DECENTRALIZED MULTI STRING PV SYSTEM WITH INTEGRATED ZVT CELL RAFAEL C. BELTRAME, MATHEUS I. DESCONZI, HÉLIO L. HEY Power Electronics and Control Research Group, Federal University of Santa Maria 97105-900, Santa Maria, RS, BRAZIL E-mails: beltrame@mail.ufsm.br, matheusiensen@gmail.com, hey@smail.ufsm.br MÁRIO L. DA SILVA MARTINS, HELDER T. CAMARA Power Analysis and Processing Research Group, Federal University of Technology – Parana 85503-390, Pato Branco, PR, BRAZIL E-mails: mlucio@utfpr.edu.br, contato@htc.eng.br Abstract⎯ This paper exploits the advantages and limitations of using a cascaded connection of DC-DC step-up stages for a series string of PV panels, in a single-phase residential/commercial grid connected installation. It is demonstrated that when multistring PV systems are adopted in order to minimize shadowing problems, it is required to employ some approach to reduce the switching losses (turn-on losses of MOSFETs and diode reverse recovery losses) of the DC-DC step-up stages. As a multi-string PV system is normally comprised by several DC-DC step-up stages, integrated soft-switching topologies are attractive due to their compactness, reliability and low cost. Thus, this work proposes the use of an integrated ZVT cell, which assists all the DCDC step-up stages and employs a very compact circuit, enabling to minimize the switching losses, improving the system efficiency. The proposed cell makes use of a magnetically-coupled auxiliary voltage source implemented by adding a secondary winding on the input inductors. In order to validate the proposed topology, experimental results are presented. Keywords⎯ Multi-string PV system, DC-DC step up converter, zero-voltage-transition. 1 Introduction Recent advances in power electronics and semiconductor technology associated with favorable incentives, such as low-interest loans and tax incentives, has led to growth of the PV market mostly due to the proliferation of grid connected PV systems, Cory (2008). PV systems connected to the utility do not require back-up system and charge regulator, i.e., they are commonly simpler, cheaper and more reliable when compared to a stand alone one. Nonetheless, the wide utilization of PV is still limited by the initial installation cost. Likewise other intermittent renewable sources, to optimize the PV array area and make the implantation of grid connected PV systems economically viable, it is essential to drain the maximum power produced by the PV array. It can be done by means of a maximum power point track algorithm, Femia (2005) and Yu (2002), commonly applied to the front-end DC-DC stage. Indeed, PV modules cannot produce identical energy levels in a PV array for several reasons, including dissimilarities of panel production, different temperatures and irradiations due to panel orientation, aging and, in some cases, shading, Carannante (2009). As the PV panels in a series string are constrained to all conduct the same current, the least efficient panel sets the string current. In long string arrays, this situation results in the lost of the maximum power point of the array, which reduces the overall efficiency of the array and, in extreme cases, it can also cause the degeneration of the panels due to the hot spot phenomenon. To overwhelm these problems, and also ensure the maximum production of electricity even when shading or other dissimilarities are present, decentralized PV systems have been proposed in the literature. Some works presented demonstrate that a gain of up to 16% of generated energy could be achieved on decentralized PV systems, Imhoff (2008). In general, decentralized PV systems split the maximum power point tracker (MPPT) stage, which can be dedicated exclusively to a single or a group of PV panels. They can be grouped in three categories, namely module integrated systems, Meinhardt (1999), and string and multi-string systems, Meinhardt (2001). The current, voltage and power ratings of DC-DC step-up MPPT semiconductors, as well as their voltage gains