Study of a 3kW High-Efficient Wide-Bandgap DC- DC Power Converter for Solar Power Integration in 400V DC Distribution Networks

Transcription

1 IEEE PEDS 2017, Honolulu, USA December 2017 Study of a 3kW High-Efficient Wide-Bandgap DC- DC Power Converter for Solar Power Integration in 400V DC Distribution Networks Yucheng Zhang, Yashwanth Bezawada Old Dominion University, VA, USA yzhang@odu.edu Ruiyun Fu Mercer University, GA, USA Weisong Tian, and Robb M. Winter South Dakota School of Mines and Technology, SD, USA Abstract- Direct current (DC) distribution network has a higher efficiency in energy delivery over alternative current (AC) network, when there is a high penetration of dc renewable energy resources and energy storage systems. In this paper, the energy conversion efficiency (η) of a 3kW Silicon Carbide (SiC)-based Boost Cascaded Buck-Boost power converter is analyzed for solar power integration in 400Vdc distribution networks, compared to a Silicon (Si)-based BoCBB prototype. We found that: 1) there is about 2% reduction in η when the switching frequency of MOSFETs increases from 20 khz to 100 khz in Si-based prototype. Comparatively, the efficiency reduction with SiCbased MOSFETs is limited within 1% in the same case; 2) under the same switching frequency, the η increases 2.4% in average by replacing Si MOSFETs with SiC MOSFETs. The differences in η increase along the switching frequency. In addition, the switching frequency of 50 khz is optimal in overall system performance for SiC-based prototype, which makes a good tradeoff between high η, good power quality, and compact size. These conclusions are verified by theoretical analysis and experiment tests. The experiment tests demonstrated a high efficiency of 97% in peak in power conversion. I. INTRODUCTION In the March of 2016, the updated analysis at the National Renewable Energy Laboratory (NREL), Department of Energy (DOE) in the United States, revealed a technical potential of 1,118 gigawatts of capacity and 1,432 terawatt-hours of annual energy generation, close to 40 % of the nation's electricity needs [1]. The worldwide small-scale roof-top solar power installation was 23 gigawatts at the end of 2013, and it was estimated to continue growing at above 20 additional gigawatts per year until 2018 [2]. In the United States, the total photovoltaic (PV) installation (including the solar panels on rooftops as well as in ground-mounted solar farms) reached gigawatts at the end of 2015, and the added capacity in 2015 was 7.3 gigawatts (28.5% of the total) [3]. With more and more solar power generation integrated into the utility and development of micro grids, efficiency gain by applying dc distribution networks drew people s attention. The work in [4] supported by the Pacific Northwest National Laboratory (PNNL), DOE revealed that it was shown that fuel cells or other local dc generation that feed directly into a premise dc bus could have favorable conversion losses. Several twoswitch buck-boost typed DC-DC converters were proposed for solar power generation, including 1) Buck-Cascaded Buck- Boost, 2) Boost-Cascaded Buck-Boost, 3) Buck-Interleaved Buck-Boost, and 4) Boost-Interleaved Buck-Boost DC-DC converters [5] [6]. Due to its low inductor-current stress and small size of passive energy-storage components, a Boost Cascaded Buck- Boost (BoCBB) power converter is adopted here for solar power integration study in 400V DC distribution networks. In this paper, we focus on the study of energy conversion efficiency for a 3kW SiC-based BoCBB power converter. In section II, a preliminary work is performed to specify the parameters of passive components and power losses. After that, the η of SiC-based and Si-based BoCBB converters are analyzed and compared when the switching frequency of MOSFETs is 20 khz, 50 KHz, and 100 KHz, respectively. These analyzes are validated by theoretical calculation in section III and experiment tests in section IV. II. PRELIMINARIES A. Specification of Components The preliminary work specifies the parameters of passive components for a 3kW BoCBB power converter connected into 400V DC distribution networks. In order to regulate the output voltage to the target 400 Vdc and