SiC-MOSFET Composite Boost Converter with 22 kw/l Power Density for Electric Vehicle Application

Transcription

1 SiC-MOSFET Composite Boost Converter with 22 kw/l Power Density for Electric Vehicle Application Hyeokjin Kim, Hua Chen, Dragan Maksimović, Robert Erickson Department of Electrical, Computer and Energy Engineering University of Colorado Boulder Boulder, Colorado, Zach Cole, Brandon Passmore, Kraig Olejniczak Wolfspeed, A Cree company Fayetteville, AR, United States Abstract A SiC-MOSFET composite boost converter for an electric vehicle power train application exhibits a volumetric power density of 22 kw/l and gravimetric power density of 20 kw/kg. The composite converter architecture, which is composed of partial-power boost, buck, and dual active bridge modules, leads to a 60% reduction in CAFE average losses, to a 280% improvement in power density, and to a 76% reduction in magnetics volume compared to the conventional Si-IGBT boost converter. These gains were achieved with the help of optimization based on a comprehensive loss model including SiC-MOSFET switching loss and magnetic losses based on the FEM method simulated in FEMM. Experimental results for the 22 kw/l SiC-MOSFET composite converter project 97.5% average efficiency on US06 driving cycle and a CAFE average efficiency of 97.8%. I. Introduction Reduction in losses of the power conversion unit (PCU) in electric vehicle (EV) or hybrid electric vehicle (HEV) powertrains translates directly into increased MPGe (Mile-per- Gallon equivalent) and downsizing of cooling capacity associated with thermal management. To increase peak efficiency and achieve high CAFE (corporate average fuel economy) average efficiency, the composite converter architecture has been introduced [1]. By utilization of pass-through modes of the composite converter and optimization by reallocating silicon semi-conductor die and magnetics, the composite converter achieved 98.1% CAFE average efficiency. While the composite converter achieves CAFE average loss reduction by a factor of three relative to the conventional Si-IGBT boost converter [2], volume reduction in magnetics is challenging owing to the low switching frequency of the optimum design. One approach associated with this magnetic volume reduction without loss of efficiency involves wide band-gap devices such as SiC- MOSFET to utilize their superior switching performances [3]. In the case of boost converter employed in EV powertrain application, even though the wide band-gap devices enables the reduction in magnetics and improvement on peak efficiency, the low efficiency is still observed under the operating points requiring high voltage conversion ratio owing to high magnetic loss. Also large volume of capacitors are necessary to meet maximum voltage requirement and peak RMS capacitor current. As a result, SiC-MOSFET conventional boost converter exhibits low average efficiencies on EPA standard driving cycles, relative to the Si-MOSFET composite boost converter, and employs similar volume of capacitors, compared to the conventional Si-IGBT boost converter. In this paper, 22 kw/l volumetric power density and 97.8% CAFE average efficiency are demonstrated in a SiC-MOSFET composite boost converter. The composite boost converter architecture achieves improvement in peak efficiency and high average efficiencies on EPA standard driving cycles such as US06, UDDS, or HWFET compared to the conventional boost converter so that the improvement on CAFE average efficiency is enabled. Compared to the conventional Si-IGBT boost converter whose power density is reported as 5.7 kw/l [2], the SiC-MOSFET composite converter achieves a 76% magnetic volume reduction, a 280% power density improvement, and a 60% CAFE average loss reduction. The composite boost converter architecture exhibits the reduction in the capacitor RMS current rating and voltage rating which directly translates into a capacitor volume, compared to the conventional boost converter. Also the magnetic volume reduction is demonstrated through the superior switching capability of SiC-MOSFET device. With the SiC-MOSFET switching device and composite boost converter, the volume reductions in capacitor and magnetic are enabled and this achievement is explained in Chapter II. To facilitate design optimization, the comprehensive loss model including switching loss and magnetic loss is developed and the loss model is explained in Chapter III. With the developed comprehensive loss model, the design of SiC-MOSFET composite boost converter and the design result are explained in Chapter IV. Based on the design result, the laboratory prototype board of 22 kw/l volumetric power density is fabricated and experimental results are demonstrated in chapter V.

