High-voltage High-frequency Transformer Design for a 7.2kV to 120V/240V 20kVA Solid State Transformer Yu Du, Seunghun Baek, Subhashish Bhattacharya, Alex Q. Huang FREEDM Systems Center, Department of Electrical and Computer Engineering, North Carolina State University 1791 Varsity Drive, Suite 100, Keystone Science Center, Raleigh, NC, 27695, U.S.A.
[email protected] Abstract--Solid state transformer (SST) exhibits good features such as high power density, small volume and weight, controlled power factor, voltage sag ride through, etc. compared with traditional line frequency transformer. The 7.2kV AC to 120V/240V AC 20kVA solid state transformer is a key component of the future renewable electric energy delivery and management (FREEDM) systems as the interface between the 7.2kV distribution grid and the low voltage residential micro-grid. Three cascaded 6.7kVA high-voltage high-frequency transformers operating at 3kHz are employed to convert voltage from 3800V high voltage DC link of each cascaded stage to 400V low voltage DC link. The transformer is required to withstand at least 15kV high frequency voltage insulation continuously. Transformer magnetic core materials were reviewed and compared. Winding layout alternatives for leakage, magnetizing inductance and insulation were compared. An insulation strategy based on split core and separate winding structure with inserted insulation layer between the C cores was proposed. One 6.7kVA high voltage high frequency transformer prototype was built and the test results were reported.
parallel to provide a low voltage DC link for the third stage, which is a three-leg two-phase inverter. The transformer in each stage provides voltage transfer between 3.8kV in primary side high voltage DC link and 400V in secondary side low voltage DC link. The switching frequency of dual active bridge (DAB) DC-DC converter is 3 kHz in Gen-I SST due to the limit of the frequency capability of 6.5kV silicon IGBT’s. The specification for the transformer in SST is summarized in Table I.
I. INTRODUCTION The solid state transformer is potentially beneficial for future power system. For example, with increased residential electric load capacity, the power rating of distribution transformer should be improved. However, there is limited underground space to accommodate a larger transformer. So increased power density of distribution transformers is a good option to deliver more power [1]. Solid state transformer (SST) also exhibits good features such as high power density, small volume and weight, controlled power factor, voltage sag ride through, etc. compared to traditional line frequency transformer [2-3]. Due to environment and safety consideration, oil free or dry type insulated high frequency transformer is highly preferred for solid state transformer and is often considered as one advantage of SST. The topology of a 7.2kV to 120/240V 20kVA FREEDM Gen-I solid state transformer (SST) is shown in Fig.1 [4]. The input is single phase 7.2kV AC line from the distribution grid, and the output is 120V/240V two phase AC terminals with a neutral line. The high voltage side AC-DC stage consists of three cascaded full bridges. The AC-DC stage is then followed by the DCDC stage of three dual active bridges with output in 978-1-4244-5226-2/10/$26.00 ©2010 IEEE
Fig.1: The topology of the 7.2kV to 120V/240V 20kVA Gen-I Solid State Transformer (SST) TABLE I: SPECIFICATIONS OF THE HIGH VOLTAGE HIGH FREQUENCY TRANSFORMER IN GEN-I SST Number of transformers Switching frequency fs Bi-directional DC-DC topology High voltage DC link VH Low voltage DC link VL Power rating per transformer Minimal insulation voltage Max. allowed temperature rise
3 3kHz Dual active bridges (DAB) 3800V 400V 6.7kVA 15kV (continuous) 55°C (nature air cooling)
There are three major challenges to design this highvoltage and high-frequency transformer. Due to the 3kHz switching frequency, the required leakage inductance for optimal operation of DAB is about 36mH, which is referred to the high voltage side, if no external inductor is needed. The second challenge is to provide a sufficient magnetizing inductance; otherwise the amplitude of magnetizing current is comparable to load current and will cause additional power loss. The third challenge is the insulation design since oil-free or dry type insulation 487 is required in a compact structure. These issues are
addressed in the following sections. But before discussing these trade offs, several high frequency soft magnetic materials were reviewed and compared for the solid state transformer application. II. MAGNETIC CORE MATERIAL SELECTION Several types of magnetic core materials can be considered for the high frequency transformer in SST applications, as shown in Table II. TABLE II: SEVERAL HIGH FREQUENCY MAGNETIC CORE MATERIALS UNDER CONSIDERATION Saturation Strip Material Series Density (T) Thickness Permeability @ 25°C (um) Silicon Steel Arnold Arnon5 1.48 127 High Amorphous Metglas 2605SA1 1.56 25 High Ferrite Ferroxcube 3C96 0.44 N/A Medium Ferrite Epcos N97 0.41 N/A Medium NanoHitachi Finemet 1.23 18 High crystalline FT-3M NanoVAC 1.20 N/A High crystalline Vitroperm500F NanoMagmet Namglass 1.23 25 High crystalline 4