are quite different for each decentralized system. Depending on the semiconductor technology and the required voltage gain, the conduction and switching losses can be particularly distinct from one decentralized system DC-DC step-up stage to another. As a multi-string PV system can be comprised by several DC-DC step-up stages, to reduce the turnon losses of MOSFETs and diode reverse recovery losses of all DC-DC stage is a quite complex task since the auxiliary circuitry may increase significantly the converter size, weight and cost. Additionally, a large semiconductor count could reduce the reliability of the PV system. This paper exploits the advantages and limitations of using a cascaded connection of DC-DC step-up stages for a series string of PV panels, in a singlephase residential/commercial grid connected installation with an integrated soft-switching topology, Schuch (2009). The proposed Zero-Voltage Transition ACC which assists all the DC-DC step-up stages and employs a very compact circuit, enabling to Table 1. Parameters for the Hypothetical Standard Module. minimize both the turn-on losses of MOSFETs and diode reverse recovery losses. In order to validate the proposed topology theoretical analysis and experimental results are presented. 2 Loss Analysis of Multi-string Grid-Connected PV System Multi-string PV system combines the advantage of higher energy yield of a string PV system with the lower costs of a central inverter PV system, as shown in Figure 1, allowing the use of individual MPPT algorithm for each one, Myrzik (2003), reducing the possibility of shadowing mismatches that may occur in a single long string PV module configuration. From the efficiency perspective, the series connection of the DC-DC stages reduces their voltage gain and, consequently, related conduction losses. In addition, the DC-DC stage outputs connected in series may provide a DC bus voltage that matches directly with those used to feed standard inverter stages. Figure 2 shows the efficiency comparison of multi-string PV systems with one, two, three and four input DC-DC step-up stages. It can be seen the efficiency is proportional to the number of input stages. It occurs because the larger the number of stages is, the lower the input DC-DC converter gain is for the same DC bus voltage (defined as 250 V), reducing the RMS current through the transistor and the related conduction losses. Additionally, the switching losses (turn-on losses of MOSFET and diode reverse recovery losses) are minimized because the voltage applied on semiconductor devices is reduced. For the theoretical curve, the efficiency was estimated by using the loss models presented in Appendix A. To simplify the analysis, the systems make use of the Hypothetical Standard Module (HSM) specified in Table 1. The step-up stage parameters and components are defined in Table 2. It is also considered that the step-up stage is operating in the maximum power point (MPP) of the PV arrays. A MOSFET is employed in step-up converter because it allows the increase of the converter switching frequency, and associated volume, weight and, mainly, cost reduction of the entire system. The MOSFET utilized were the N-Channel Standard Power MOSFET IRFP460 (500V/20A). The di- Parameter Value VOC (ISC) 21.9 V (8.02 A) Peak power voltage (current) 17.6 V (7.39 A) Peak power (No. of panels of the array) 130 W (3) Table 2. Parameters for the input DC-DC step-up stages. Parameter Value Switching frequency, fs 100 kHz Input filter inductor, L (N of turns) 285 μH (22 turns) Ferrite core NEE – 65/26 Maximum flux density, BMax 0.2 T Copper wire 3 x 63 x 37 AWG (Litz) Output capacitor, C 470 μF Main switch (MOSFET) IRFP460 Main diode 8ETH06 ode utilized was the Ultrafast Switching Mode Power Rectifier 8ETH06 (600V/8A). Also in Figure 2 experimental data from a laboratory prototype is depicted, it can be seen by the plotted curves that the estimated losses presents a good agreement with the data measured from the laboratory prototype. The distribution of the losses in the DC-DC step-up semiconductors for a multi-string PV system composed by two stages (93.6%, vide Figure 2) is shown in Figure 3. It can be concluded that in the MOSFET based PV system, the transistor turn-on and turn-off losses represents about 23% of the semiconductor losses, and the diode switching losses represents 35%, confirming the importance in reducing switching losses. 