2 II. Size reduction of passive components The conventional boost converter employed in EV or HEV powertrain is comprised of capacitors, magnetics, semiconductor devices, and peripheral circuits such as gate driver or sensor circuitries. The capacitors and magnetics occupy significant part of boost converter module volume [2] and the capacitor volume is proportional to the voltage rating and peak magnitude of capacitor root-mean-square(rms) current. In the case of conventional boost converter, the magnitude of capacitor RMS current is proportional to the load power and voltage conversion ratio of input to output. Based on the assumption that the rated power can be delivered to the load over the all range of operating point, the magnitude of capacitor RMS current is maximized when voltage conversion ratio is two while the rated power is delivered to the load [4]. The RMS current rating and capacitance per volume versus voltage rating of Metallized Polypropylene Film Capacitors (Film capacitor) is shown in Fig. 1. As the voltage rating of capacitor is higher, the RMS current capability and capacitance per volume is degraded so that higher voltage rating necessitates larger volume of capacitor to meet the requirement of RMS current rating or capacitance RMS[A]/Vol[mm 3 ] C[uF]/Vol[mm 3 ] Vrating [V] Fig. 1. Capacitor RMS current rating per volume and capacitance per volume as function of voltage rating The boost converter employed in EV or HEV powertrain is required to operate wide range of power and voltage conversion ratio, however, the rated power is delivered to the load at high motor speed region. Since the motor shaft power is the function of torque and angular speed but maximum torque is delivered only at low speed region, the rated power can be delivered at high speed region. Based on the assumption that 250V of battery is employed and the maximum inverter DC bus voltage is 800V, the motor torque, load power and required DC bus voltage for the inverter as function of motor speed resulted from the 30kW rated EV powertrain model [5] is shown in Fig. 2. The required bus voltage is obtained from following equation to achieve minimum converter and inverter losses Torque [Nm] Vbus [10V] RPM Fig. 2. Motor torque, shaft power, and required inverter DC bus voltage as function of motor RPM based on 30kW EV powertrain model V bus,re f = α(vds 2 + V2 qs) (1) V ds and V qs are the voltage of d and q stationary axis, and α is the number to achieve minimum converter and inverter losses while avoid field weakening control. In this case, α is taken to be 3. At the low voltage conversion ratio of boost converter, the maximum deliverable power is proportional to the motor RPM. This load condition such as EV powertrain causes the capacitor RMS current magnitude of boost converter proportional to the voltage conversion ratio owing to low peak power at low voltage conversion ratio. With this load condition, the peak output capacitor RMS current and peak output power of boost converter versus motor RPM at 250 battery voltage is shown in Fig. 3. The peak RMS current is observed at high voltage conversion ratio of battery to bus. Assuming that a conventional boost converter is designed to meet the specifications shown in Fig. 2 and 3, the voltage rating of output capacitor must be higher than maximum DC bus voltage, 800V, and capacitor RMS current rating must be higher than peak RMS current, 64A. With the composite boost converter architecture [1] shown in Fig. 4, the capacitor RMS current and voltage rating is reduced. The composite boost converter architecture consists of three dissimilar converter modules, buck, boost, and dual active bridge (DAB) operated as DC transformer (DCX). Since the output voltage of boost module employed in composite boost converter is operated within 400V and the boost module processes partial system power, voltage rating and peak capacitor RMS current are reduced. Based on the RMS current capability per volume as function of voltage rating shown

3 Efficiency [%] However, the magnetic volume of the Si-MOSFET composite converter is essentially unchanged, compared to the conventional boost converter owing to the low switching frequency that occurs in the optimized design [5]. To reduce magnetic volume employed in half bridge converter, the SiC- MOSFET device can be employed with the high switching frequency without loss of efficiency owing to superior switching performance compared to the Si-MOSFET or Si-IGBT device. The efficiency comparison of SiC-MOSFET composite, Si-MOSFET composite, SiC-MOSFET conventional, and Si-IGBT conventional boost converter at 250 V in, 650 V bus versus power is shown in Fig. 5. Compared to the Si-IGBT conventional boost converter, the SiC-MOSFET conventional boost converter achieves a reduction in losses over a wide range of operating points, particularly at low power and the resulting magnetic volume is reduced by 60%. Nonetheless, Capacitor RMS current [A] Vbus [V] Fig. 3. Peak RMS current of output capacitor employed in boost converter and load power as function of output voltage