One of the important considerations for high frequency transformer design is the core loss. A widely used calculation of specific core loss is Steinmetz equation,
Pcv = K ⋅ f α ⋅ Bacβ
(1)
where Bac is the peak flux amplitude, Pcv is the timeaverage power loss per unit volume, and f is the frequency of sinusoidal excitation, and K, α, and β are constants found by curve fitting [5]. The Steinmetz equation coefficients were extracted for the magnetic core materials listed in Table II based on the core loss graph from manufacturer datasheets. These coefficients are listed in Table III and the specific core loss at 3kHz with 0.3T peak flux density are also compared in this table, where the unit for Pcv is mW/cm3, and f and Bac use kHz and Tesla respectively. TABLE III: EXTRACTED STEINMETZ EQUATION COEFFICIENTS FOR DIFFERENT MAGNETIC CORE MATERIALS
K 3 (mW/cm ) Silicon Steel 278.4 Amorphous 46.7 Ferrite A 42.8 Ferrite B 44.0 Nanocrystalline A 8.03 Nanocrystalline B 2.48 Nanocrystalline C 3.75 Core Materials
α
β
1.39 1.51 1.53 1.36 1.62 1.80 1.71
1.80 1.74 2.98 2.72 1.98 2.08 1.97
Loss at 3kHz 0.3T (mW/cm3) 147 30.2 6.3 7.4 4.39 1.47 2.30
One typical soft magnetic core material used from several tens kilo-hertz to several mega-hertz range for high-frequency high-voltage transformers is ferrite [6-8]. It can be found in Table III that the ferrite core loss is low at 3kHz. And the cost of ferrite core materials is also relatively low. However, there are two major disadvantages for ferrite materials. The first one is the low saturation flux density. At room temperature, it is around 0.4T and it becomes worse when core temperature increases. This can potentially result in a larger core size
at 3kHz frequency due to the limited peak flux density. The second disadvantage is that ferrite cores are brittle. The size of the high voltage high frequency transformer in SST is not compact since the switching frequency is only 3kHz with 6.5kV silicon IGBT’s. The mechanical strength of ferrite might not be strong enough to support the structure. Therefore, the ferrite core materials are not considered here. Silicon steel materials have high saturation flux density (about 1.5T) and high permeability. Some manufactures provide laminations with gauge down to 1mil or 25um. However, compared with other transformer magnetic materials such as ferrite, nanocrystalline and amorphous cores, their specific loss is relatively high at 3kHz. This will result in a larger core size with small peak flux density or result in reduced efficiency with high flux amplitude and smaller core size. The nanocrystalline core materials are very promising for high frequency transformers. The nanocrystalline core is metallic tape-wound core made of nanocrystalline soft magnetic material. It exhibits high saturation flux density (1.2T) and extremely low specific loss up to about 100 kilo hertz frequency. Table III shows the specific core loss from several nanocrystalline core manufacturers, which exhibit lowest core loss at 3kHz frequency. On the other hand, 1.2T saturation flux density allows the selection of high operating peak flux amplitude. Therefore, with nanocrystalline core materials, the high frequency transformer can potentially be designed to be extremely compact and efficient at 3kHz. However, there are also several disadvantages for nanocrystalline cores. The first one is the relatively high material cost. Secondly, although special shapes in oval or rectangular design with or without cut can be produced, the standard off-the-shelf core shape is toroidal uncut tape-wound core. The leakage inductance is small due to the toroidal geometry and low number of turns. While the required leakage inductance is quite large since the DAB switching frequency is 3kHz. Another option is amorphous materials. Amorphous cores also exhibit high saturation induction (1.56T) and high permeability. High peak flux density can potentially be selected at 3kHz as well. The core is laminated with 1mil amorphous tape. The specific core loss is several times lower than silicon steel but still higher than that of nanocrystalline and ferrite cores. However, the cost of amorphous core is relatively low and the performance to cost factor is excellent at 3kHz frequency for this SST transformer. The mechanical strength is good enough to build larger transformers. In addition, cut C-core with large geometry and I-core formers are commercially available, offering more design flexibilities. Therefore, amorphous core material is selected for the 6.5kV silicon IGBT based SST transformer. The B-H curve of selected amorphous core is shown in Fig.2. The magnetic core material used in this project is Metglas® amorphous Alloy 2605SA1. The selected core geometry parameters are listed in Fig.3. It can be found that the core fill factor is 82%. The core is AMCC-1000 of Powerlite series from Metglas® and it is the largest 488
out-of-shell amorphous core. Several pairs of C cores can be paralleled to provide sufficient core cross section area.