3 Proposed Multi-String Grid-Connected PV System with Integrated ZVT Cell As the switching frequency of the step-up stages increases, the large reverse-recovery currents of the diodes affect the system efficiency and also produce electromagnetic interference (EMI) noise. To overcome 96 Efficiency (%) 95 94 93 92 Experimental 91 Estimated 90 Figure 1. Decentralized multi-string PV system with a cascaded connection of DC-DC step-up stages. 1 2 3 Number of Input Stages 4 Figure 2. Efficiency comparison of hard switched multi-string PV system. 27% 35% Trans. Cond. Diode Swit. Trans. T-on 6% Trans. T-off Diode Cond. 17% 14% Figure 3. Semiconductor power loss balance in the PV system. those problems, various passive, Tseng (2007), Li (2009), and active approaches have been proposed for the DC-DC step-up converter, Jovanovic (2005), Martins (2005) and Wang (2005). Among these techniques, the zero-voltage-transition (ZVT) technique is especially suitable to overcome the aforementioned drawbacks because it reduces the diode reverse recovery losses by controlling the di/dt slope of the current during the turn-off of the diode. Furthermore, it effectively minimizes the switching losses and also promotes the absorption of the parasitic capacitances energy of semiconductor devices, Martins (2004), minimizing the turn-on capacitive losses when majority carrier devices such as MOSFETs are employed. There are several ways to implement the ZVT auxiliary commutation circuit (ACC), depending on how the auxiliary voltage source of the ACC is synthesized. In converters that present a filter inductor, as the DCDC step-up converter, an auxiliary voltage source implemented by a magnetic coupling with the inductor filter, as proposed by Lee (1994), is advantageous because the voltage source is implemented in the same magnetic core as the filter inductor, and the auxiliary inductor can be implemented by means of the leakage inductance of these inductors, Russi (2008). This way, these features result in a very compact topology in that there are no extra magnetic components. 3.1 The Integration Concept The inclusion of ACCs increases the number of additional components, and therefore the cost and the complexity of the overall system. It is a critical problem for systems with several power conversion stages, such as decentralized PV multi-string systems, as can be seen in Figure 1. In these systems, it would be necessary to include an independent ACC for each power conversion stage, decreasing the usefulness of these circuits. Aiming to reduce the number of additional auxiliary devices, the integration concept of ACCs can be employed, which is suitable to systems with several power conversion stages. This concept is based on sharing an ACC among different converters or stages, aiming to reduce the number of additional auxiliary devices, Schuch (2009). Therefore, it is proposed the use of the unpublished integrated ZVT cell shown in Figure 4, which employs a very compact circuit and can assist all the DC-DC step-up stages. The proposed ACC makes use of a magnetically-coupled auxiliary voltage source implemented by adding a secondary winding on the input inductors. As the ACC is galvanic isolated from the main power circuit, it can be shared by all the DCDC step-up converters with none short-circuit risk. The ZVT cell presented in Figure 4 makes possible to minimize diode reverse recovery losses of D1 to Dn by controlling the di/dt slope of their currents and to assist the turn-on of S1 to Sn simply by using a phase difference of 360º n among their command signals, where n is the number of DC-DC step-up stages, or by synchronizing their turn-on instants. 