based on 30kW EV powertrain in Fig. 1, the volume reduction in output capacitor of boost module is 41%, relative to the output capacitor of conventional boost converter. Even though the extra capacitors are necessary for DAB module input and output, the peak magnitude of capacitor RMS current of DAB converter is very small [6] owing to the nature of DAB operation under the voltage conversion ratio of input to output close to the transformer turns ratio. The resulting net capacitor volume of composite boost converter is reduced by a factor of 1.4, compared to the net capacitor volume of conventional boost converter. V IN Buck Mbuck(D) DCX 1:NDCX Boost Mboost(D) Fig. 4. Composite boost converter architecture [1] + V BUS SiC-MOSFET Composite boost Si-MOSFET Composite boost SiC-MOSFET Conv. boost Si-IGBT Conv. boost Power / Prated Fig. 5. Efficiency comparison of SiC-MOSFET composite, Si-MOSFET composite, SiC-MOSFET conventional, and Si-IGBT conventional boost converter at 250 V in, 650 V bus versus power the efficiency of the SiC-MOSFET conventional boost converter is inferior to the Si-MOSFET composite converter, owing to high magnetics losses at operating points requiring high voltage conversion ratio. In this paper, high efficiency and high power density are demonstrated with SiC-MOSFET composite boost converter. Through the composite converter architecture and SiC-MOSFET switching device, capacitor and magnetic volumes are reduced, and high average efficiencies on EPA standard driving cycles are achieved, compared to the conventional boost converter. III. Comprehensive loss model In this paper, the SiC-MOSFET composite boost converter is designed to demonstrate not only high power density, but also high average efficiency on US06 driving cycle. To facilitate converter module design optimization, a comprehensive loss model is developed. The model includes semiconductor loss comprised of switching and conduction losses, and magnetic loss including DC, AC copper winding losses and core loss. A. Semiconductor loss Piecewise linear (PWL) function model [7] is employed for switching loss calculation. The advantage of PWL switching

4 Efficiency [%] Efficiency [%] 99 loss model is the use of parameters listed in manufacturer s datasheet and parameters of gate-driver circuit so that available switching devices can be evaluated. The developed PWL switching loss model is based on following assumptions: The switching loss is negligible when the switching device is operated under zero-voltage switching (ZVS). Parasitic circuit inductance and any ringing during switching transition are neglected. Based on the PWL switching loss model, instantaneous loss of each switching interval is calculated and average switching loss over one switching period is estimated. Also, the conduction loss of semiconductor can be expressed P conduction = R DS I RMS 2 R DS is the on-state resistance of MOSFET and I RMS is the RMS current flowing through the MOSFET. To verify the semiconductor loss model, a prototype board configured as boost converter is fabricated using Cree 900V/10mΩ SiC- MOSFET [8] packaged in HT-4000 module [9] which is shown in Fig. 6. The comparison of measured efficiency (2) Measured efficiency Loss model Fig. 7. SiC-MOSFET boost module loss model (red dashed line) and measured efficiency (black dots) at 200V in and 208V out as function of power Fig. 6. Photo of Cree HT-4000 SiC-MOSFET full bridge package and US quarter coin and switching loss model efficiency is shown in Fig. 7 and 8 at operating point of 200V in /209V out and 300V in /313V out with 200kHz of switching frequency. Under the voltage conversion ratio close to unity, the switching loss dominates the converter loss, while magnetic loss is negligible and conduction loss can be easily predicted. The developed loss model shows good agreement with measured data over wide range of power and voltage. B. Magnetic loss model The magnetic loss model consists of DC and AC copper winding loss, and core loss. Core loss is calculated according to the igse method [10]. DC copper winding loss is proportional to the square of the DC winding current. For the AC winding copper loss, 2D FEM simulation tool, FEMM [11], is employed to estimate the loss resulted from skin, proximity, and fringing effect. The magnetic loss is measured Measured efficiency Loss model Fig. 8. SiC-MOSFET boost module loss model (red dashed line) and measured efficiency (black dots) 300V in and 313V out as function of power with the prototype converter configured as a boost converter under 240kHz of switching frequency and high voltage conversion ratio with light load. Under this operating condition, the semiconductor is operated under zero-voltage switching so that the switching loss is negligible and magnetic loss dominates the converter loss. The comparison of measured efficiency and loss model efficiency is shown in Fig. 9 at four different operating conditions, 100V in /146V out, 150V in /220V out, 200V in /294V out, and 250V in /368V out. The magnetic loss model shows good agreement with measured data.

