
Equations (19)–(24) reveal the dc gain characteristics for
Mode II operation. It is interesting that Mode II operation has Fig. 9. Series of example of dc gain curves for a new resonant converter with
similar dc gain expression to Mode I but with different parame- different Q value (Mode I).
ters for the resonant tank. A series of example of dc gain curves
of Mode II operation under different load conditions (with differ- V. DC GAIN VERIFICATION AND COMPARISON
ent Q values) are plotted in Fig. 8. For very light load conditions
(small Q), the gain has a large peak. On the contrary, the gain To verify the dc gain expression derived in section IV, a
becomes flat under heavy load conditions (large Q). Similar to series of simulations have been performed for different Vpv for
an LLC converter, the dc characteristic of Mode II operation a given load condition. The converter’s switching frequency fs
can be divided into ZVS and ZCS regions, and the converter is recorded. Equation (19) is used to calculate the dc gain result
should be prevented from entering the ZCS region. With proper at a given fs for the same operation condition. Through the
choice of the resonant tank, Mode II operation can stay in the comparison between the dc gain from simulation (Msimulation )
ZVS region for Vpv and Ppv variations. The ZVS region can be and the theoretical analysis result (Mcalculation ), the accuracy
further divided into regions I and II due to slightly operation of (19) can be evaluated. Table III shows the comparison results
differences. In practical designs, the converter has unity gain at for a 50% load condition where Msimulation is defined by
Vpv = Vnom and the converter enters Mode II operation only Vout · N
Msimulation = . (25)
when Vpv ≤ Vnom . Therefore, it is impossible for the proposed Vpv /2
resonant converter to work in region I after entering Mode II
operation. Mode II operation can only be active in region II. From Table III, Mcalculation matches with Msimulation very
Furthermore, the discussion about Mode II operation in the last well. Therefore, (19) is accurate enough for engineering design
section is dedicated for region II. On the contrary, Mode I op- of the proposed converter. Furthermore, a comparison of the dc
eration can only be active in region I (see Fig. 9) because the gain between Mode I and II operations is conducted in order to
required dc gain should be lower than 1 in Mode I (Vpv > Vnom ). reveal the general dc gain features of the proposed converter.
32π 2 ·Cr ·Lm r ·Rout ·fSW ·N 2
2
M= .
(32π 2 ·Cr ·Lm r ·Rout ·fSW ·N 2 − 8·Rout ·N 2 + 32·π 2 ·Cr ·Llkg ·Rout ·fSW ·N 2 )2 + (−2·π 3 ·Lm r ·fSW + 8·π 5 ·Cr ·Lm r ·Llkg ·fSW )2
2 2 3
(18)
8. 904 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 26, NO. 3, MARCH 2011
TABLE III TABLE IV
DC GAIN COMPARISON BETWEEN SIMULATION AND CALCULATION LIST OF PARAMETERS OF THE PROPOSED CONVERTER FOR GAIN ANALYSIS
The normalized frequency fn has a different base for Mode I
and II operations since they have different fr :
fSW
fn M o deI = ,
fr M o deI
1
where fr M o deI = (26)
2π (Llkg + Lr ) · Cr
fSW 1
fn M o deII = , where fr M o deII = .
fr M o deII 2π Llkg · Cr
(27)
For further analysis, fn needs to be unified using the same
base, for fn M o deI :
fSW fr M o deII
fn M o deI = ·
fr M o deII fr M o deI
fr M o deII
= fn M o deII · = α · fn M o deII . (28)
fr M o deI
Both the dc gain expressions for Modes I and II can be written
as functions of fn M o deII , as shown (29) and (30), at the bottom
of this page.
Table IV gives the resonant tank parameters for example de-
sign. For comparison, the equations for calculating several key
parameters are also listed in Table IV. The gain curves for the
two operation modes can be plotted in the same figure, as shown
in Fig. 10. Fig. 10. DC gain comparison between Modes I and II at 50% rated power.