determined by (3) with d=1 to minimize the reactive power in the circuit and maximize the zero voltage switching (ZVS) range at light load condition [10]. N H VH 3800 (3) = = = 9.5 NL
VL
400
The number of turns can be calculated by (4), NH =
10 4 ⋅ VH 4 ⋅ Bac ⋅ f s ⋅ Ac
(4)
Fig. 2: B-H curve of selected amorphous core material 2605SA1 [9]
Fig.3: Geometry of selected C-C core pair AMCC1000 [9]
The Steinmetz equation coefficients of specific core loss formula of U-cores AMCC1000 under square-wave input voltage are recalculated by collecting power loss data in terms of frequency f and peak flux density Bac because the Steinmetz equation provided on datasheet is the test result from no-cut toroidal cores under sinusoidal voltage excitation. Square-wave voltage inputs are injected on one winding while the other winding is open and the power losses are carefully measured. The coefficients of the formula are achieved by curve-fitting power loss data, as shown in Table IV. TABLE IV: COEFFICIENTS OF SPECIFIC CORE LOSS OF 2605SA1 3
K (mW/cm ) Datasheet (sinusoidal) Measurement (square input)
α
β
Pcv at 3kHz, 0.25T
46.7
1.51 1.74
21.99 (mW/cm3)
106.3
1.19 1.97
25.61(mW/cm3)
III. TRANSFORMER DESIGN TRADE OFF A. Selection of Transformer Turns Ratio The selection of the transformer turns ratio and required leakage inductance has great impact on the performance of dual active bridges (DAB) DC-DC converter. For example, from DAB conduction loss point of view, the reactive power in the circuit should be minimized. The ratio of the reactive power Q to output power P for DAB converter is shown in Fig.4, where the reactive power is defined by the product of rms current through the leakage inductance and rms voltage across it. The parameter d is defined as, V ⋅N (2) d= H L VL ⋅ N H where VH is high voltage DC-link voltage, VL is low voltage DC-link voltage, NH is the number of turns for high voltage winding and NL is the number of turns for low voltage winding. Based on Fig.4, the transformer turns ratio is
Fig.4: The ratio of reactive power to output power in dual active bridges (DAB) DC-DC converter
B. Major Design Considerations There are several challenges to design this high voltage and high frequency transformer. The first one is to provide sufficient leakage inductance such that no external inductor is required in dual active bridges DCDC converter. The required series inductance for a dual active bridge converter can be calculated in (5): Ls =
VH 2 ⋅ d ⋅ π 2 ⋅ f s ⋅ Po
⋅ Φ ⋅ (π − Φ )
(5)
where Ls is the required leakage inductance (referred to high voltage side); VH is input voltage (3.8kV); d is defined in (2) and equal to 1 here; fs is switching frequency (3kHz); Po is output power of DAB (6.7kW) and Φ is the phase-shift angle between two H-bridges [10]. From Fig.4, the smaller the phase-shift angle is, the lower the ratio of reactive power can be obtained when d is equal to 1. However, practically the neither transformer turns ratio nor the voltage regulation can be controlled very accurately. 10% variation of d (d=0.9 or d=1.1) can result in otherwise increased reactive power ratio with small phase-shift angles. One suggested maximal phaseshift angle which is corresponding to 6.7kW full load is 20° or π/9. Therefore, the DAB requires an inductance of 36mH as referred to the high voltage side for optimal operation. It is highly preferred that the inductance is provided by the leakage inductance of the transformer or it can be integrated into the transformer for lower volume, weight and cost. It should be noted that in this scenario, the leakage inductance should be increased since the input voltage is high and switching frequency and output power are relatively low, instead of minimizing the leakage inductance in some low voltage, high frequency and high power applications [11]. The second challenge is to have high magnetizing 489 inductance; otherwise the magnetizing current is
comparable to load current and will increase the power stage loss. At 3kHz frequency, 250mH magnetizing inductance, which is referred to primary side, is required to limit the magnetizing current lower than one-forth of the primary winding full load current. On the other hand, this level of magnetizing current will potentially help the ZVS of switches at light load condition [10]. The third challenge is high voltage insulation. Transformer oil is used for conventional high voltage transformer as insulation and cooling medium. This might cause environmental, safety and maintenance issues, especially for distribution transformer. So oil free or dry type high frequency transformer is highly preferred for SST. But it is difficult for insulation design in such a small core window area with air insulation, which requires large air clearance distance that is comparable to the core window geometry. In addition, the winding loss or winding AC resistance should be considered. The winding AC resistance can be predicted by Dowell’s model [13]. When the number of winding layers increases, the magnetic leakage energy stored in the winding area also increase, which results in a great increase of winding AC resistance. For high voltage transformers, due to the insulation clearance, winding interleave cannot be used and the high voltage winding is put separately with low voltage winding. The leakage magnetic energy storage is more than that of the conventional low voltage transformer. This helps to increase the leakage inductance but the winding AC resistance and loss will also increase a lot. Two transformer winding structures can be considered. The first one is more widely used, as shown in Fig.5. The high voltage and low voltage windings are divided into two equal parts. Two low voltage windings surround the transformer core legs and then two high voltage windings surround two low voltage windings respectively. The other option is to separate the high voltage winding from low voltage winding, as shown in Fig.6. A solid insulation layer is inserted between two C cores.