realize maximum power point tracking (MPPT) in response to the unpredictable climate change, the BoCBB power converter is adopted here due to its superiorities in transient and steady-state performance over other candidates. The BoCBB acts as a boost converter at low input voltage from PV panel and a buck converter at high input voltage vice versa. The rated voltage of PV panel is 480 Vdc in this study. The input power and V/I curve of PV panel relates to the ambient temperature and irradiance. It should be noted that classic buck-boost converter cannot be used in this application because of the poor switch utilization, achieving a maximum of 25% at a duty ratio of 50%, when V in = V out (for continuous conduction mode) [7]. The topology of BoCBB power converter is shown in Fig. 1. During the boost-mode operation, L boost acts as a boost inductor and L buck acts as an output current filter. Similarly, during the buck-mode operation, L buck acts as a buck inductor and L boost acts as an input current filter. MOSFET switches are modulated according to the input voltage from PV panel (V in) /17/$ IEEE 680

2 The target connection point of this study is a 400V dc distribution network (V out = 400 V). To control MOSFET switches, i.e. S boost and S buck, the duty cycle (D) of the converter is specified as: when V in > V out, S boost = 0, D = Sbuck = ; When V in <= V out, D = S boost =1-, Sbuck = 1. The output voltage is regulated to 400 Vdc, as shown in Fig. 2, with pulse width modulation (PWM). loss defined in (1) and switching loss defined in (2). The turnon and turn-off delay times of MOSFETs are counted into switching loss calculations.. = ( ) (1). =. + (2) where is through-current in amps, is acrossvoltage in volts, ( ) is on-resistance in ohms, and are rise time and fall time in seconds, respectively, for MOSFETs; is duty cycle of conduction; is switching frequency in Hz. Fig. 1. Topology of BoCBB power converter. Losses in Power Inductors Power inductors operate as choking inductors to limit current harmonics. Two types of losses are presented in power inductors: core loss and low-frequency copper loss. For power inductors and transformers, high operating flux density leads to reduced size, weight and cost. Silicon steel and similar materials exhibit saturation flux densities of 1.5 T to 2 T, but also exhibit high core losses. In case of low-resistance materials, eddy losses is high. The core loss can be approximately calculated from (3) [8]. Fig. 2. Output voltage of BoCBB power converter regulated to 400Vdc. When the switching frequency of MOSFETs is 50 khz, the parameters of L boost and L buck are 5.01 mh and 4.48 mh, respectively to limit the peak-peak current oscillation under 15% of the RMS current. Similarly, the inductances are a double for 20 khz switching and a half for 100 khz switching of these values, according to X L=2πf*L. B. Power Losses Analysis In order to improve the η of the system, it is essential to find the expression of power losses and thus identify the optimal tradeoff in system design. In the power losses analysis here, these major losses are considered in the η calculation: a) Power Losses associated with MOSFETs and Diodes: the majority of power losses are conduction loss and switching loss: P MOSFET = P swit. + P cond.; b) Power Losses associated with inductors: another major power loss is the inductor loss. Similar to typical magnetic devices, two types of power losses are considered here: 1) ferrite core loss, and 2) low frequency copper loss [8]. Losses in Power MOSFETs Losses of power MOSFETs mainly consist of conduction = ( ) (3) where is area across of the core in ;, and are decided by selecting magnetic cores, and here =2.7, which usually lies in the range of [ ]. For the dc dc solar converter in boost mode, = ; for the dc-dc solar converter in buck mode, = ( ). For the intended, is the period of ac signal in seconds. equals 1/ for the solar converters operating at a specified switching frequency; is the number of turns in winding and can be specified by magnetic core selection and power inductor design procedure. These equations are used for the calculation of energy conversion efficiency analysis in section III. Table I shows