From Fig. 10, the two curves reach their peaks at the same
frequency fn M defined by
fM 1 2) The frequency difference becomes larger with higher input
fn M = = . voltage. Fig. 10 takes Vpv = 22 V and Vpv = 32 V as ex-
fn M o deII 2π (Lr + Llkg + Lm ) · Cr · fn M o deII
(31) amples. It shows the switching frequency almost doubles
Similar to the LLC resonant converter, operation in the region if the converter operates in Mode II with 32-V input.
where fn < fn M is forbidden. In the region fn M < fn < 3) The gain curve of Mode II becomes much flatter at high
f0 , MM o deI is always higher than MM o deII . On the contrary, frequency. The gain is almost constant and stops decreas-
MM o deI becomes lower than MM o deII in region fn > f0 . For ing. Considering that higher Vm pp requires smaller dc
a desired dc gain in the latter region, the following conclusion gain, this implies that the PV panel voltage may be out of
can be drawn. regulation in Mode II when Vm pp is too high. Therefore,
1) Mode II operation needs a higher switching frequency it is reasonable to keep the converter operating in Mode I
than Mode I operation. when Vm pp is higher than a certain value.
1
MM o deI (fn M o deII ) = (29)
(1 + λM o deI − (λM o deI /(α · fn M odeII
)2 ))2 + Q2 o deI (α · fn
M M o deII − (1/α · fn M odeII
))2
1
MM o deII (fn M o deII ) = . (30)
(1 + λM o deII − 2
(λM o deII /fn M odeII
))2 + Q2 o deII (fn
M M odeII
− (1/fn M odeII
))2
9. LIANG et al.: HIGH-EFFICIENCY PV MODULE-INTEGRATED DC/DC CONVERTER FOR PV ENERGY HARVEST IN FREEDM SYSTEMS 905
TABLE V TABLE VI
CIRCUIT PARAMETERS FOR EXPERIMENT LOSS BREAKDOWN OF THE PROPOSED CONVERTER IN MODE II WITH 10% OF
P ra te d (V pv ≤ 32 V)
TABLE VII
LOSS BREAKDOWN OF THE LLC CONVERTER WITH 10% OF P ra te d
(V pv ≤ 32 V)
Fig. 11. Efficiency improvement of the proposed converter in Mode II
operation.
VI. DESIGN EXAMPLE AND EFFICIENCY ANALYSIS
The MIC will be operated with PV panels that normally have
Vm pp of around 22–40 V. Vnom for this design is 32 V and
Prated is equal to 240 W. The transformer primary side is the
low-voltage side and it has high resonant current circulating.
In order to minimize the conduction loss, a 75-V MOSFET Fig. 12. System diagram for the experiment with a work flow chart for the
with low Rdson is preferred and multistrand Litz wire should be dc/dc controller.
used to reduce the ac resistance of the primary winding of the
transformer. There is no strict limitation on volume and size for
MICs. Thus, a lower switching frequency fs (<200 kHz) can
be adopted to benefit the converter efficiency.
Table V gives component parameters for the MIC prototype.
The threshold voltage Vth for operation mode decision is chosen
to be equal to Vnom . One can design Cr , Lr , Lm , and Tx1
with a conventional design procedure for an LLC converter.
Then, a secondary winding is added to Lr such that it forms
the transformer Tx2 . The devices D3 , D4 , and S3 in HWR
have the same current rating as D1 and D2 in voltage doubler.
Considering that a practical transformer has a certain leakage
inductance, the value of Llkg can be chosen to be 5–15% of
(Lr + Lm ).
A comprehensive loss analysis has been conducted to eval-
uate the efficiency of the designed converter. For comparison,
the efficiency of a traditional LLC resonant converter with the
same circuit parameters is also analyzed. Their efficiency dif- Fig. 13. Picture of a 240-W MIC prototype.
ference is plotted in Fig. 11 for 5–50% of Prated . The efficiency
10. 906 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 26, NO. 3, MARCH 2011
Fig. 14. Waveforms of an MIC prototype: (a) Mode I (ch1: 10 V/div; ch4: 10 A/div; t = 4 μs) and (b) Mode II (ch1: 50 V/div; ch2: 200 V/div; ch3: 1 A/div;
ch4: 10 A/div).