3.2 Integrated ZVT Cell Operation In order to explain the ACC operation, for simplicity it is considered only one DC-DC step-up converter, as presented in Figure 5, where the coupledinductor was replaced by its cantilever model, Maksimovic (2000). Furthermore, the input inductor (L1) was approximated by a constant current source (IL1). This approximation can be made if the magnetizing inductance of the coupled-inductor (Lm) is much larger than the leakage inductance (Lx) and the ripple of the current through L in a switching period is small. Mode 1 ( t ≤ t0 ) : previous to the ACC operation, S is turned off and D1 is conducting IL1, as presented in Figure 5 (a) and in Figure 6 for t ≤ t0 . During this mode, the voltage on Cs1 (vS1) and the current through Lx (iLx) are Vo1 and zero, respectively. Mode 2 ( t0 < t ≤ t1 ) : aiming to commutate IL1 from D1 to S1, the auxiliary switch Sx is turned on at “t0”. The circuit configuration during this mode is represented in Figure 5 (b) and the main theoretical waveforms in Figure 6 for t0 < t ≤ t1 . The voltage on Cs1 is Vo1 and the current though Lx increases linearly with a slope given by N (Vo1 − Vi1 ) Lx . It is important to notice that the turns ratio (N) must be positive in order to apply a positive voltage on Lx. This mode ends at “t1”, when NiLx(t) reaches IL1 and D1 turns off with a controlled di/dt slope. The time duration of this mode, defined as Δt1 = t1 − t0 , can be calculated by (1). PV string Vin Dxn N Lin CSx N Li1 Ci1 a1 S1 Dn Csn Con Sn Vi1 Dx1 Sx Cin an Von VBUS D1 Cs1 Co1 Vo1 Figure 4. Proposed multi-string PV system with integrated ZVT. Δt1 = Lx I L1 N (Vo1 − Vi1 ) Where (1) 2 ωn = Mode 3 ( t1 < t ≤ t2 ) : this mode begins at “t1”, when D1 turns off, and both Cs1 and Lx go to a resonant process. The circuit configuration during this mode is represented in Figure 5 (c) and the main theoretical waveforms in Figure 6 for t1 < t ≤ t2 . The voltage on Cs1 goes zero during the resonant process. The current peak though Lx can be calculated by (2). Vo1 − Vi1 I L1 + Z N iLx ( max ) = (2) Where Z= Lx . Cs1 (3) (4) This mode ends at “t2”, when vS1(t) reduces to zero and is clamped by the antiparallel diode of S1. Thus, the time duration of this stage, defined as Δt2 = t2 − t1 , can be calculated by (5). Δt2 = ⎛ Vi1 ⎞ arccos ⎜ ⎟ ωn ⎝ Vi1 − Vo1 ⎠ 1 Dx1 Lx iLx NvL1 CSx vSx Sx IL1 vL1 ~ ~ Vo1 Vi1 Dx1 Nv iLxL1 NiLx CSx vSx ~ L1 iD1 Vo1 Vi1 ~ ~ vL1 Cs1 vS1 S1 (e) Mode 5t ( tt3 < t ≤ t tt4 ) .t 0 1 2 3 4 ~ a1 iS1I ~ D1 Lx NvL1 NiLxIL1/N a1 iL IL1 vL1 S1 IL1 Vo1 Vi1 7 Lx Lx iLx IL1 vL1 CSx vSx Sx NiLx D1 Cs1 vS1 S1 a1 Vo1 Vi1 IL1 vL1 D1 Cs1 vS1 S1 Vo1 (d) Mode 4 ( t2 < t ≤ t3 ) . Dx1 CSx vSx Sx Lx S1 iLx NvL1 NiLx a1 Vo1 Vi1 iLx NvL1 NiLx IL1 vL1 NvL1 D1 t t 6 CSx vSx Sx (c) Mode 3 ( t1 < t ≤ t2 ) . t (f)t Mode t t 6 ( t4 < t ≤ t5 ) . 5 iLx Dx1 Cts1 vS1 (8) Dx1 a1 Vo1 t iLx2NS(xVo1-V Ci1Sx) vSx Lx Vo1 − Vi1 I sin (ωn Δt2 ) + L1 . Z N suming gradually IL1, the circuit goes to the configuration represented in Figure 5 (e). The circuit operation is similar to that in the previous mode. The main theoretical waveforms are presented in Figure 6 for t3 < t ≤ t4 . This mode ends at “t4”, when iLx(t) reduces to zero and Dx1 turns off under a controlled di/dt slope. It is important to highlight that after the turn-off of Dx1, Sx can be turned off under zero-current-switching (ZCS). Thus, an IGBT can be employed as auxiliary D1 Cs1 vS1 t (7) N 2 Vi1 Mode 5 ( t3 < t ≤ t4 ) : at “t3”, when S1 starts as- NvL1 S1 Dx1 ~ Sx Vi1 iLx ( t2 ) = (b) Mode 2 ( t0 < t ≤t t1 ) . ~ Mode 1 ( t ≤ t0 ) . v(a) S1 Lx ( N iLx ( t2 ) − I L1 ) Where NiLx a1 Cs1 vS1 S1 Lx vSxiLx CSx vSx D1 GS1IL1 GSx Δt3 = Dx1 Sx NvL1 NiLx vL1 iLx Lx (6) “t2”, the antiparallel diode of S1 stars conducting, as presented in Figure 5 (d). The voltage on Cs1 remains clamped at zero volts and the current though Lx decreases linearly with a slope given by − NVi1 Lx . It is important to notice that N must be positive in