in terms of the leakage inductance, magnetizing inductance, ease of insulation and winding loss, which are four key aspects of the high voltage high frequency transformer for SST applications. C. Leakage and Magnetizing Inductance The leakage inductance of the high voltage high frequency transformer was compared based on two types of winding layout in Max3D FEA simulation. With coaxial winding structure as shown in Fig.5, the distance between high voltage and low voltage winding is small and the magnetic field intensity is also relatively small because the winding is divided into two parts. There is no energy stored in the major part of the space of the window. With separate winding structure as shown in Fig.6, the high voltage winding and the low voltage winding are far away from each other and then there is a lot of magnetic energy stored in the window space. Therefore, larger leakage inductance is expected from the separate winding structure. Based on the Max3D FEA simulation results, with the coaxial winding structure, the leakage inductance is 5.5mH and the magnetizing inductance with 0.1mm air gap is 1476mH. This leakage inductance is much lower than required value. However, with the separate winding structure, the leakage inductance increases to 35.5mH and 1.0mm solid insulation layer thickness provides 273mH magnetizing inductance, which is enough to mitigate magnetizing current. It should be noted that 82% core stack fill factor needs to be considered in simulation. If the insulation layer or gap thickness is too small, the dielectric stress of the gap is too high to insulate the high voltage based on the proposed HV insulation strategy.
D. High Voltage Insulation Strategy SST high voltage and high frequency transformer is designed as dry-type transformer to meet the demand for environmentally friendly product. Hence, there is one additional important concern about the insulation for safe and reliable operation of a transformer. When the voltage insulation is taken into consideration, at least 15kV insulation has to be obtained between the high voltage winding and low voltage winding in continuous operation. The maximal winding voltage insulation requirement occurs in the top-most transformer as shown in Fig.1, in which the voltage potential referred to ground can be raised up to 11.4kV. Air is normally a good insulator but it begins to break down under electric field strength of Fig.5: Coaxial high voltage and low voltage winding layout more than about 3kV/mm at 1cm distance, which can lead to spark or arc that bridges the gap between conductors. NEMA's (National Electrical Manufacturers Association) standards TR 1 mentions that the minimum clearance between live parts of dry type transformer is 5 inches for distribution transformers under 15kV but the realizable minimum distance for SST application is investigated to reduce the size and volume of the transformer. Obviously 5-inch clearance distance cannot be met by the limited window size of the high frequency Fig.6: Separate high voltage and low voltage winding layout transformer. Therefore, additional voltage insulation strategy should be proposed. The above two winding layout structures are compared 490 For co-axial winding layout, 15kV has to be supported
by the space between the high voltage and low voltage windings. It can also be found that in core window area two 15kV insulation structures between the high voltage and low voltage windings should be used. In addition, large clearance is required for 15kV level insulation if air is used as the insulator. Therefore, it is very difficult to insulate windings with the compact window geometry in the high frequency transformer. One of the oil-free insulation strategies is proposed and shown in Fig.6. A thin insulation layer (red), such as polypropylene or epoxy plate, is inserted into the gap between two C cores. The insulation layer mitigates the isolation requirement between winding and core, if the primary core is electrically connected to negative terminal of DC-link in its belonged stage, and the secondary core is electrically connected to negative DClink terminal of the adjacently next lower cascaded stage in high voltage side. Thus, the primary winding to ground high voltage is partly supported by the inserted insulation layer and winding to core insulation requirement is mitigated. With this insulation structure, the winding to core insulation voltage is reduced to 3.8kV instead of 11.4kV, which can be easily met by bobbin and selected wires with HV insulation coating. In worst scenario, the inserted insulation layer will support a maximal primary core to secondary core voltage of 3.8kV. With 1mm thickness, the dielectric strength of inserted insulation layer can be good enough with appropriate edge termination design to avoid sharp edges or corners, if 20kV/mm maximal dielectric strength is considered. E. Summary of Comparison The advantages and disadvantages for two winding structures are summarized in Table V. Since it is easier to obtain the required leakage inductance value and the required insulation voltage, while the magnetizing inductance is acceptable, the separate winding layout structure is used in this project. In the prototype, the black wire is low voltage winding and blue wire is high voltage winding.