Fig. 15. Waveforms to verify the ZVS operation in Mode II (ch1: 10 V/div; ch2: 20 V/div; ch4: 10 A/div). (a) V in = 22 V, 20% of P ra te d (verify upper side
switch ZVS) and (b) V in = 22 V, 20% of P ra te d (verify lower side switch ZVS).
improvement drops when Ppv increases. When Ppv approaches operation reduces the transformer core loss by causing smaller
50% of Prated , the efficiency improvement is reduced to almost variation of the magnetic field strength in a switching period.
zero. Therefore, there is no benefit to keep converter running As a result, the total loss is dramatically reduced by Mode II
in Mode II when Ppv > 50% of Prated and mode change is operation.
required.
To get a better understanding of the efficiency improvement
in Mode II operation, a loss breakdown is conducted for both VII. EXPERIMENTAL RESULTS
Mode II operation and normal LLC operation with Vpv < 32 V An experimental prototype has been built to verify the per-
and Ppv = 10% of Prated . Tables VI and VII give the analysis formance of the proposed converter. Fig. 12 depicts the system
results. As discussed in the previous section, Mode II operation diagram for experiment and Fig. 13 shows a picture of the pro-
will increase the switching frequency. Thus, the switching loss totype. An MPPT controller implemented in a microcontroller
of MOSFET may increase due to the increase in the number of will provide a reference voltage Vpv ref that will be used by the
switching events. However, the data in Table VI show a signifi- dc/dc controller to determine the converter’s operation mode
cant decrease in the total switching loss. This is because higher based on the criteria described in Table II. The dc/dc controller
frequency operation leads to a much lower resonant current will check Vpv and Ppv every few minutes and its operation
through the MOSFET during its turn-off event. Due to the same follows the work flow chart in Fig. 12.
reason, the MOSFET conduction loss and transformer copper Fig. 14 shows the operation waveforms of MIC prototype in
loss are also greatly reduced. Moreover, the higher frequency Modes I and II. In Mode II, only the positive part of current
11. LIANG et al.: HIGH-EFFICIENCY PV MODULE-INTEGRATED DC/DC CONVERTER FOR PV ENERGY HARVEST IN FREEDM SYSTEMS 907
converter’s performance have been validated by the experiment
results from a 240-W prototype. Future work includes the com-
pletion of an advanced energy controller design for the MIC that
can receive commands from the IEM and allows for a flexible
control of the power generation profile.
ACKNOWLEDGMENT
The authors would like to thank Edward Van Brunt’s help
during the manuscript revision. This work made use of ERC
shared facilities supported by the National Science Foundation
Fig. 16. Measured efficiency improvements with HWR (Mode II) for 5–50% under Award Number EEC-0812121.
of P ra te d (V pv ≤ 32 V).
REFERENCES
[1] B. K. Bose, “Global warming: Energy, environmental pollution, and the
impact of power electronics,” IEEE Ind. Electron. Mag., vol. 4, no. 1,
pp. 6–17, Mar. 2010.
[2] European Renewable Energy Council (2004, May). Renewable Energy
Scenario to 2040 [Online]. Available: http://www.erec-renewables.org/
documents/targets_2040/EREC_Scenario%202040.pdf
[3] Q. Li and P. Wolfs, “A review of the single phase photovoltaic module in-
tegrated converter topologies with three different DC link configurations,”
IEEE Trans. Power Electron., vol. 23, no. 3, pp. 1320–1333, May 2008.
[4] IMS Research. (2010, Apr. 15). [Online]. Available: http://www.pv-tech.
org/lib/printable/8848/
[5] A. Q. Huang, “Renewable energy system research and education at the
NSF FREEDM systems center,” in Proc. IEEE Power Energy Soc. (PES)
Gen. Meeting, Calgary, AB, Canada, 2009, pp. 1–6.
[6] H. Laaksonen, “Protection principles for future microgrids,” IEEE Trans.
Fig. 17. Efficiency measurement results for the designed MIC prototype. Power Electron., vol. 25, no. 12, pp. 2910–2918, Dec. 2010.