5 Efficiency [%] Vin / 146Vout 150Vin / 220Vout 200Vin / 294Vout 250Vin / 368Vout Fig. 10. US06 driving cycle density plot of operating points [5] Fig. 9. Magnetic loss model and measured efficiency of boost converter at 100V in /146V out, 150V in /220V out, 200V in /294V out, 250V in /368V out as function of power. Semiconductor loss is negligible due to ZVS and light load IV. Design of SiC-MOSFET composite boost converter Based on the comprehensive loss model discussed in section III, the composite boost converter is designed to demonstrate high volumetric and gravimetric power density, and high US06 average efficiency as well. The specifications of input and output of boost converter are listed in Table I. TABLE I SiC-MOSFET composite boost converter specification should be taken account. However, this brute force method takes prohibitively large amount of computational effort. To reduce a computational effort without loss of optimization result accuracy, the weighted loss method [5] is employed. For the buck and boost modules, interleaving control scheme is employed to reduce magnetic volume and capacitor RMS current which directly translates into capacitor volume. As a result, the capacitor volume is reduced by 17% and the magnetic volume is reduced by 25%, compared to the result of non-interleaved design. The optimization results are listed in Table II. Input voltage DC bus voltage V 800V maximum TABLE II SiC-MOSFET composite converter magnetics design summary With the 900V/10mΩ SiC-MOSFET module consisting of two half-bridge circuitries, an optimum parameter set including switching frequency and magnetics are optimized. To facilitate minimum average loss over US06 driving cycle, EV powertrain simulation model is developed based on the Nissan LEAF vehicle. The required motor power and bus voltage are extracted from the simulation result and the resulting density plot of US06 driving cycle is shown in Fig. 10 on the normalized motor power versus required bus voltage plane. Higher frequency counts of operating points are represented by darker shadings. Unlike the required bus voltage distributed over the range between boost converter input voltage and maximum bus voltage, the load power is distributed over the range of power less than 50% of system power. This tendency can be observed on UDDS or HWFET driving cycles as well [5]. Therefore, a boost converter with high efficiency at light load is able to improve average efficiency on not only US06 driving, but also other EPA standard driving cycles. To optimize a boost converter over US06 driving cycle, all operating points Buck / Boost module Switching frequency 240 khz Inductance 5.2 µh for Boost, 3.4 µh for Buck Inductor core size and material Two ferrite, EILP 43 DCX module Switching frequency 240 khz Transformer turns ratio 4:6 Transformer core size and material Ferrite, EILP 64 For capacitor design, single or multiple capacitors can be configured to meet the peak magnitude of RMS current. To minimize the volume of composite boost converter module and also meet the peak capacitor RMS current, an exhaustive search script is developed on Matlab to find an optimum set of capacitors. The TDK s Metallized Polypropylene Film capacitors are incorporated into the parameter library and resulting capacitors are listed in Table. III. The optimized design of SiC-MOSFET composite boost converter achieves a peak power of 39kW with a predicted total volume of 1.8 liters.

6 TABLE III SiC-MOSFET composite converter capacitors design summary Input Boost output DAB input DAB output V. Experimental results 2x 6.8µF / 450V 3x 3.3µF / 630V 2x 5.0µF / 450V 1x 4.7µF / 630V Based on the converter optimization summarized in Section IV, the SiC-MOSFET composite boost converter shown in Fig. 11 has been fabricated and tested. The prototype board consists of driver board, power board, SiC-MOSFET module, coldplate, capacitors, and magnetics. The 3-dimensional exploded view of SiC-MOSFET composite boost converter is shown in Fig. 12. To reduce resistive conductor loss of PCB trace, power board utilizes heavy copper traces, while standard copper traces are employed on driver board to utilize the small monolithic ICs. DAB primary and secondary, boost, and buck magnetic current at 250V in /650V bus with 50W. Under this light load condition, switches of DAB primary and secondary achieves ZVS through the resonant phenomenon between the output capacitor of MOSFET and magnetizing inductance