the design of a 10.8mH inductor for 20kHz switching frequency experiment. Ferrite core EE65 is selected according to the value of. For 50 khz and 100 khz switching operation, EE40 core is selected due to smaller request of L. Inductance (L) TABLE I SPECIFICATION OF AN INDUCTOR DESIGN Core Type Number of Turns (N) 10.8 mh EE Total Length of Air Gap 5.0 mm (1.5 mm mm) AWG In the lab-manufactured inductor shown in Fig. 3, some air gaps are made in consideration of: a) Higher mmf can be tolerated without saturation; b) Reduced core losses, with higher quality factor (Q); c) Less sensitive in inductance affected by changing current and temperature; d) Multiple choices in core selection

3 Fig. 3. A power inductor of 10.8 mh manufactured for lab experiment. III. ENERGY CONVERSION EFFICIENCY ANALYSIS Firstly, the efficiency of 3kW SiC-based BoCBB converter is analyzed theoretically. The efficiency of Si-based BoCBB converter is also calculated and used as a baseline for comparison. Table II compares the characteristic parameters of SiC and Silicon MOSFETs. Tables III-VI show the results of power losses and conversion efficiency of the BoCBB DC-DC converter in the cases with SiC MOSFETs (model: CREE C2M028120D) and Si MOSFETs (model: Fairchild FQA8N100C) operating in 20 khz, 50 khz, and 100 khz in buck mode (V in > 400 Vdc) and boost mode (V in < 400 Vdc), respectively. TABLE II COMPARISON OF SIC AND SILICON MOSFETS PARAMETERS Parameters SiC MOSFET CREE C2M028120D Silicon MOSFET (Fairchild FQA8N100C) t fall 21.7 ns 202 ns t rise 12.8 ns 145 ns V F 3.3 V 1.4 V R DS(ON) 280 mω 1.45 Ω In buck mode, the boost switch (S boost) is consistently in OFF state. So there is no switching or conduction losses in S boost (in Fig.1). When the converter operates in the boost mode, the buck switch (S buck) is always in ON state. Thus, there is no switching loss and only conduction loss, which should be considered for S buck in boost operation. This characteristic in the BoCBB topology causes additional conduction loss, which makes it different from the η calculation of traditional buckboost power converters. Our case study reveals that this additional conduction loss causes an increase in overall power loss within the range of [15.5%, 25.8%], depending on switching frequency, in boost operation. Under different switching frequencies in MOSFET modulation, the changes in inductor value are taken into consideration and thus the N of the inductor changes, which contributes to copper loss. The conduction losses of Schottky diodes and MOSFETs are kindly independent of switching frequency because the fall and rise times are relatively very small compared to the complete switching period, which makes this error is negligible. For Schottky diodes, only conduction losses are considered and switching losses are negligible. The core losses in inductors are not affected by diodes and MOSFETs, since the ΔB is small. TABLE III POWER LOSS WITH SI MOSFET IN BUCK MODE (D = 0.9) Core losses 2.75*10 0.5*10 5.3*10 Copper losses Switching losses Conduction Buck losses side Diode losses Total Losses η (%) 97.35% 96.68% 95.59% TABLE IV POWER LOSS WITH SIC MOSFET IN BUCK MODE (D = 0.9) Core losses 2.75*10 0.5*10 5.3*10 Copper losses Switching losses Conduction Buck losses side Diode losses Total Losses η (%) 99.45% 99.4% 99.33% TABLE V POWER LOSS WITH SI MOSFET IN BOOST MODE (D = 0.1) Core losses 1.15* *10 1.6*10 Copper losses Switching losses Buck Conductio side n losses Boost side Diode losses Total Losses W η (%) 96.92% 96.52% 95.87% TABLE VI POWER LOSS WITH SIC MOSFET IN BOOST MODE (D = 0.1) Core losses 1.15* *10 1.6*10 Copper losses Switching losses Buck Conduction side losses Boost side Diode losses Total Losses η (%) 99.35% 99.31% 98.25% Fig. 4 indicates the η gain by replacing Si MOSFETs with SiC MOSFETs. It is noticed that, under the buck mode operation, the η gain keeps increasing. But under the boost mode operation, the η gain increases from 20 khz to 50 khz and then decreases from 50 khz to 100 khz. An inflection point appears at 50 khz in the curve and this is caused by less drop of η in SiC prototype. It means the effect of SiC devices on the improvement of η gradually reduce when the switching frequency is higher than 50 khz. 682