[7] Y. Fang and X. Ma, “A novel PV micro-inverter with coupled inductors
and double boost topology,” IEEE Trans. Power Electron., vol. PP, no. 99,
Isec TX1 will flow through the HWR. Since Isec TX1 returns to p. 1, 2010.
zero before each half cycle ends, ZCS turn off of diodes D1 –D3 [8] B. Yang, W. Li, Y. Zhao, and X. He, “Design and analysis of a grid-
connected photovoltaic power system,” IEEE Trans. Power Electron.,
is realized. Fig. 15 verifies the ZVS feature of the proposed vol. 25, no. 4, pp. 992–1000, Apr. 2010.
converter for Mode II operation. It clearly shows that ZVS turn [9] T. Kerekes, M. Liserre, R. Teodorescu, C. Klumpner, and M. Sumner,
on is achieved for both the high-side and the low-side MOSFETs “Evaluation of three-phase transformerless photovoltaic inverter topolo-
gies,” IEEE Trans. Power Electron., vol. 24, no. 9, pp. 2202–2211, Sep.
in Mode II. 2009.
The efficiency of the proposed converter under different Vpv [10] S. B. Kjaer, J. K. Pedersen, and F. Blaabjerg, “A review of single-phase
and Ppv is measured in the experiment. For comparison, the effi- grid-connected inverters for photovoltaic modules,” IEEE Trans. Ind.
Appl., vol. 41, no. 5, pp. 1292–1306, Sep./Oct. 2005.
ciency of operation without the HWR (normal LLC operation) is [11] S. B. Kjaer, J. K. Pedersen, and F. Blaabjerg, “Power inverter topologies
also recorded. Fig. 16 shows the efficiency difference between for photovoltaic modules—a review,” in Proc. IEEE Ind. Appl. Soc. (IAS)
Mode II operation and normal LLC operation. From Fig. 16, Annu. Meeting, Dec. 2002, vol. 2, pp. 782–788.
[12] J. M. A. Myrzik and M. Calais, “String and module integrated inverters for
the maximum efficiency improvement happens at 5% of Prated single-phase grid connected photovoltaic systems—A review,” presented
for all input conditions. For this condition, over 10% improve- at the IEEE Power Tech. Conf., Bologna, Italy, 2003.
ment is achieved. With an increase of the load, the efficiency [13] Q. Li and P. Worlfs, “A review of the single phase photovoltaic module in-
tegrated converter topologies with three different DC link configurations,”
improvement drops. Fig. 17 gives the complete efficiency data IEEE Trans. Power Electron., vol. 23, no. 3, pp. 1320–1333, May 2008.
for the MIC prototype. A high efficiency of 96.5% occurs in [14] Y. Xue, L. Chang, S. B. Kjaer, J. Bordonau, and T. Shimizu, “Topolo-
Mode II with Vpv = 32 V and Ppv = 50% of Prated . The highest gies of single-phase inverters for small distributed power generators: An
overview,” IEEE Trans. Power Electron., vol. 19, no. 5, pp. 1305–1314,
weighted efficiency is 95.8% in the experiment. Sep. 2004.
[15] W.-Y. Choi, B.-H. Kwon, and J.-S. Lai, “High-efficiency grid-connected
photovoltaic module integrated converter system,” in Proc. IEEE . 35th
VIII. CONCLUSION AND FUTURE WORK Annu. Conf. Ind. Electron. (IECON), Porto, Portugal, 2009.
The PV converters can take advantage of the 400-V dc bus [16] E. Roman, R. Alonso, P. Ibanez, S. Elorduizapatarietxe, and D. Goitia,
“Intelligent PV module for grid-connected PV systems,” IEEE Trans.
in FREEDM systems to reduce its complexity as well as costs Ind. Appl., vol. 53, no. 4, pp. 1066–1073, Jun. 2006.
to the end user. The parallel connected dc MICs are good can- [17] G. R. Walker and J. C. Pierce, “Photovoltaic DC–DC module integrated
didates for this application. In this paper, a high-efficiency dual converter for novel cascaded and bypass grid connection topologies-design
and optimization,” in Proc. IEEE 37th Power Electron. Spec. Conf.