current generated by winding capacitance is very small compared to load current and does not have much impact on circuit. But the impact of the parasitic capacitance on common mode current and EMI needs to be further studied. TABLE VI: MAJOR SPECIFICATIONS OF THE PROTOTYPE TRANSFORMER Switching frequency High voltage winding Low voltage winding Power rating Core cross section area Core window area Peak flux density Number of high voltage winding turns Number of low voltage winding turns High voltage wire insulation layer Magnetizing inductance (referred to high voltage winding side) Leakage inductance (referred to high voltage winding side)
3kHz 3.8kV (Square wave) 400V (Square wave) 6.7kVA 69cm2 42 cm2 0.242T (>0.3T very noisy) 190 20 0.45mm PFA coating [12] Tested:287mH Simulated: 273mH Tested: 36.0mH Simulated: 35.5mH
TABLE VII: MEASUREMENT OF PARASITIC CAPACITANCE High voltage winding capacitance Low voltage winding capacitance HV winding to core capacitance LV winding to core capacitance
35pF 24pF 175pF 240pF
Fig.7: Prototype of 3.8kV to 400V 6.7kVA 3kHz transformer for SST
TABLE V: COMPARISON OF TWO WINDING LAYOUT STRUCTURE Leakage inductance Winding AC resistance Magnetizing inductance Insulation
Co-axial winding Low Medium High Difficult
Separate winding High High Medium Easier
IV. PROTOTYPE AND TEST RESULTS One high voltage high frequency transformer prototype was built to verify the design. The major specifications of the transformer are listed in Table VI and the prototype is shown in Fig.7. The parasitic capacitance of the prototype transformer was measured and the results are listed in Table VII. The parasitic capacitance is important for high voltage high frequency transformer design since the parasitic capacitance will generate high reactive current and also common mode current in the circuit. In our case, since the switching frequency is relatively low (3kHz) and the high voltage winding capacitance is 35pF, the reactive
Fig.8: High voltage winding AC resistance
The AC resistance of the high voltage wire is also important since the number of high voltage winding turns is large and there are so many layers as well. The Dowell’s model [13] is used to predict the AC resistance at high frequency range for high voltage winding, as shown in Fig.8. The winding AC resistance at 3kHz is about four times of the DC resistance. The AC resistance measurement was also conducted and the measured winding resistance coincides well with model prediction in lower frequency range but it increases faster in higher frequency range. The prototype transformer was tested at 400V 3kHz input in low voltage side with 7kW resistor bank load connected to high voltage winding. The test waveforms 491 were shown in Fig.9. It can be found that L/R time
constant is 85uS and the corresponding leakage inductance is 35.7mH, which is very close to 36mH measured value by LRC meter. The high voltage winding was excited to 3.8kV. The loss and efficiency test results are listed in Table VIII.