4 Fig. 4. The gain in η by replacing Si MOSFETs with SiC MOSFETs under different switching frequencies. IV. EXPERIMENT ANALYSIS The energy conversion efficiency analysis in section III is further validated by experimental tests. Fig. 5 is the schematic design of power circuit and driver circuit design for BoCBB prototype in Cadsoft Eagle. The prototype is shown in Fig. 6, and the testbed of a 3kW BoCBB prototype was set up as shown Fig. 7. A 4kW, 600V Magna dc power supply is controlled by a Magnapower Photovoltaic Power Profile Emulation 2.0 module to emulate the behavior of PV panel. Fig. 7. Testbed of BoCBB converter. Fig. 8 is the experimental waveforms of current ripples (in blue) and voltage ripple across filtering capacitor (in pink) in boost operation. The top orange waveforms are high switching signals modulated to control MOSFETs. For these cases, power inductors are 10.8 mh with EE65 core for 20 khz, 5.01 mh with EE40 core for 50 khz, and 2.6 mh with EE40 core for 100 khz. The duty cycle D is controlled within [0.1, 0.5] to regulate output voltage to 400 Vdc, depending on input DC voltage. Fig. 8 shows the waveforms for the cases of D = 0.5 and D = 0.1. (a) Switching frequency of 20 khz (b) Switching frequency of 50 khz Fig. 5. Schematic design of driver circuit. Fig. 6. PCB and prototype of BoCBB converter. (c) Switching frequency of 100 khz Fig. 8 Experimental waveforms of ripple current through L Boost and ripple voltage across C2, when duty cycle D = 0.5 (left) and D = 0.1 (right). 683

5 Fig. 9 is a comparison of energy conversion efficiency under different switching frequencies at 20 khz, 50kHz, and 100 khz, and duty cycles of PWM within [0.1, 0.5]. 3) The switching frequency of 50 khz is recommended for SiC-based power converters, which makes a good tradeoff between high energy conversion efficiency, good power quality, and compact size E_20kHz E_50kHz E_100kHz ACKNOWLEDGMENT This material is based upon work supported by the Air Force Civil Engineer Center (AFCEC) under Contract No. FA C Any opinions, findings and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the Air Force Civil Engineer Center (AFCEC). Approved by AFCEC/PA for public release and distribution. PA #201617, 14 Jul 16. Fig. 9. Comparison in η under switching frequency of 20 khz, 50 khz, and 100 khz and duty cycle within [0.1, 0.5]. V. CONCLUSION By performing theoretical analysis and experiment tests on the energy conversion efficiency of a 3kW SiC-based BoCBB power converters for solar power integration into 400Vdc distribution networks, we found that: 1) There is about 2% reduction in efficiency, when the switching frequency of MOSFETs increases from 20 khz to 100 khz in Si-based prototype. Comparatively, the efficiency reduction with SiC-based prototype is limited within 1% in the same case; 2) Under the same switching frequency, the efficiency can be increased by 2.4% in average by replacing Si MOSFETs with SiC MOSFETs. This difference increases along the switching frequency; REFERENCES [1] NREL website: last accessed on June 16th [2] Go Solar California website: last accessed on June 16th [3] International Energy Agency (2014), "Technology Roadmap: Solar Photovoltaic Energy" archived from the original on 7 October [4] Donald J. Hammerstrom, AC Versus DC Distribution Systems-Did We Get it Right?, Power Engineering Society General Meeting, 2007 IEEE, pp:1-5, Tampa, FL, June [5] Chien-Hsuan chang, Chun-An cheng, En-Chih Chang, and Hung-Liang Cheng. Design and implementation of a two-switch Buck-Boost typed inverter with universal and high-efficiency features, 9th international conference on power electronics-ecce Asia, pp: 1-5, June 2015, Seoul Korea. [6] J. Chen, D. Maksimovic, and R. W. Erickson, Analysis and design of a low-stress buck-boost converter in universal-input PFC applications, IEEE Trans. Power Electronics, Vol.21, No.2, pp: , March [7] N. Mohan, T. M. Undeland, and W. P. Robbins, Power electronics: converters, applications, and design, 2nd Ed. New York; Brisbane: J. Wiley, [8] Robert W. Erickson, Dragan Masksimovic. Fundamentals of Power Electronics second edition. ISBN: Kluwer Academic/Plenum Publishers, New York. 684