mode resonant converter topology is proposed for dc MICs. The (PESC), Jun. 2006, pp. 1–7.
new resonant converter can change resonant modes adaptively [18] G. R. Walker and P. C. Sernia, “Cascaded DC–DC converter connection
depending on the PV panel operation conditions. A detailed of photovoltaic modules,” IEEE Trans. Power Electron., vol. 19, no. 4,
pp. 1130–1139, Jul. 2004.
theoretical analysis of the converter operation and its dc gain [19] B. Liu, C. Liang, and S. Duan, “Design considerations and topology
features is presented in this paper. The analysis and the new selection for dc-module-based building integrated photovoltaic system,”
12. 908 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 26, NO. 3, MARCH 2011
in Proc. 3rd IEEE Conf. Ind. Electron. Appl. (ICIEA), Jun. 2008, pp. 1066– [43] A. Lohner, T. Meyer, and A. Nagel, “A new panel-integratable inverter
1070. concept for grid-connected photovoltaic systems,” in Proc. IEEE Int.
[20] Solar Edge System Overview. (Jan. 2011). [Online]. Available: Symp. Ind. Electron. (ISIE), vol. 2, Warsaw, Poland, Jun. 1996, pp. 827–
http://www.solaredge.com/files/pdfs/se_system_overview.pdf 831.
[21] R. A. Messenger and J. Ventre, Photovoltaic Systems Engineering, [44] B. Lu et al., “Optimal design methodology for LLC resonant converter,” in
2nd ed. Boca Raton, FL: CRC Press, 2003. Proc. IEEE 21st Appl. Power Electron. Conf. Expo. (APEC-06), pp. 19–23.
[22] W. Herrmann, W. Wiesner, and W. Vaassen, “Hot spot investigation on PV [45] B. Yang, “Topology investigation for front end DC/DC power conversion
modules—New concepts for a test standard and consequences for module for distributed power system,” Ph.D. dissertation, Virginia Polytech. Inst.
design with respect to bypass diodes,” in Proc. 26th IEEE Photovoltaic State Univ., Blacksburg, VA, 2003.
Spec. Conf. (PSC), Sep./Oct. 1997, pp. 1129–1132. [46] Y. Gu, L. Hang, Z. Lu, Z. Qian, and D. Xu, “Voltage doubler application
[23] T. Boutot and L. Chang, “Development of a single-phase inverter for in isolated resonant converters,” in Proc. IEEE 31st Annu. Conf. Ind.
small wind turbines,” in Proc. IEEE Canadian Conf. Electrical Computer Electron. Soc. (IECON), Nov. 2005, pp. 1184–1188.
Engineering (CCECE’98), Waterloo, ON, Canada, May 24–28, pp. 305– [47] R. W. Erickson and D. Maksimovic, Fundamentals of Power Electronics,
308. 2nd ed. Springer Science+Business Media, 2001.
[24] R. O. C´ ceres and I. Barbi, “A boost dc–ac converter: Analysis, design, and
a [48] J. F. Lazar and R. Martinelli, “Steady-state analysis of the LLC series
experimentation,” IEEE Trans. Power Electron., vol. 14, no. 1, pp. 134– resonant converter,” in Proc. IEEE 16th Annu. Conf. Appl. Power Electron.
141, Jan. 1999. Conf. Expo. (APEC), Anaheim, CA, Mar. 2001, pp. 728–735.
[25] S. Saha and V. P. Sundarsingh, “Novel grid-connected photovoltaic in- [49] S. De Simone, C. Adragna, C. Spini, and G. Gattavari, “Design-oriented
verter,” in Proc. Inst. Elect. Eng., Mar. 1996, vol. 143, pp. 219–224. steady-state analysis of LLC resonant converters based on FHA,” in Proc.
[26] F. Kang, C. Kim, S. Park, and H. Park, “Interface circuit for photovoltaic Int. Symp. Power Electron., Electr. Drives, Autom. Motion (SPEEDAM),
system based on buck–boost current–source PWM inverter,” in Proc. IEEE 2006, pp. 16–23.