of transformer so that the switching loss of DAB is minimized. Also, the buck module is operated under the pass-through mode. As a result, the switching loss and magnetic loss of buck module are eliminated. Itx(dcx/sec) Vsw(dcx/sec) Vsw(dcx/pri) Itx(dcx/pri) Vsw(boost) IL(boost) Vsw(buck) IL(buck) Fig. 11. SiC-MOSFET composite boost converter prototype board. Top view (Left), bottom view with US $1 bank bill (Right) SiC-MOSFET module Magnetics Cold-plate Capacitors Power PCB Driver PCB Fig D drawing of SiC-MOSFET composite boost converter composed of driver board, power board, cold-plate, capacitors, and magnetics Fig. 13 shows the measured waveforms of DAB primary and secondary, boost and buck switching node voltages, and Fig. 13. DAB primary and secondary switching node voltages and transformer primary and secondary currents, and boost and buck switching node voltages and inductor currents at 250V in /650V bus with 50W Also, the measured waveforms including interleaved scheme of boost module at 250V in /650V bus with 12kW is shown in Fig. 14. Since the DAB is operated under the voltage conversion ratio close to the transformer turns ratio, the AC copper winding loss of magnetic is minimized over wide operating range. The operating condition and estimated losses of individual modules, and resulting efficiencies of composite boost converter, and conventional Si-IGBT boost converter at the operating points shown in Fig. 13 and 14 are listed in Table. IV. The comprehensive loss model shows good agreement with measured data. Compared to the Si-IGBT conventional boost converter [2], the improvement on efficiency at light load is demonstrated with composite boost converter. This high efficiency on light load leads substantial improvement on average efficiencies over EPA standard driving cycles. The comparison of measured efficiency, the loss model efficiency, and conventional SiC-MOSFET boost converter at the operating point of 250V in /650V bus as function of power is shown in Fig. 15. Compared to the conventional boost

7 Efficiency [%] Vsw(dcx/pri) Vsw(dcx/sec) Itx(dcx/sec) Itx(dcx/pri) converter, the composite boost converter achieves the improvement on efficiency under light load and maintains high efficiency over wide range of power Vsw(boost) IL(boost) Vsw(buck) IL(buck) Vsw(boost-1) IL(boost-1) Vsw(boost-2) Comprehensive loss model Conventional Si-IGBT boost Measured effciency Fig. 15. Comprehensive loss model efficiency, measured efficiency of SiC- MOSFET composite boost converter and Si-IGBT conventional boost converter efficiency IL(boost-2) Fig. 14. DAB primary and secondary switching node voltages and transformer primary and secondary currents, boost and buck switching node voltages and inductor currents, and boost module 1 and boost module 2 switching node voltages and inductor currents at 250V in /650V bus with 12kW TABLE IV Operating conditions and estimated losses of individual module, and resulting efficiency of composite boost converter and conventional boost converter [2] Operating point 250V in /650V bus 250V in /650V bus 50 W 12 kw Boost operating condition 250V in /275V out /21W 250V in /275V out /5kW Boost loss 11W 94W Buck operating condition 250V in /250V out /29W 250V in /250V out /7kW Buck loss 0W 5W DAB operating condition 250V in /375V out /29W 250V in /375V out /7kW DCX loss 18W 111W Net loss 29W 210W Estimated efficiency 63.3% 98.2% Measured efficiency 66.2% 97.3% Conv. boost op. 250V in /650V bus /50W 250V in /650V bus /12kW Conv. boost loss 337W 463W Conv. boost efficiency 12.9% 96.3% The comparison of estimated average EPA standard driving cycles efficiency and CAFE (Corporate Average Fuel Economy) average efficiency and quality factor, Q = P P out / P loss, switching frequency, and magnetic volume of conventional boost converter with Si-IGBT [2] or SiC- MOSFET, and composite boost converter with Si-MOSFET [5] or SiC-MOSFET are listed in Table. V. For higher quality factor, Q, either higher output power can be achieved without modification of cooling capacity, or reduction in cooling capacity associated with manufacturing cost can be enabled without degradation of output power. Compared to the conventional Si-IGBT or SiC-MOSFET boost converter, composite boost converter including Si-MOSFET or SiC-MOSFET exhibits the improvement on average efficiencies on EPA standard driving cycles. Also, the reduction in magnetic volume is demonstrated with SiC-MOSFET composite boost converter. VI. Conclusions This paper is focused on the design of a high power density and high average efficiency on US06 driving cycle SiC-MOSFET composite boost converter. The composite boost converter achieves 22 kw/l volumetric power density, as well as 97.5% average efficiency on US06 driving cycle and 97.8% CAFE averaged efficiency as well. Relative to the conventional 5.7 kw/l Si-IGBT converter, magnetics volume is reduced by 76%, power density is improved by 280%, and CAFE average loss is reduced by 60%. Also, CAFE average loss reduction by a factor of 1.8 is achieved, compared to the CAFE average loss of SiC-MOSFET conventional boost