order to apply a negative voltage over Lx. The main theoretical waveforms are presented in Figure 6 for t2 < t ≤ t3 . This mode ends at “t3”, when NiLx(t) reduces under IL1 and S1 starts assuming gradually IL1. The time duration of this mode, defined as Δt3 = t3 − t2 , can be calculated by (7). Dx1 a1 Vi1 (5) . Mode 4 ( t2 < t ≤ t3 ) : after vS1(t) reaching zero at In order to achieve the zero-voltage-switching (ZVS) condition to S1, it is mandatory that vS1(t) reduces to zero volts. It can be demonstrated that the restriction (4) must be observed in order to satisfy the aforementioned ZVS condition. The restriction (4) implies that the converter gain must be always higher than two. Vo1 ≥2 Vi1 N Lx Cs1 NiLx D1 Cs1 vS1 CSx vSx Sx a1 Vo1 Vi1 (g) Mode 7 ( t5 < t ≤ t6 ) . Figure Single DC-DC step-up stage with ZVT cell operation modes. Figure 6. Theoretical waveforms of the5.proposed topology. IL1 vL1 S1 D1 Cs1 vS1 (h) Mode 8 ( t6 < t ≤ t7 ) . Vo1 switch. The time duration of this mode, defined as Δt4 = t4 − t3 , can be calculated by (9). Δt4 = Lx I L1 N 2 Vi1 turns to the configuration of Figure 5 (a) (mode 1). The voltage over Sx at “t7” has the level expressed in (11). vSx ( t7 ) = 2 N (Vo1 − Vi1 ) (9) 4 Mode 6 ( t4 < t ≤ t5 ) : after Dx1 turning off at “t4”, S1 conducts IL1. Thus, vS1(t) and iLx(t) are both zero. During this mode, represented in Figure 5 (f), the PWM modulation is implemented. This mode ends at “t5”, when S1 is turned off. The main theoretical waveforms are presented in Figure 6 for t4 < t ≤ t5 . Mode 7 ( t5 < t ≤ t6 ) : this mode begins at “t5”, when S1 is turned off and IL1 is transferred to the snubber capacitor Cs1. This way, Cs1 is charged linearly with a slope given by I L1 Cs1 . The circuit configuration during this mode is represented in Figure 5 (g) and the main theoretical waveforms are presented in Figure 6 for t5 < t ≤ t6 . This stage ends at “t6”, when vS 1 ( t ) = Vo1 and D1 starts conducting IL1, clamping vS1(t) at Vo1. Additionally, the time duration of this mode, defined as Δt5 = t6 − t5 , can be calculated by (10) C V Δt5 = s1 o1 I L1 Design Methodology The design methodology for the auxiliary ZVT circuit is based on the selection of Lx and N in order to satisfy the restrictions that guarantee the proper ACC operation and also limit the voltage and current stresses in the DC-DC step-up stage semiconductor devices. 4.1 Selection of N As mentioned during the analysis of the operation modes 2 and 4, N must be positive in order to guarantee the magnetizing and demagnetizing conditions to Lx, as expressed in (12). Furthermore, in order to comply with the maximum voltages of Dx1 (vDx(max)) and Sx (vSx(max)), which occur during mode 1, (13) and (14) must be satisfied. N ( min ) (10) N ( max ) Mode 8 ( t6 < t ≤ t7 ) : this mode only exists due to the intrinsic capacitance of Sx (CSx). In the following analysis, it is assumed that the linear charge of Cs1 (snubber commutation) was finished. Thus, the voltage applied over L1 is Vi1 − Vo1 at “t6”. This voltage, reflected to the ACC side, turns Dx1 on and enables Lx and CSx to start a resonant process. The circuit configuration during this mode is represented in Figure 5 (h) and the main theoretical waveforms in Figure 6 for t6 < t ≤ t7 . This mode ends at “t7”, when iLx(t) reduces to zero and is Dx1 turned off. This way, the circuit re- ~ ~ GS1 t ~ GSx Vo1 ~ vS1 vSx ~ ~ iLx 2N(Vo1-Vi1) ~ ~ iS1 IL1 t0 t 1 t 2 t3 t4 t5 t6 t 7 Figure 6. Theoretical waveforms of the proposed topology. N ( max ) Mag = N ( min ) vSx ( max ) = >0 (12) vSx ( max ) (13) 2 (Vo1 − Vi1 ) = vDx ( max ) Des vDx ( max ) (14) (Vo1 − Vi1 ) 4.2 Selection of Lx The minimum value of Lx that comply with the maximum di/dt slope during the turn off of D1 (di/dtD) can be calculated from (1), as expressed in (15). In the same way, the minimum value of Lx that comply with the di/dt slope of Dx1 (di/dtDx) can be calculated by (16). Furthermore, the minimum value of Lx that limits iLx(max) to the levels of auxiliary semiconductor devices (iACC(max)) can be calculated by (17). On the other hand, the maximum value of Lx that guarantees that the magnetization time will not be larger than a fraction of the switching time (∆t1(max)), offsetting the duty cycle of the converter, can be calculated by (18). Lx ( min ) t didt D Lx ( min ) t = N 2 (Vo1 − Vi1 ) didt Dx (15) didtD = N Vi1 didtDx (16) 2 t iL ~ iD1 IL1/N t (11) t t Lx ( min ) iLx ( max ) Lx ( max ) ⎛ ⎞ Vo1 − Vi1 =⎜ ⎟⎟ Cs1 ⎜i ⎝ ACC ( max ) − I L1 N ⎠ Δt1( max ) = N 2 (Vo1 − Vi1 ) Δt1( max ) I L1 (17) (18) 4.3 Design Abacus It is possible to find a set of N and Lx combinations that comply with the specifications of the con- Table 3. Converter specification and semiconductor limitations. Table 4. Semiconductor devices and ACC specifications. Parameter Value Parameter Value Note Input power (Pin) 390 W Main switch (S1) IRFP460 Discrete MOSFET Main diode (D1) 8ETH06 Discrete Diode Switching frequency (fs) 100 kHz Input voltage (Vi1) / Output voltage (Vo1) 52.8 V / 250 V Maximum voltage over Sx (vSx(max)) / Dx (vDx(max)) 480 V / 480 V Auxiliary switch (Sx) IRG4BC20UD Discrete IGBT Auxiliary diode (Dx1) Discrete Diode 8ETH06 Maximum current through the ACC (iACC(max)) 13 A Ferrite core NEE – 65/26 – Maximum di/dt though D (di/dtD) and Dx1 (di/dtDx) 100 A/μs Cooper Wire 63 x 37 AWG Litz Snubber capacitor (Cs1) - COSS 480 pF Input inductor (L1) 285μH – Auxiliary inductor (Lx) 7.1μH Leakage 1.21 N (pri) = 22 x 3 N (sec) = 27 x 1 verter and the limitations (maximum voltages and currents) of the semiconductor devices. Thus, the restrictions (12) to (18) are plotted in Figure 7 using the specifications summarized in Table 3. This way, from Figure 7 was selected Lx = 7.1μH and N = 1.21 in order to satisfy the restrictions defined in Table 3. 5 System Implementation and Experimental Results In order to verify the reliability of the proposed system, an experimental analysis is carried out for a single DC-DC step-up converter employing the integrated ZVT cell presented in Figure 4. The semiconductors employed in the prototype are summarized in Table 4, as well as some additional ACC specifications. For this analysis the PV panels were emulated by a constant voltage source with a voltage level equal to the MPP voltage, and the MPP current was drained from the voltage source by adjusting the DC-DC step-up converter duty-cycle. Figure 8 presents the experimental waveforms of the converter operating at nominal power. In Figure 8 (a)-(b) are presented the stead-state waveforms of the command signals of both main (vGS1(t)) and auxiliary switches (vGSx(t)), as well as their voltage (vS1(t) and vSx1(t), respectively), and the current through the auxiliary inductor (iLx(t)). It is possible to see in Figure 8 (a)-(b) that the ACC intervention demands a small portion of the switching period. It must be noted that the linear rising of iLx(t) (with di/dt controlled by Lx) corresponds to a linear falling of the current though Auxiliary Inductor, Lx (μH) 15 Turns ratio (N) D1, reducing its reverse recovery losses. It is possible to see in Figure 8 (b) that Sx1 is turned off after iLx(t) reaching zero, characterizing a ZCS commutation. Furthermore, Figure 8 (c) presents the details of the turn-on commutation of S1. As can be seen in Figure 8 (c), the main switch S1 is turned on after vs1(t) vS1 1 iLx 3 Ch1: 100 V/div Ch3: 5 A/div vGSx 4 vSx 1 iLx 3 Ch1: 100 V/div Ch3: 5 A/div vGS1 Des 10 Time: 2 us/div Ch4: 20 V/div (b) vSx(max) Mag Time: 2 us/div Ch4: 20 V/div (a) Δt1(max) 12.5 vGS1 4 4 iACC(max) di/dtD ZVS vS1 vDx(max) 7.5 Project 1 5 2.5 0 iLx di/dtDx 0 0.5 1 1.5 2 Turns Ratio, N Figure 7. Design abacus. 3 2.5 3 Ch1: 100 V/div Ch3: 5 A/div Time: 500 ns/div Ch4: 20 V/div (c) Figure 8. Experimental results. (a) Main circuit in stead-state operation. (b) Auxiliary circuit in stead-state operation. (c) ZVS commutation detail. retical and experimentally. A design procedure to determine its auxiliary components have been presented and discussed. The proposed topology was validated by experimental results obtained from a single DCDC step-up converter operating at 390 W and 100 kHz. The experimental results proved the effectiveness of the converter. 