Ind. Electron. Conf. (IECON), 2002, pp. 3257–3261. [50] Application Notes, “Half-bridge LLC resonant converter design using
[27] K. Chomsuwan, P. Prisuwanna, and V. Monyakul, “Photovoltaic grid con- FSFR-series Fairchild Power Switch (FPS),” Application Note AN-4151
nected inverter using two-switch buck–boost converter,” in Proc. IEEE (2007). [Online]. Available: http://www.fairchildsemi.com/an/AN/AN-
Photovoltaic Spec. Conf., 2002, pp. 1527–1530. 4151.pdf
[28] S. Funabiki, T. Tanaka, and T. Nishi, “A new buck–boost-operation-based [51] T. Esram and P. L. Chapman, “Comparison of photovoltaic array maximum
sinusoidal inverter circuit,” in Proc. IEEE Power Electronics Spec. Conf. power point tracking techniques,” IEEE Trans. Energy Convers., vol. 22,
(PESC), 2002, pp. 1624–1629. no. 2, pp. 439–449, Jun. 2007.
[29] J. Myrzik, “Static converter unit for photovoltaic or single-phase applica-
tions,” German Patent DE 19603823A1, Aug. 14, 1996.
[30] J. M. A. Myrzik, “Power conditioning of low-voltage generators with
transformerless grid connected inverter topologies,” in Proc. Eur. Conf.
Power Electron. Appl., 1997, pp. 2.625–2.630.
[31] J. Myrzik and P. Zacharias, “New inverter technology and harmonic dis-
tortion problems in modular PV systems,” in Proc. Eur. Photovoltaic Solar
Energy Conf. Exhib., 1997, pp. 2207–2210. Zhigang Liang (S’10) was born in Sichuan, China,
[32] J. M. A. Myrzik, “Novel inverter topologies for single-phase standalone or in 1981. He received the B.S. and M.S. degrees
grid-connected photovoltaic systems,” in Proc. Power Electronics Drive in electrical engineering from Zhejiang University,
Systems (PEDS), 2001, pp. 103–108. Hangzhou, China, in 2003 and 2006, respectively.
[33] D. C. Martins and R. Demonti, “Photovoltaic energy processing for utility He is currently working toward the Ph.D. degree
connected system,” in Proc. IEEE Ind.l Electron. Soc. (IECON), 2001, in the Future Renewable Electric Energy Delivery
pp. 1292–1296. and Management (FREEDM) Systems Center, North
[34] D. C. Martins and R. Demonti, “Grid connected PV system using two Carolina State University, Raleigh.
energy processing stages,” in Proc. IEEE Photovoltaic Spec. Conf., 2002, From 2006 to 2007, he was a System Engineer with
pp. 1649–1652. Monolithic Power Systems (MPS), Inc., Hangzhou,
[35] T. Shimizu, K. Wada, and N. Nakamura, “A flyback-type single phase China. His research interests include high-efficiency
utility interactive inverter with low-frequency ripple current reduction on power conversion, micro inverters and MICs for Photovoltaic applications, and
the DC input for an AC photovoltaic module system,” in Proc. IEEE Power energy management in dc microgrid.
Electron. Spec. Conf. (PESC), 2002, pp. 1483–1488.
[36] T. Shimizu, K. Wada, and N. Nakamura, “Flyback-type single-phase utility
interactive inverter with power pulsation decoupling on the dc input for an
ac photovoltaic module system,” IEEE Trans. Power Electron., vol. 21,
no. 5, pp. 1264–1272, Sep. 2006.