8 Converter TABLE V Comparison of switching frequency, US06, CAFE average efficiency, converter quality factor (Q) and magnetic volume Si-IGBT Si-MOSFET SiC-MOSFET SiC-MOSFET Conv. boost [2] Composite boost [1] Conv. boost Composite boost Switching frequency 10 khz 20 khz 240 khz 240 khz US06 average efficiency 93.3% 98.3% 96.8% 97.5% UDDS average efficiency 97.1% 98.2% 97.7% 98.0% HWFET average efficiency 91.8% 98.0% 94.1% 97.6% CAFE average efficiency 94.7% 98.1% 96.1% 97.8% CAFE Q factor Magnetic volume [ml] converter. For the design optimization, a comprehensive loss model is developed. The complete comprehensive loss model is experimentally verified through a prototype SiC-MOSFET boost module and shows good agreement with measure data. To optimize SiC-MOSFET composite boost on US06 driving cycle, weighted loss method is employed and the resulting converter leads to a predicted 39 kw peak power and volume of 1.8 liters. Prototype board is fabricated and experimental results shows the improvement on efficiency over wide range of power and voltage, relative to the conventional Si-IGBT and SiC-MOSFET boost converter. Furthermore, the reduction in loss at light load is achieved so that improvement on average efficiencies on EPA standard driving cycles is enabled. The prototype SiC-MOSFET composite boost converter exhibits 22 kw/l of volumetric power density and 20 kw/kg of gravimetric power density. [8] V. Pala, G. Wang, B. Hull, S. Allen, J. Casady, and J. Palmour, Recordlow 10mΩ sic mosfets in to-247, rated at 900v, in 2016 IEEE Applied Power Electronics Conference and Exposition (APEC), March 2016, pp [9] Z. Cole, B. McGee, J. Stabach, S. Storkov, B. Whitaker, D. Martin, G. Falling, R. Shaw, and B. Passmore, High temperature, wide bandgap full-bridge power modules for high frequency applications, in Integrated Power Packaging (IWIPP), 2015 IEEE International Workshop on, May 2015, pp [10] V. Kapil, R. S. Charles, A. Tarek, and T. Hernan, Accurate prediction of ferrite core loss with nonsinusoidal waveforms using only steinmetz parameters, in Computers in Power Electronics (COMPEL), Proceedings IEEE Workshop on, June 2002, pp [11] Finite element method magnetics. [Online]. Available: Acknowledgement The information, data, or work presented herein was funded in part by the Office of Energy Efficiency and Renewable Energy (EERE), U.S. Department of Energy, under Award Number DE-EE References [1] H. Chen, K. Sabi, H. Kim, T. Harada, R. Erickson, and D. Maksimovic, A 98.7% efficient composite converter architecture with applicationtailored efficiency characteristic, Power Electronics, IEEE Transactions on, vol. 31, no. 1, pp , [2] T. Burress, Evaluation of the 2010 Toyota Prius hybrid synergy drive system, Oak Ridge National Laboratory (ORNL); Power Electronics and Electric Machinery Research Facility, Technical Report ORNL/TM- 2010/253. [3] T. Zhao, J. Wang, A. Q. Huang, and A. Agarwal, Comparisons of SiC-MOSFET and Si-IGBT based motor drive systems, in Industry Applications Conference, nd IAS Annual Meeting. Conference Record of the 2007 IEEE, 2007, pp [4] R. W. Erickson and D. Maksimovic, Fundamentals of Power Electronics. Springer, [5] H. Kim, H. Chen, D. Maksimovic, and R. Erickson, Boost composite converter design based on drive cycle weighted losses in electric vehicle powertrain applications, in Energy Conversion Congress and Exposition (ECCE), 2016 IEEE, Sept [6] D. Costinett, D. Maksimovic, and R. Zane, Design and control for high efficiency in high step-down dual active bridge converters operating at high switching frequency, Power Electronics, IEEE Transactions on, vol. 28, no. 8, pp , [7] H. Kim, H. Chen, D. Maksimovic, and R. Erickson, Design of a high efficiency 30 kw boost composite converter, in Energy Conversion Congress and Exposition (ECCE), 2015 IEEE, Sept 2015, pp