additional reactive current in the circuit. The measured core loss is 81W and transformer efficiency is close to 97% from half to full load range. The measured maximal surface temperature rise is 35°C with nature air cooling. The partial discharge issue and the performance of proposed insulation structure should be evaluated in the future work. ACKNOWLEDGMENT This work was supported by ERC Program of the U.S. National Science Foundation under Award Number EEC08212121. REFERENCES
Fig.9: Test waveforms of prototype transformer with 400V input voltage on LV winding and 3.8kV output with 7kW resistor load TABLE VIII: LOSS AND EFFICIENCY OF PROTOTYPE TRANSFORMER Core loss HV winding LV winding Total Load Efficiency (W) loss (W) loss (W) (W) 3.9kW 80.6 16.4 28.7 126 96.9% 7.0kW 80.6 62.1 85.4 228 96.8%
Fig.10: The measurement of the prototype transformer temperature rise at 7kW load with nature air cooling
The temperature rise of the transformer at 7kW load with nature air cooling was measured and shown in Fig.10. The temperature rise is not an issue and the measured maximal surface temperature rise is 35°C at the low voltage winding. V. CONCLUSIONS The magnetic core materials were reviewed and compared in this paper for the high voltage and high frequency transformer in SST applications. The amorphous core material is selected to build the 3kHz 6.7kVA transformer prototype, due to its acceptable specific loss, good mechanical strength and relatively low cost. The transformer design trade off was conducted in terms of the leakage inductance, magnetizing inductance, winding AC resistance and voltage insulation. The separate winding layout structure with an inserted thin insulation layer between two C core pairs is more attractive because it provides required leakage and magnetizing inductance and it is easier to insulate 15kV voltage, although the winding AC resistance is higher. One 3.8kV to 400V 3kHz 6.7kVA transformer prototype was constructed and tested with full voltage and load. The measured leakage and magnetizing inductance meet the design specification and the parasitic capacitance does not have great impact in terms of the generating
[1] M. Rabinowitz, “Power Systems of the Future,” in IEEE Power Engineering Review, vol.20, issue 3, March 2000, pp.10-29. [2] L. Yang, T. Zhao, J. Wang, A.Q. Huang, “Design and Analysis of a 270kW Five-level DC/DC Converter for Solid State Transformer Using 10kV SiC Power Devices,” in IEEE 2007 Power Electronics Specialists Conference, 2007, pp.245-251. [3] E.R.Jr. Ronan, S.D. Sudhoff, S.F. Glover, D.L. Galloway, “Application of Power Electronics to the Distribution Transformer,” in IEEE 2000 Applied Power Electronics Conference, 2000, vol.2, pp.861-867. [4] S. Bhattacharya, T. Zhao, G. Wang, S. Dutta, S. Baek, Y. Du, B. Parkhideh, X. Zhou, Alex Q. Huang, “Design and Development of Generation-I Silicon Based Solid State Transformer,” in IEEE 2010 Applied Power Electronics Conference and Expositions, February 2010, pp. 16661673. [5] S. Mulder, “Power Ferrite Loss Formulas for Transformer Design”, Power Conversion & Intelligent Motion, vol. 21, no. 7, July 1995, pp. 22–31. [6] J. Biela, D. Aggeler, S. Inoue, H. Akagi, J. W. Kolar, “Bidirectional Isolated DC-DC Converter for Nextgeneration Power Distribution - Comparison of Converters Using Si and SiC Devices,” in 2007 Power Conversion Conference, 2007, vol. 127, pp.510-517. [7] T. Filchev, D. Cook, P. Wheeler, A. Van den Bossche, J. Clare, V. Valchev, “High Power, High Voltage, High Frequency Transformer/Rectifier for HV Industrial Applications,” in EPE 2008 Power Electronics and Motion Control Conference, 2008, pp.1326-1331. [8] D. Aggeler, J. Biela, J.W. Kolar, “A Compact, High Voltage 25kW, 50kHz DC-DC Converter Based on SiC JFETs”, in IEEE 2008 Applied Power Electronics Conference and Exposition, 2008, pp.801 – 807. [9] http://www.metglas.com [10] M. H. Kheraluwala, R. W. Gasgoigne, D. M. Divan, E. Bauman, “Performance Characterization of a High Power Dual Active Bridge DC-DC Converter,” IEEE 1990 Industry Applications Society Annual Meeting, Oct. 1990, vol.2, pp. 1267-1273. [11] M. S. Rahuls, C. W. Novotny, D. M. Divan, R. R. Bacon, R. W. Gascoigne, “Multiturn High-Frequency Coaxial Winding Power Transformers” IEEE transaction on Industry Application, vol. 31, no.1, Jan/Feb 1995 [12] http://www.newenglandwire.com [13] P.L. Dowell, “Effects of eddy currents in transformer windings”, Proceedings of the Institution of Electrical Engineers, vol.113, no.8, August 1966, pp.1387-1394. 492