[37] N. P. Papanikolaou, E. C. Tatakis, A. Critsis, and D. Klimis, “Simplified
high frequency converters in decentralized grid-connected PV systems: Rong Guo (M’10) was born in Hunan, China, in
A novel low-cost solution,” in Proc. Eur. Conf. Power Electron. Appl., 1982. She received the B.S. degree in electrical engi-
[CD-ROM], 2003. neering and automation from Xi’an Jiaotong Univer-
[38] N. Kasa, T. Iida, and A. K. S. Bhat, “Zero-voltage transition flyback sity, Xi’an, China, in 2003, the M.S. degree in power
inverter for small scale photovoltaic power system,” in Proc. IEEE Power electronics from Zhejiang University, Hangzhou,
Electron. Spec. Conf. (PESC), 2005, pp. 2098–2103. China, in 2006, and the Ph.D. degree in electrical
[39] A. Fernandez, J. Sebastian, M. M. Hernando, M. Arias, and G. Perez, “Sin- engineering from North Carolina State University,
gle stage inverter for a direct ac connection of a photovoltaic cell module,” Raleigh, in 2010.
in Proc. IEEE Power Electron. Spec. Conf. (PESC), 2006, pp. 93–98. She is currently an Application Engineer at the
[40] S.-J. Jang, C.-Y. Won, B.-K. Lee, and J. Hur, “Fuel cell generation system Rhode Island IC Design Center, International Rectier,
with a new active clamping current-fed half-bridge converter,” IEEE Warwick, RI, engaged on the denition and applica-
Trans. Energy Convers., vol. 22, no. 2, pp. 332–340, Jun. 2007. tion of multiphase dc/dc converter ICs for servers and desktop computers. Her
[41] S.-K. Han, H.-K. Yoon, G.-W. Moon, M.-J. Youn, Y.-H. Kim, and research interests include high-frequency power conversion, analog IC design,
K.-H. Lee, “A new active clamping zero-voltage switching PWM current- and lighting technology.
fed half-bridge converter,” IEEE Trans. Power Electron., vol. 20, no. 6,
pp. 1271–1279, Nov. 2005.
[42] M. Meinhardt, T. O’Donnell, H. Schneider, J. Flannery, C. O. Mathuna,
P. Zacharias, and T. Krieger, “Miniaturised “low profile” module inte-
grated converter for photovoltaic applications with integrated magnetic
components,” in Proc. IEEE Applied Power Electronics Conference Ex-
position (APEC), Mar. 1999, vol. 1, pp. 305–311.
13. LIANG et al.: HIGH-EFFICIENCY PV MODULE-INTEGRATED DC/DC CONVERTER FOR PV ENERGY HARVEST IN FREEDM SYSTEMS 909
Jun Li (S’07) was born in Liaoning, China, in 1981. Alex Q. Huang (S’91–M’94–SM’96–F’05) received
He received the B.S. degree in automation from the B.Sc. degree in electrical engineering from
Tianjin University, Tianjin, China, in 2004, the M.S. Zhejiang University, Hangzhou, China, in 1983, the
degree in power electronics from Zhejiang Univer- M.Sc. degree in electrical engineering from the
sity, Hangzhou, China, in 2006, and the Ph.D. degree Chengdu Institute of Radio Engineering, Chengdu,
in power electronics from North Carolina State Uni- China, in 1986, and the Ph.D. degree from
versity, Raleigh, in 2010. Cambridge University, Cambridge, U.K., in 1992.
He is currently a Senior R&D Engineer in ABB From 1994 to 2004, he was a Professor with the
U.S. Corporate Research Center, Raleigh, NC. His Center for Power Electronics Systems, Virginia Poly-
research interests include topology and control of technic Institute and State University, Blacksburg.
high-power multilevel converters for MV drives and Since 2004, he has been a Professor of Electrical
renewable energy generation. Engineering with North Carolina State University (NCSU), Raleigh, and the
Director of NCSU’s Semiconductor Power Electronics Center. He is also the
Progress Energy Distinguished Professor and the Director of the new National
Science Foundation’s Engineering Research Center for Future Renewable Elec-
tric Energy Delivery and Management Systems, Department of Electrical and
Computer Engineering, North Carolina State University, Raleigh. His research
areas are power management, emerging applications of power electronics, and
power semiconductor devices. He has published more than 200 papers in jour-
nals and conference proceedings, and holds 14 U.S. patents.
Prof. Huang is the recipient of the NSF CAREER Award and the prestigious
R&D 100 Award.