94.5 Efficiency (%) 94 93.5 93 92.5 Soft-Switched 92 91.5 200 Appendix A Hard-Switched 250 300 350 M PP Power (W) 400 Figure 9. Converter efficiency in function of solar irradiation. In a range from 600 to 1000 W/m2. reaches zero, characterizing a ZVS commutation. Figure 9 shows the measured efficiency of the DC-DC step-up prototypes evaluated experimentally as a function of PV modules power for a set of irradiances defined a priori. It can be observed that the efficiencies are inversely proportional to the PV module power, since the conduction losses increase with the module current. In spite of it, the Integrated ZVT DCDC prototype presented higher efficiency for the entire module power range. The real efficiency gain of the ZVT DC-DC prototype can be seen in Figure 10. 6 Conclusions This paper presented and analyses an integrated ZVT ACC that uses a magnetically-coupled auxiliary voltage source implemented by adding a secondary winding on the input inductors. The proposed ACC assists all the DC-DC step-up converters of a decentralized multi-string PV system employing a very compact circuit, reducing the cost and size of the overall system, and improving its performance by reducing the turn-on capacitive losses of the main switches and the diode reverse recovery losses. Thus, the benefits of using multi-string PV systems, such as reducing the possibility of shadowing mismatches and increasing the generated energy are kept without penalizing the system reliability and cost. The proposed integrated ZVT was analyzed theo- The forward voltage drop on both main transistor and main diode can be estimated by (19), where the coefficients A, B and C, summarized in Table 5, can be obtained using the data extracted from the device datasheet, and idev is the instantaneous current through the device. vdev ( idev ) = A idev B + C This way, the transistor and diode conduction losses can be calculated by (20), where Ts is the switching period, Beltrame (2009). Pcnd = 1 Ts Ts ∫ v ( i ) i ( t ) dt dev dev dev (20) 0 The transistor switching losses can be divided in turn-on and turn-off losses. The turn-on losses are composed by the overlap between the voltage and current though the transistor, McMurray (1980), and by the turnon capacitive losses, Beltrame (2009). On the other hand, the turn-off losses are mainly due to the overlap between the voltage and current. Thus, the turn-on capacitive losses can be calculated by (21), where Vo is the output voltage of the DC-DC step-up converter. Additionally, the turn-off losses can be calculated by (22). The parameters Coss, tf and tr are defined in Table 5. Pon = 1 1 idevVo tr f s + CossVo 2 f s 2 2 (21) 1 idevVo t f f s (22) 2 The switching losses of the diode are basically due to the reverse recovery phenomenon. Then, the switching losses of the diode can be estimated by (23), Schönberger (2008), where Qrr is defined in Table 5. Poff = Prr = QrrVo Efficiency Gain 0,4 0,35 0,3 0,25 0,2 0,15 0,1 0,05 0 (19) (23) Finally, the losses observed in the inductor can be estimated by (24), where ILrms is the RMS current though the inductor and RL is the DC resistance of the cooper wire. It is important to notice that only the cooper losses Table 5. Parameters employed to estimate the converter losses. Device 600 700 800 900 1000 Irradiation Figure 10. Integrated ZVT efficiency gain in function of the irradiation in W/m2. Main switch (IRGP460) Parameter Value A; B; C 0.083; 1.380; 0.557 Coss 480 pF tr; tf 81 ns; 65 ns Main diode (8ETH06) A; B; C 0.969; 0.291; 0.235 Qrr 115 nC Inductor RL 39.61 mΩ were take into account to calculate the inductor losses. PL = I Lrms 2 RL (24) Acknowledgment The authors would like to express their gratitude to “Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – CAPES” and “Conselho Nacional de Desenvolvimento Científico e Tecnológico – CNPQ” (proc. nº 307798/2009-7 and proc. nº 478154/2009-7) for financial support. References Beltrame, R.C., Candido, D.B., Martins, M.L.S., Pinheiro, J.R. and Hey, H.L., “Comparison between integrated and simplified ZVT topologies for three-phase voltage-source inverters,” in Proc. IEEE Industrial Electronics Society Conf., 2009, pp. 7-12. Carannante, G., Fraddanno, C., Pagano, M. and Piegari, L., “Experimental Performance of MPPT Algorithm for Photovoltaic Sources Subject to Inhomogeneous Insolation”, IEEE Trans. on Industrial Electronics, Vol. 56, No. 11, pp. 4374-4380, 2009. Cory, K., Coughlin, J., Jenkin, T., Pater, J. and Swezey, B., “Innovations in Wind and Solar PV Financing”, Technical Report NREL/TP-67042919, 2008. Femia, N., Petrone, G., Spagnuolo, G. and Vitelli, M., “Optimization of Perturb and Observe Maximum Power Point Tracking Method”, IEEE Trans. on Power Electronics, Vol. 20, No. 4, pp. 963-973, July 2005. Imhoff, J., Pinheiro, J.R., Russi, J.L., Brum, D., Gules, R. and Hey, H.L., “DC-DC Converters in a multistring configuration for stand-alone photovoltaic systems”, in Proc. IEEE Power Electronics Specialists Conf., 2008, pp. 2806-2812. Jovanovic, M.M. and Jang, Y., “State-of-the-Art, SinglePhase, Active Power-Factor-Correction Techniques for High-Power Applications – An Overview”, IEEE Trans. on Industrial Electronics, Vol. 52, No. 3, pp. 701-708, June 2005. Li, R.T.H., Chung, H.S.H. and Sung, A.K.T., “Passive lossless snubber with minimum voltage and current stress for boost PFC”, in Proc. IEEE Energy Conversion Congress and Exposition, 2009, pp. 940-947. Maksimovic, D., Erickson, R.W. and Griesbach, C., “Modeling of cross-regulation in converters containing coupled inductors,” IEEE Trans. on Power Electronics, Vol. 15, No. 4, pp. 607615, 2000. Martins, M.L.S. and Hey, H.L., “Self-commutated auxiliary circuit ZVT PWM converters”, IEEE Trans. on Power Electronics, Vol. 19, No. 6, pp. 1435-1445, Nov. 2004. Martins, M.L.S. , Russi, J.L. and Hey, H.L., “Zerovoltage transition PWM converters: a classification methodology”, in Electric Power Applications, IEE Proceedings, Vol. 152, No. 2, pp. 323-334, 2005. McMurray, W., “Selection of Snubbers and Clamps to Optimize the Design of Transistor Switching Converters”, IEEE Trans. on Industry Applications, Vol. IA-16, No.4, pp. 513-523, July 1980. Meinhardt, M., O’Donnell, T., Schneider, H., Flannery, J., Mathuna, C.O., Zacharias, P. and Krieger, T., “Miniaturised ‘low profile’ module integrated converter for photovoltaic applications with integrated magnetic components”, in Proc. IEEE APEC, Vol. 1, 1999, pp. 305-311. Meinhardt, M., Wimmer, D. and Cramer, G., “Multistring-converter: The next step in evolution of string-converter”, in Proc. IEEE EPE, 2001. Myrzik, J.M.A. and Calais, M., “String and module integrated inverters for single-phase grid connected photovoltaic systems – a review”, in Proc. IEEE Power Tech Conf., 2003. Russi, J.L., Martins, M.L.S. and Hey, H.L., “Coupled-filter-inductor soft-switching techniques: principles and topologies”, IEEE Trans. on Industrial Electronics, Vol. 55, No. 9, pp. 3361-3373, Sept. 2008. Schuch L., Rech C., Pinheiro J.R., “Integrated auxiliary commutation circuits: a generalised approach,” Power Electronics, IET, Vol. 2, pp. 42-51, 2009. Schonberger, J., Feix, G., “Modelling turn-off losses in power diodes”, in Control and Modeling for Power Electronics, 2008, pp. 1-6. Tseng, S.-Y., Shiang, J.-Z., Chang, H.H., Jwo, W.-S. and Hsieh, C.-T., “A Novel Turn-On/Off Snubber for Interleaved Boost Converters” in Proc. IEEE Power Electronics Specialists Conf., 2007, pp. 2341-2347. Wang, C.M., “A novel zero-voltage-switching PWM boost rectifier with high power factor and low conduction losses,” IEEE Trans. Industrial Electronics, Vol. 52, No. 2, pp. 427–435, Apr. 2005. Yu, G.J., Jung, Y.S., Choi, J.Y., Choy, I., Song, J.H. and Kim, G.S., “A novel two-mode MPPT control algorithm based on comparative study of existing algorithms”, in Photovoltaic Specialists Conf., 2002, pp. 1531-1534.